src/session.c

   1 /* This file is part of Pazpar2.
   2    Copyright (C) 2006-2013 Index Data
   3
   4 Pazpar2 is free software; you can redistribute it and/or modify it under
   5 the terms of the GNU General Public License as published by the Free
   6 Software Foundation; either version 2, or (at your option) any later
   7 version.
   8
   9 Pazpar2 is distributed in the hope that it will be useful, but WITHOUT ANY
  10 WARRANTY; without even the implied warranty of MERCHANTABILITY or
  11 FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License
  12 for more details.
  13
  14 You should have received a copy of the GNU General Public License
  15 along with this program; if not, write to the Free Software
  16 Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
  17
  18 */
  19
  20 /** \file session.c
  21     \brief high-level logic; mostly user sessions and settings
  22 */
  23
  24 #if HAVE_CONFIG_H
  25 #include <config.h>
  26 #endif
  27
  28 #include <time.h>
  29 #include <stdlib.h>
  30 #include <stdio.h>
  31 #include <string.h>
  32 #if HAVE_SYS_TIME_H
  33 #include <sys/time.h>
  34 #endif
  35 #if HAVE_UNISTD_H
  36 #include <unistd.h>
  37 #endif
  38 #ifdef WIN32
  39 #include <windows.h>
  40 #endif
  41 #include <signal.h>
  42 #include <ctype.h>
  43 #include <assert.h>
  44 #include <math.h>
  45
  46 #include <yaz/marcdisp.h>
  47 #include <yaz/comstack.h>
  48 #include <yaz/tcpip.h>
  49 #include <yaz/proto.h>
  50 #include <yaz/readconf.h>
  51 #include <yaz/pquery.h>
  52 #include <yaz/otherinfo.h>
  53 #include <yaz/yaz-util.h>
  54 #include <yaz/nmem.h>
  55 #include <yaz/query-charset.h>
  56 #include <yaz/querytowrbuf.h>
  57 #include <yaz/oid_db.h>
  58 #include <yaz/snprintf.h>
  59
  60 #define USE_TIMING 1
  61 #if USE_TIMING
  62 #include <yaz/timing.h>
  63 #endif
  64
  65 #include "ppmutex.h"
  66 #include "parameters.h"
  67 #include "session.h"
  68 #include "eventl.h"
  69 #include "http.h"
  70 #include "termlists.h"
  71 #include "reclists.h"
  72 #include "relevance.h"
  73 #include "database.h"
  74 #include "client.h"
  75 #include "settings.h"
  76 #include "normalize7bit.h"
  77
  78 #include <libxml/tree.h>
  79
  80 #define MAX_CHUNK 15
  81
  82 #define MAX(a,b) ((a)>(b)?(a):(b))
  83
  84 // Note: Some things in this structure will eventually move to configuration
  85 struct parameters global_parameters =
  86 {
  87     0,   // dump_records
  88     0,   // debug_mode
  89     0,   // predictable sessions
  90 };
  91
  92 struct client_list {
  93     struct client *client;
  94     struct client_list *next;
  95 };
  96
  97 /* session counting (1) , disable client counting (0) */
  98 static YAZ_MUTEX g_session_mutex = 0;
  99 static int no_sessions = 0;
 100 static int no_session_total = 0;
 101
 102 static int session_use(int delta)
 103 {
 104     int sessions;
 105     if (!g_session_mutex)
 106         yaz_mutex_create(&g_session_mutex);
 107     yaz_mutex_enter(g_session_mutex);
 108     no_sessions += delta;
 109     if (delta > 0)
 110         no_session_total += delta;
 111     sessions = no_sessions;
 112     yaz_mutex_leave(g_session_mutex);
 113     yaz_log(YLOG_DEBUG, "%s sessions=%d", delta == 0 ? "" : (delta > 0 ? "INC" : "DEC"), no_sessions);
 114     return sessions;
 115 }
 116
 117 int sessions_count(void)
 118 {
 119     return session_use(0);
 120 }
 121
 122 int session_count_total(void)
 123 {
 124     int total = 0;
 125     if (!g_session_mutex)
 126         return 0;
 127     yaz_mutex_enter(g_session_mutex);
 128     total = no_session_total;
 129     yaz_mutex_leave(g_session_mutex);
 130     return total;
 131 }
 132
 133 static void log_xml_doc(xmlDoc *doc)
 134 {
 135     FILE *lf = yaz_log_file();
 136     xmlChar *result = 0;
 137     int len = 0;
 138 #if LIBXML_VERSION >= 20600
 139     xmlDocDumpFormatMemory(doc, &result, &len, 1);
 140 #else
 141     xmlDocDumpMemory(doc, &result, &len);
 142 #endif
 143     if (lf && len)
 144     {
 145         (void) fwrite(result, 1, len, lf);
 146         fprintf(lf, "\n");
 147     }
 148     xmlFree(result);
 149 }
 150
 151 static void session_enter(struct session *s, const char *caller)
 152 {
 153     if (caller)
 154         session_log(s, YLOG_DEBUG, "Session lock by %s", caller);
 155     yaz_mutex_enter(s->session_mutex);
 156 }
 157
 158 static void session_leave(struct session *s, const char *caller)
 159 {
 160     yaz_mutex_leave(s->session_mutex);
 161     if (caller)
 162         session_log(s, YLOG_DEBUG, "Session unlock by %s", caller);
 163 }
 164
 165 static void session_normalize_facet(struct session *s, const char *type,
 166                                     const char *value,
 167                                     WRBUF display_wrbuf,
 168                                     WRBUF facet_wrbuf)
 169 {
 170     struct conf_service *service = s->service;
 171     pp2_charset_token_t prt;
 172     const char *facet_component;
 173     int i;
 174     const char *icu_chain_id = 0;
 175
 176     for (i = 0; i < service->num_metadata; i++)
 177         if (!strcmp((service->metadata + i)->name, type))
 178             icu_chain_id = (service->metadata + i)->facetrule;
 179     if (!icu_chain_id)
 180         icu_chain_id = "facet";
 181     prt = pp2_charset_token_create(service->charsets, icu_chain_id);
 182     if (!prt)
 183     {
 184         session_log(s, YLOG_FATAL,
 185                     "Unknown ICU chain '%s' for facet of type '%s'",
 186                 icu_chain_id, type);
 187         wrbuf_destroy(facet_wrbuf);
 188         wrbuf_destroy(display_wrbuf);
 189         return;
 190     }
 191     pp2_charset_token_first(prt, value, 0);
 192     while ((facet_component = pp2_charset_token_next(prt)))
 193     {
 194         const char *display_component;
 195         if (*facet_component)
 196         {
 197             if (wrbuf_len(facet_wrbuf))
 198                 wrbuf_puts(facet_wrbuf, " ");
 199             wrbuf_puts(facet_wrbuf, facet_component);
 200         }
 201         display_component = pp2_get_display(prt);
 202         if (display_component)
 203         {
 204             if (wrbuf_len(display_wrbuf))
 205                 wrbuf_puts(display_wrbuf, " ");
 206             wrbuf_puts(display_wrbuf, display_component);
 207         }
 208     }
 209     pp2_charset_token_destroy(prt);
 210 }
 211
 212 void add_facet(struct session *s, const char *type, const char *value, int count)
 213 {
 214     WRBUF facet_wrbuf = wrbuf_alloc();
 215     WRBUF display_wrbuf = wrbuf_alloc();
 216
 217     session_normalize_facet(s, type, value, display_wrbuf, facet_wrbuf);
 218
 219     if (wrbuf_len(facet_wrbuf))
 220     {
 221         int i;
 222         for (i = 0; i < s->num_termlists; i++)
 223             if (!strcmp(s->termlists[i].name, type))
 224                 break;
 225         if (i == s->num_termlists)
 226         {
 227             if (i == SESSION_MAX_TERMLISTS)
 228             {
 229                 session_log(s, YLOG_FATAL, "Too many termlists");
 230                 wrbuf_destroy(facet_wrbuf);
 231                 wrbuf_destroy(display_wrbuf);
 232                 return;
 233             }
 234
 235             s->termlists[i].name = nmem_strdup(s->nmem, type);
 236             s->termlists[i].termlist = termlist_create(s->nmem);
 237             s->num_termlists = i + 1;
 238         }
 239
 240 #if 0
 241         session_log(s, YLOG_LOG, "Facets for %s: %s norm:%s (%d)", type, value, wrbuf_cstr(facet_wrbuf), count);
 242 #endif
 243         termlist_insert(s->termlists[i].termlist, wrbuf_cstr(display_wrbuf),
 244                         wrbuf_cstr(facet_wrbuf), count);
 245     }
 246     wrbuf_destroy(facet_wrbuf);
 247     wrbuf_destroy(display_wrbuf);
 248 }
 249
 250 static xmlDoc *record_to_xml(struct session *se,
 251                              struct session_database *sdb, const char *rec)
 252 {
 253     struct database *db = sdb->database;
 254     xmlDoc *rdoc = 0;
 255
 256     rdoc = xmlParseMemory(rec, strlen(rec));
 257
 258     if (!rdoc)
 259     {
 260         session_log(se, YLOG_WARN, "Non-wellformed XML");
 261         return 0;
 262     }
 263
 264     if (global_parameters.dump_records)
 265     {
 266         session_log(se, YLOG_LOG, "Un-normalized record from %s", db->id);
 267         log_xml_doc(rdoc);
 268     }
 269
 270     return rdoc;
 271 }
 272
 273 #define MAX_XSLT_ARGS 16
 274
 275 // Add static values from session database settings if applicable
 276 static void insert_settings_parameters(struct session_database *sdb,
 277                                        struct conf_service *service,
 278                                        char **parms,
 279                                        NMEM nmem)
 280 {
 281     int i;
 282     int nparms = 0;
 283     int offset = 0;
 284
 285     for (i = 0; i < service->num_metadata; i++)
 286     {
 287         struct conf_metadata *md = &service->metadata[i];
 288         int setting;
 289
 290         if (md->setting == Metadata_setting_parameter &&
 291             (setting = settings_lookup_offset(service, md->name)) >= 0)
 292         {
 293             const char *val = session_setting_oneval(sdb, setting);
 294             if (val && nparms < MAX_XSLT_ARGS)
 295             {
 296                 char *buf;
 297                 int len = strlen(val);
 298                 buf = nmem_malloc(nmem, len + 3);
 299                 buf[0] = '\'';
 300                 strcpy(buf + 1, val);
 301                 buf[len+1] = '\'';
 302                 buf[len+2] = '\0';
 303                 parms[offset++] = md->name;
 304                 parms[offset++] = buf;
 305                 nparms++;
 306             }
 307         }
 308     }
 309     parms[offset] = 0;
 310 }
 311
 312 // Add static values from session database settings if applicable
 313 static void insert_settings_values(struct session_database *sdb, xmlDoc *doc,
 314     struct conf_service *service)
 315 {
 316     int i;
 317
 318     for (i = 0; i < service->num_metadata; i++)
 319     {
 320         struct conf_metadata *md = &service->metadata[i];
 321         int offset;
 322
 323         if (md->setting == Metadata_setting_postproc &&
 324             (offset = settings_lookup_offset(service, md->name)) >= 0)
 325         {
 326             const char *val = session_setting_oneval(sdb, offset);
 327             if (val)
 328             {
 329                 xmlNode *r = xmlDocGetRootElement(doc);
 330                 xmlNode *n = xmlNewTextChild(r, 0, (xmlChar *) "metadata",
 331                                              (xmlChar *) val);
 332                 xmlSetProp(n, (xmlChar *) "type", (xmlChar *) md->name);
 333             }
 334         }
 335     }
 336 }
 337
 338 static xmlDoc *normalize_record(struct session *se,
 339                                 struct session_database *sdb,
 340                                 struct conf_service *service,
 341                                 const char *rec, NMEM nmem)
 342 {
 343     xmlDoc *rdoc = record_to_xml(se, sdb, rec);
 344
 345     if (rdoc)
 346     {
 347         char *parms[MAX_XSLT_ARGS*2+1];
 348
 349         insert_settings_parameters(sdb, service, parms, nmem);
 350
 351         if (normalize_record_transform(sdb->map, &rdoc, (const char **)parms))
 352         {
 353             session_log(se, YLOG_WARN, "Normalize failed");
 354         }
 355         else
 356         {
 357             insert_settings_values(sdb, rdoc, service);
 358
 359             if (global_parameters.dump_records)
 360             {
 361                 session_log(se, YLOG_LOG, "Normalized record from %s",
 362                             sdb->database->id);
 363                 log_xml_doc(rdoc);
 364             }
 365         }
 366     }
 367     return rdoc;
 368 }
 369
 370 void session_settings_dump(struct session *se,
 371                            struct session_database *db,
 372                            WRBUF w)
 373 {
 374     if (db->settings)
 375     {
 376         int i, num = db->num_settings;
 377         for (i = 0; i < num; i++)
 378         {
 379             struct setting *s = db->settings[i];
 380             for (;s ; s = s->next)
 381             {
 382                 wrbuf_puts(w, "<set name=\"");
 383                 wrbuf_xmlputs(w, s->name);
 384                 wrbuf_puts(w, "\" value=\"");
 385                 wrbuf_xmlputs(w, s->value);
 386                 wrbuf_puts(w, "\"/>");
 387             }
 388             if (db->settings[i])
 389                 wrbuf_puts(w, "\n");
 390         }
 391     }
 392 }
 393
 394 // Retrieve first defined value for 'name' for given database.
 395 // Will be extended to take into account user associated with session
 396 const char *session_setting_oneval(struct session_database *db, int offset)
 397 {
 398     if (offset >= db->num_settings || !db->settings[offset])
 399         return "";
 400     return db->settings[offset]->value;
 401 }
 402
 403 // Prepare XSLT stylesheets for record normalization
 404 // Structures are allocated on the session_wide nmem to avoid having
 405 // to recompute this for every search. This would lead
 406 // to leaking if a single session was to repeatedly change the PZ_XSLT
 407 // setting. However, this is not a realistic use scenario.
 408 static int prepare_map(struct session *se, struct session_database *sdb)
 409 {
 410     if (sdb->settings && !sdb->map)
 411     {
 412         const char *s;
 413
 414         if (sdb->settings[PZ_XSLT] &&
 415             (s = session_setting_oneval(sdb, PZ_XSLT)))
 416         {
 417             char auto_stylesheet[256];
 418
 419             if (!strcmp(s, "auto"))
 420             {
 421                 const char *request_syntax = session_setting_oneval(
 422                     sdb, PZ_REQUESTSYNTAX);
 423                 if (request_syntax)
 424                 {
 425                     char *cp;
 426                     yaz_snprintf(auto_stylesheet, sizeof(auto_stylesheet),
 427                                  "%s.xsl", request_syntax);
 428                     for (cp = auto_stylesheet; *cp; cp++)
 429                     {
 430                         /* deliberately only consider ASCII */
 431                         if (*cp > 32 && *cp < 127)
 432                             *cp = tolower(*cp);
 433                     }
 434                     s = auto_stylesheet;
 435                 }
 436                 else
 437                 {
 438                     session_log(se, YLOG_WARN,
 439                                 "No pz:requestsyntax for auto stylesheet");
 440                 }
 441             }
 442             sdb->map = normalize_cache_get(se->normalize_cache,
 443                                            se->service, s);
 444             if (!sdb->map)
 445                 return -1;
 446         }
 447     }
 448     return 0;
 449 }
 450
 451 // called if watch should be removed because http_channel is to be destroyed
 452 static void session_watch_cancel(void *data, struct http_channel *c,
 453                                  void *data2)
 454 {
 455     struct session_watchentry *ent = data;
 456
 457     ent->fun = 0;
 458     ent->data = 0;
 459     ent->obs = 0;
 460 }
 461
 462 // set watch. Returns 0=OK, -1 if watch is already set
 463 int session_set_watch(struct session *s, int what,
 464                       session_watchfun fun, void *data,
 465                       struct http_channel *chan)
 466 {
 467     int ret;
 468     session_enter(s, "session_set_watch");
 469     if (s->watchlist[what].fun)
 470         ret = -1;
 471     else
 472     {
 473
 474         s->watchlist[what].fun = fun;
 475         s->watchlist[what].data = data;
 476         s->watchlist[what].obs = http_add_observer(chan, &s->watchlist[what],
 477                                                    session_watch_cancel);
 478         ret = 0;
 479     }
 480     session_leave(s, "session_set_watch");
 481     return ret;
 482 }
 483
 484 void session_alert_watch(struct session *s, int what)
 485 {
 486     assert(s);
 487     session_enter(s, "session_alert_watch");
 488     if (s->watchlist[what].fun)
 489     {
 490         /* our watch is no longer associated with http_channel */
 491         void *data;
 492         session_watchfun fun;
 493
 494         http_remove_observer(s->watchlist[what].obs);
 495         fun  = s->watchlist[what].fun;
 496         data = s->watchlist[what].data;
 497
 498         /* reset watch before fun is invoked - in case fun wants to set
 499            it again */
 500         s->watchlist[what].fun = 0;
 501         s->watchlist[what].data = 0;
 502         s->watchlist[what].obs = 0;
 503
 504         session_leave(s, "session_alert_watch");
 505         session_log(s, YLOG_DEBUG,
 506                     "Alert Watch: %d calling function: %p", what, fun);
 507         fun(data);
 508     }
 509     else
 510         session_leave(s,"session_alert_watch");
 511 }
 512
 513 //callback for grep_databases
 514 static void select_targets_callback(struct session *se,
 515                                     struct session_database *db)
 516 {
 517     struct client *cl;
 518     struct client_list *l;
 519
 520     for (l = se->clients_cached; l; l = l->next)
 521         if (client_get_database(l->client) == db)
 522             break;
 523
 524     if (l)
 525         cl = l->client;
 526     else
 527     {
 528         cl = client_create(db->database->id);
 529         client_set_database(cl, db);
 530
 531         l = xmalloc(sizeof(*l));
 532         l->client = cl;
 533         l->next = se->clients_cached;
 534         se->clients_cached = l;
 535     }
 536     /* set session always. If may be 0 if client is not active */
 537     client_set_session(cl, se);
 538
 539     l = xmalloc(sizeof(*l));
 540     l->client = cl;
 541     l->next = se->clients_active;
 542     se->clients_active = l;
 543 }
 544
 545 static void session_reset_active_clients(struct session *se,
 546                                          struct client_list *new_list)
 547 {
 548     struct client_list *l;
 549
 550     session_enter(se, "session_reset_active_clients");
 551     l = se->clients_active;
 552     se->clients_active = new_list;
 553     session_leave(se, "session_reset_active_clients");
 554
 555     while (l)
 556     {
 557         struct client_list *l_next = l->next;
 558
 559         client_lock(l->client);
 560         client_set_session(l->client, 0); /* mark client inactive */
 561         client_unlock(l->client);
 562
 563         xfree(l);
 564         l = l_next;
 565     }
 566 }
 567
 568 static void session_remove_cached_clients(struct session *se)
 569 {
 570     struct client_list *l;
 571
 572     session_reset_active_clients(se, 0);
 573
 574     session_enter(se, "session_remove_cached_clients");
 575     l = se->clients_cached;
 576     se->clients_cached = 0;
 577     session_leave(se, "session_remove_cached_clients");
 578
 579     while (l)
 580     {
 581         struct client_list *l_next = l->next;
 582         client_lock(l->client);
 583         client_set_session(l->client, 0);
 584         client_set_database(l->client, 0);
 585         client_unlock(l->client);
 586         client_destroy(l->client);
 587         xfree(l);
 588         l = l_next;
 589     }
 590 }
 591
 592 // Associates a set of clients with a session;
 593 // Note: Session-databases represent databases with per-session
 594 // setting overrides
 595 static int select_targets(struct session *se, const char *filter)
 596 {
 597     return session_grep_databases(se, filter, select_targets_callback);
 598 }
 599
 600 int session_active_clients(struct session *s)
 601 {
 602     struct client_list *l;
 603     int res = 0;
 604
 605     for (l = s->clients_active; l; l = l->next)
 606         if (client_is_active(l->client))
 607             res++;
 608
 609     return res;
 610 }
 611
 612 int session_is_preferred_clients_ready(struct session *s)
 613 {
 614     struct client_list *l;
 615     int res = 0;
 616
 617     for (l = s->clients_active; l; l = l->next)
 618         if (client_is_active_preferred(l->client))
 619             res++;
 620     session_log(s, YLOG_DEBUG, "Has %d active preferred clients.", res);
 621     return res == 0;
 622 }
 623
 624 static void session_clear_set(struct session *se, struct reclist_sortparms *sp)
 625 {
 626     reclist_destroy(se->reclist);
 627     if (nmem_total(se->nmem))
 628         session_log(se, YLOG_DEBUG, "NMEN operation usage %zd",
 629                     nmem_total(se->nmem));
 630     nmem_reset(se->nmem);
 631     se->total_records = se->total_merged = 0;
 632     se->num_termlists = 0;
 633
 634     relevance_clear(se->relevance);
 635
 636     /* reset list of sorted results and clear to relevance search */
 637     se->sorted_results = nmem_malloc(se->nmem, sizeof(*se->sorted_results));
 638     se->sorted_results->name = nmem_strdup(se->nmem, sp->name);
 639     se->sorted_results->increasing = sp->increasing;
 640     se->sorted_results->type = sp->type;
 641     se->sorted_results->next = 0;
 642
 643     session_log(se, YLOG_DEBUG, "clear_set session_sort: field=%s increasing=%d type=%d configured",
 644             sp->name, sp->increasing, sp->type);
 645
 646     se->reclist = reclist_create(se->nmem);
 647 }
 648
 649 static void session_sort_unlocked(struct session *se,
 650                                   struct reclist_sortparms *sp,
 651                                   const char *mergekey)
 652 {
 653     struct client_list *l;
 654     const char *field = sp->name;
 655     int increasing = sp->increasing;
 656     int type  = sp->type;
 657     int clients_research = 0;
 658
 659     session_log(se, YLOG_DEBUG, "session_sort field=%s increasing=%d type=%d",
 660                 field, increasing, type);
 661
 662     if (mergekey && strcmp(se->mergekey, mergekey))
 663     {
 664         /* new mergekey must research/reingest anyway */
 665         assert(mergekey);
 666         xfree(se->mergekey);
 667         se->mergekey = *mergekey ? xstrdup(mergekey) : 0;
 668         clients_research = 1;
 669         session_log(se, YLOG_DEBUG, "session_sort: new mergekey = %s",
 670                     mergekey);
 671     }
 672     if (clients_research == 0)
 673     {
 674         struct reclist_sortparms *sr;
 675         for (sr = se->sorted_results; sr; sr = sr->next)
 676             if (!reclist_sortparms_cmp(sr, sp))
 677                 break;
 678         if (sr)
 679         {
 680             session_log(se, YLOG_DEBUG, "session_sort: field=%s increasing=%d type=%d already fetched",
 681                         field, increasing, type);
 682             return;
 683         }
 684     }
 685     session_log(se, YLOG_DEBUG, "session_sort: field=%s increasing=%d type=%d must fetch",
 686                 field, increasing, type);
 687
 688     // We need to reset reclist on every sort that changes the records, not just for position
 689     // So if just one client requires new searching, we need to clear set.
 690     // Ask each of the client if sorting requires re-search due to native sort
 691     // If it does it will require us to
 692     for (l = se->clients_active; l; l = l->next)
 693     {
 694         struct client *cl = l->client;
 695         // Assume no re-search is required.
 696         client_parse_init(cl, 1);
 697         clients_research += client_parse_sort(cl, sp);
 698     }
 699     if (clients_research)
 700     {
 701         session_log(se, YLOG_DEBUG,
 702                     "session_sort: reset results due to %d clients researching",
 703                     clients_research);
 704         session_clear_set(se, sp);
 705     }
 706     else
 707     {
 708         // A new sorting based on same record set
 709         struct reclist_sortparms *sr = nmem_malloc(se->nmem, sizeof(*sr));
 710         sr->name = nmem_strdup(se->nmem, field);
 711         sr->increasing = increasing;
 712         sr->type = type;
 713         sr->next = se->sorted_results;
 714         se->sorted_results = sr;
 715         session_log(se, YLOG_DEBUG, "session_sort: no research/ingesting done");
 716         return ;
 717     }
 718     session_log(se, YLOG_DEBUG, "Re- search/ingesting for clients due to change in sort order");
 719
 720     for (l = se->clients_active; l; l = l->next)
 721     {
 722         struct client *cl = l->client;
 723         if (client_get_state(cl) == Client_Connecting ||
 724             client_get_state(cl) == Client_Idle ||
 725             client_get_state(cl) == Client_Working) {
 726             client_start_search(cl);
 727         }
 728         else
 729         {
 730             session_log(se, YLOG_DEBUG,
 731                         "session_sort: %s: No re-start/ingest in show. "
 732                         "Wrong client state: %d",
 733                         client_get_id(cl), client_get_state(cl));
 734         }
 735     }
 736 }
 737
 738 void session_sort(struct session *se, struct reclist_sortparms *sp,
 739                   const char *mergekey)
 740 {
 741     //session_enter(se, "session_sort");
 742     session_sort_unlocked(se, sp, mergekey);
 743     //session_leave(se, "session_sort");
 744 }
 745
 746
 747 enum pazpar2_error_code session_search(struct session *se,
 748                                        const char *query,
 749                                        const char *startrecs,
 750                                        const char *maxrecs,
 751                                        const char *filter,
 752                                        const char *limit,
 753                                        const char **addinfo,
 754                                        struct reclist_sortparms *sp,
 755                                        const char *mergekey)
 756 {
 757     int live_channels = 0;
 758     int no_working = 0;
 759     int no_failed_query = 0;
 760     int no_failed_limit = 0;
 761     struct client_list *l, *l0;
 762
 763     session_log(se, YLOG_DEBUG, "Search");
 764
 765     *addinfo = 0;
 766
 767     if (se->settings_modified) {
 768         session_remove_cached_clients(se);
 769     }
 770     else
 771         session_reset_active_clients(se, 0);
 772
 773     session_enter(se, "session_search");
 774     se->settings_modified = 0;
 775
 776     if (mergekey)
 777     {
 778         xfree(se->mergekey);
 779         se->mergekey = *mergekey ? xstrdup(mergekey) : 0;
 780     }
 781
 782     session_clear_set(se, sp);
 783     relevance_destroy(&se->relevance);
 784
 785     live_channels = select_targets(se, filter);
 786     if (!live_channels)
 787     {
 788         session_leave(se, "session_search");
 789         return PAZPAR2_NO_TARGETS;
 790     }
 791
 792     facet_limits_destroy(se->facet_limits);
 793     se->facet_limits = facet_limits_create(limit);
 794     if (!se->facet_limits)
 795     {
 796         *addinfo = "limit";
 797         session_leave(se, "session_search");
 798         return PAZPAR2_MALFORMED_PARAMETER_VALUE;
 799     }
 800
 801     l0 = se->clients_active;
 802     se->clients_active = 0;
 803     session_leave(se, "session_search");
 804
 805     for (l = l0; l; l = l->next)
 806     {
 807         int parse_ret;
 808         struct client *cl = l->client;
 809         client_parse_init(cl, 1);
 810         if (prepare_map(se, client_get_database(cl)) < 0)
 811             continue;
 812
 813         parse_ret = client_parse_query(cl, query, se->facet_limits);
 814         if (parse_ret == -1)
 815             no_failed_query++;
 816         else if (parse_ret == -2)
 817             no_failed_limit++;
 818         else if (parse_ret < 0)
 819             no_working++; /* other error, such as bad CCL map */
 820         else
 821         {
 822             client_parse_range(cl, startrecs, maxrecs);
 823             client_parse_sort(cl, sp);
 824             client_start_search(cl);
 825             no_working++;
 826         }
 827     }
 828     session_reset_active_clients(se, l0);
 829
 830     if (no_working == 0)
 831     {
 832         if (no_failed_query > 0)
 833         {
 834             *addinfo = "query";
 835             return PAZPAR2_MALFORMED_PARAMETER_VALUE;
 836         }
 837         else if (no_failed_limit > 0)
 838         {
 839             *addinfo = "limit";
 840             return PAZPAR2_MALFORMED_PARAMETER_VALUE;
 841         }
 842         else
 843             return PAZPAR2_NO_TARGETS;
 844     }
 845     session_log(se, YLOG_LOG, "session_start_search done");
 846     return PAZPAR2_NO_ERROR;
 847 }
 848
 849 // Creates a new session_database object for a database
 850 static void session_init_databases_fun(void *context, struct database *db)
 851 {
 852     struct session *se = (struct session *) context;
 853     struct session_database *new = nmem_malloc(se->session_nmem, sizeof(*new));
 854     int i;
 855
 856     new->database = db;
 857
 858     new->map = 0;
 859     assert(db->settings);
 860     new->settings = nmem_malloc(se->session_nmem,
 861                                 sizeof(struct settings *) * db->num_settings);
 862     new->num_settings = db->num_settings;
 863     for (i = 0; i < db->num_settings; i++)
 864     {
 865         struct setting *setting = db->settings[i];
 866         new->settings[i] = setting;
 867     }
 868     new->next = se->databases;
 869     se->databases = new;
 870 }
 871
 872 // Doesn't free memory associated with sdb -- nmem takes care of that
 873 static void session_database_destroy(struct session_database *sdb)
 874 {
 875     sdb->map = 0;
 876 }
 877
 878 // Initialize session_database list -- this represents this session's view
 879 // of the database list -- subject to modification by the settings ws command
 880 void session_init_databases(struct session *se)
 881 {
 882     se->databases = 0;
 883     predef_grep_databases(se, se->service, session_init_databases_fun);
 884 }
 885
 886 // Probably session_init_databases_fun should be refactored instead of
 887 // called here.
 888 static struct session_database *load_session_database(struct session *se,
 889                                                       char *id)
 890 {
 891     struct database *db = new_database_inherit_settings(id, se->session_nmem, se->service->settings);
 892     session_init_databases_fun((void*) se, db);
 893
 894     // New sdb is head of se->databases list
 895     return se->databases;
 896 }
 897
 898 // Find an existing session database. If not found, load it
 899 static struct session_database *find_session_database(struct session *se,
 900                                                       char *id)
 901 {
 902     struct session_database *sdb;
 903
 904     for (sdb = se->databases; sdb; sdb = sdb->next)
 905         if (!strcmp(sdb->database->id, id))
 906             return sdb;
 907     return load_session_database(se, id);
 908 }
 909
 910 // Apply a session override to a database
 911 void session_apply_setting(struct session *se, char *dbname, char *setting,
 912                            char *value)
 913 {
 914     struct session_database *sdb = find_session_database(se, dbname);
 915     struct conf_service *service = se->service;
 916     struct setting *new = nmem_malloc(se->session_nmem, sizeof(*new));
 917     int offset = settings_create_offset(service, setting);
 918
 919     expand_settings_array(&sdb->settings, &sdb->num_settings, offset,
 920                           se->session_nmem);
 921     new->precedence = 0;
 922     new->target = dbname;
 923     new->name = setting;
 924     new->value = value;
 925     new->next = sdb->settings[offset];
 926     sdb->settings[offset] = new;
 927
 928     se->settings_modified = 1;
 929
 930     // Force later recompute of settings-driven data structures
 931     // (happens when a search starts and client connections are prepared)
 932     switch (offset)
 933     {
 934     case PZ_XSLT:
 935         if (sdb->map)
 936         {
 937             sdb->map = 0;
 938         }
 939         break;
 940     }
 941 }
 942
 943 void session_destroy(struct session *se)
 944 {
 945     struct session_database *sdb;
 946     session_log(se, YLOG_DEBUG, "Destroying");
 947     session_use(-1);
 948     session_remove_cached_clients(se);
 949
 950     for (sdb = se->databases; sdb; sdb = sdb->next)
 951         session_database_destroy(sdb);
 952     normalize_cache_destroy(se->normalize_cache);
 953     relevance_destroy(&se->relevance);
 954     reclist_destroy(se->reclist);
 955     xfree(se->mergekey);
 956     if (nmem_total(se->nmem))
 957         session_log(se, YLOG_DEBUG, "NMEN operation usage %zd", nmem_total(se->nmem));
 958     if (nmem_total(se->session_nmem))
 959         session_log(se, YLOG_DEBUG, "NMEN session usage %zd", nmem_total(se->session_nmem));
 960     facet_limits_destroy(se->facet_limits);
 961     nmem_destroy(se->nmem);
 962     service_destroy(se->service);
 963     yaz_mutex_destroy(&se->session_mutex);
 964 }
 965
 966 size_t session_get_memory_status(struct session *session) {
 967     size_t session_nmem;
 968     if (session == 0)
 969         return 0;
 970     session_enter(session, "session_get_memory_status");
 971     session_nmem = nmem_total(session->nmem);
 972     session_leave(session, "session_get_memory_status");
 973     return session_nmem;
 974 }
 975
 976
 977 struct session *new_session(NMEM nmem, struct conf_service *service,
 978                             unsigned session_id)
 979 {
 980     int i;
 981     struct session *session = nmem_malloc(nmem, sizeof(*session));
 982
 983     char tmp_str[50];
 984
 985     sprintf(tmp_str, "session#%u", session_id);
 986
 987     session->session_id = session_id;
 988     session_log(session, YLOG_DEBUG, "New");
 989     session->service = service;
 990     session->relevance = 0;
 991     session->total_records = 0;
 992     session->number_of_warnings_unknown_elements = 0;
 993     session->number_of_warnings_unknown_metadata = 0;
 994     session->num_termlists = 0;
 995     session->reclist = reclist_create(nmem);
 996     session->clients_active = 0;
 997     session->clients_cached = 0;
 998     session->settings_modified = 0;
 999     session->session_nmem = nmem;
1000     session->nmem = nmem_create();
1001     session->databases = 0;
1002     session->sorted_results = 0;
1003     session->facet_limits = 0;
1004     session->mergekey = 0;
1005
1006     for (i = 0; i <= SESSION_WATCH_MAX; i++)
1007     {
1008         session->watchlist[i].data = 0;
1009         session->watchlist[i].fun = 0;
1010     }
1011     session->normalize_cache = normalize_cache_create();
1012     session->session_mutex = 0;
1013     pazpar2_mutex_create(&session->session_mutex, tmp_str);
1014     session_use(1);
1015     return session;
1016 }
1017
1018 const char * client_get_suggestions_xml(struct client *cl, WRBUF wrbuf);
1019
1020 static struct hitsbytarget *hitsbytarget_nb(struct session *se,
1021                                             int *count, NMEM nmem)
1022 {
1023     struct hitsbytarget *res = 0;
1024     struct client_list *l;
1025     size_t sz = 0;
1026
1027     for (l = se->clients_active; l; l = l->next)
1028         sz++;
1029
1030     res = nmem_malloc(nmem, sizeof(*res) * sz);
1031     *count = 0;
1032     for (l = se->clients_active; l; l = l->next)
1033     {
1034         struct client *cl = l->client;
1035         WRBUF w = wrbuf_alloc();
1036         const char *name = session_setting_oneval(client_get_database(cl),
1037                                                   PZ_NAME);
1038
1039         res[*count].id = client_get_id(cl);
1040         res[*count].name = *name ? name : "Unknown";
1041         res[*count].hits = client_get_hits(cl);
1042         res[*count].approximation = client_get_approximation(cl);
1043         res[*count].records = client_get_num_records(cl);
1044         res[*count].filtered = client_get_num_records_filtered(cl);
1045         res[*count].diagnostic =
1046             client_get_diagnostic(cl, &res[*count].message,
1047                                   &res[*count].addinfo);
1048         res[*count].state = client_get_state_str(cl);
1049         res[*count].connected  = client_get_connection(cl) ? 1 : 0;
1050         session_settings_dump(se, client_get_database(cl), w);
1051         res[*count].settings_xml = nmem_strdup(nmem, wrbuf_cstr(w));
1052         wrbuf_rewind(w);
1053         wrbuf_puts(w, "");
1054         res[*count].suggestions_xml = nmem_strdup(nmem, client_get_suggestions_xml(cl, w));
1055         wrbuf_destroy(w);
1056         (*count)++;
1057     }
1058     return res;
1059 }
1060
1061 struct hitsbytarget *get_hitsbytarget(struct session *se, int *count, NMEM nmem)
1062 {
1063     struct hitsbytarget *p;
1064     session_enter(se, "get_hitsbytarget");
1065     p = hitsbytarget_nb(se, count, nmem);
1066     session_leave(se, "get_hitsbytarget");
1067     return p;
1068 }
1069
1070 // Compares two hitsbytarget nodes by hitcount
1071 static int cmp_ht(const void *p1, const void *p2)
1072 {
1073     const struct hitsbytarget *h1 = p1;
1074     const struct hitsbytarget *h2 = p2;
1075     return h2->hits - h1->hits;
1076 }
1077
1078 // Compares two hitsbytarget nodes by hitcount
1079 static int cmp_ht_approx(const void *p1, const void *p2)
1080 {
1081     const struct hitsbytarget *h1 = p1;
1082     const struct hitsbytarget *h2 = p2;
1083     return h2->approximation - h1->approximation;
1084 }
1085
1086 static int targets_termlist_nb(WRBUF wrbuf, struct session *se, int num,
1087                                NMEM nmem, int version)
1088 {
1089     struct hitsbytarget *ht;
1090     int count, i;
1091
1092     ht = hitsbytarget_nb(se, &count, nmem);
1093     if (version >= 2)
1094         qsort(ht, count, sizeof(struct hitsbytarget), cmp_ht_approx);
1095     else
1096         qsort(ht, count, sizeof(struct hitsbytarget), cmp_ht);
1097     for (i = 0; i < count && i < num && ht[i].hits > 0; i++)
1098     {
1099
1100         // do only print terms which have display names
1101
1102         wrbuf_puts(wrbuf, "<term>\n");
1103
1104         wrbuf_puts(wrbuf, "<id>");
1105         wrbuf_xmlputs(wrbuf, ht[i].id);
1106         wrbuf_puts(wrbuf, "</id>\n");
1107
1108         wrbuf_puts(wrbuf, "<name>");
1109         if (!ht[i].name || !ht[i].name[0])
1110             wrbuf_xmlputs(wrbuf, "NO TARGET NAME");
1111         else
1112             wrbuf_xmlputs(wrbuf, ht[i].name);
1113         wrbuf_puts(wrbuf, "</name>\n");
1114
1115         wrbuf_printf(wrbuf, "<frequency>" ODR_INT_PRINTF "</frequency>\n",
1116                      ht[i].hits);
1117
1118         if (version >= 2) {
1119             // Should not print if we know it isn't a approximation.
1120             wrbuf_printf(wrbuf, "<approximation>" ODR_INT_PRINTF "</approximation>\n", ht[i].approximation);
1121             wrbuf_printf(wrbuf, "<records>%d</records>\n", ht[i].records - ht[i].filtered);
1122             wrbuf_printf(wrbuf, "<filtered>%d</filtered>\n", ht[i].filtered);
1123         }
1124
1125         wrbuf_puts(wrbuf, "<state>");
1126         wrbuf_xmlputs(wrbuf, ht[i].state);
1127         wrbuf_puts(wrbuf, "</state>\n");
1128
1129         wrbuf_printf(wrbuf, "<diagnostic>%d</diagnostic>\n",
1130                      ht[i].diagnostic);
1131         wrbuf_puts(wrbuf, "</term>\n");
1132     }
1133     return count;
1134 }
1135
1136 void perform_termlist(struct http_channel *c, struct session *se,
1137                       const char *name, int num, int version)
1138 {
1139     int i, j;
1140     NMEM nmem_tmp = nmem_create();
1141     char **names;
1142     int num_names = 0;
1143
1144     if (!name)
1145         name = "*";
1146
1147     nmem_strsplit(nmem_tmp, ",", name, &names, &num_names);
1148
1149     session_enter(se, "perform_termlist");
1150
1151     for (j = 0; j < num_names; j++)
1152     {
1153         const char *tname;
1154         int must_generate_empty = 1; /* bug 5350 */
1155
1156         for (i = 0; i < se->num_termlists; i++)
1157         {
1158             tname = se->termlists[i].name;
1159             if (!strcmp(names[j], tname) || !strcmp(names[j], "*"))
1160             {
1161                 struct termlist_score **p = 0;
1162                 int len;
1163
1164                 wrbuf_puts(c->wrbuf, "<list name=\"");
1165                 wrbuf_xmlputs(c->wrbuf, tname);
1166                 wrbuf_puts(c->wrbuf, "\">\n");
1167                 must_generate_empty = 0;
1168
1169                 p = termlist_highscore(se->termlists[i].termlist, &len,
1170                                        nmem_tmp);
1171                 if (p)
1172                 {
1173                     int i;
1174                     for (i = 0; i < len && i < num; i++)
1175                     {
1176                         // prevent sending empty term elements
1177                         if (!p[i]->display_term || !p[i]->display_term[0])
1178                             continue;
1179
1180                         wrbuf_puts(c->wrbuf, "<term>");
1181                         wrbuf_puts(c->wrbuf, "<name>");
1182                         wrbuf_xmlputs(c->wrbuf, p[i]->display_term);
1183                         wrbuf_puts(c->wrbuf, "</name>");
1184
1185                         wrbuf_printf(c->wrbuf,
1186                                      "<frequency>%d</frequency>",
1187                                      p[i]->frequency);
1188                         wrbuf_puts(c->wrbuf, "</term>\n");
1189                     }
1190                 }
1191                 wrbuf_puts(c->wrbuf, "</list>\n");
1192             }
1193         }
1194         tname = "xtargets";
1195         if (!strcmp(names[j], tname) || !strcmp(names[j], "*"))
1196         {
1197             wrbuf_puts(c->wrbuf, "<list name=\"");
1198             wrbuf_xmlputs(c->wrbuf, tname);
1199             wrbuf_puts(c->wrbuf, "\">\n");
1200
1201             targets_termlist_nb(c->wrbuf, se, num, c->nmem, version);
1202             wrbuf_puts(c->wrbuf, "</list>\n");
1203             must_generate_empty = 0;
1204         }
1205         if (must_generate_empty)
1206         {
1207             wrbuf_puts(c->wrbuf, "<list name=\"");
1208             wrbuf_xmlputs(c->wrbuf, names[j]);
1209             wrbuf_puts(c->wrbuf, "\"/>\n");
1210         }
1211     }
1212     session_leave(se, "perform_termlist");
1213     nmem_destroy(nmem_tmp);
1214 }
1215
1216 #ifdef MISSING_HEADERS
1217 void report_nmem_stats(void)
1218 {
1219     size_t in_use, is_free;
1220
1221     nmem_get_memory_in_use(&in_use);
1222     nmem_get_memory_free(&is_free);
1223
1224     yaz_log(YLOG_LOG, "nmem stat: use=%ld free=%ld",
1225             (long) in_use, (long) is_free);
1226 }
1227 #endif
1228
1229 struct record_cluster *show_single_start(struct session *se, const char *id,
1230                                          struct record_cluster **prev_r,
1231                                          struct record_cluster **next_r)
1232 {
1233     struct record_cluster *r = 0;
1234
1235     session_enter(se, "show_single_start");
1236     *prev_r = 0;
1237     *next_r = 0;
1238     reclist_limit(se->reclist, se);
1239
1240     reclist_enter(se->reclist);
1241     while ((r = reclist_read_record(se->reclist)))
1242     {
1243         if (!strcmp(r->recid, id))
1244         {
1245             *next_r = reclist_read_record(se->reclist);
1246             break;
1247         }
1248         *prev_r = r;
1249     }
1250     reclist_leave(se->reclist);
1251     if (!r)
1252         session_leave(se, "show_single_start");
1253     return r;
1254 }
1255
1256 void show_single_stop(struct session *se, struct record_cluster *rec)
1257 {
1258     session_leave(se, "show_single_stop");
1259 }
1260
1261
1262 int session_fetch_more(struct session *se)
1263 {
1264     struct client_list *l;
1265     int ret = 0;
1266
1267     for (l = se->clients_active; l; l = l->next)
1268     {
1269         struct client *cl = l->client;
1270         if (client_get_state(cl) == Client_Idle)
1271         {
1272             if (client_fetch_more(cl))
1273             {
1274                 session_log(se, YLOG_LOG, "%s: more to fetch",
1275                             client_get_id(cl));
1276                 ret = 1;
1277             }
1278             else
1279             {
1280                 session_log(se, YLOG_LOG, "%s: no more to fetch",
1281                             client_get_id(cl));
1282             }
1283         }
1284         else
1285         {
1286             session_log(se, YLOG_LOG, "%s: no fetch due to state=%s",
1287                         client_get_id(cl), client_get_state_str(cl));
1288         }
1289
1290     }
1291     return ret;
1292 }
1293
1294 struct record_cluster **show_range_start(struct session *se,
1295                                          struct reclist_sortparms *sp,
1296                                          int start, int *num, int *total,
1297                                          Odr_int *sumhits, Odr_int *approx_hits,
1298                                          void (*show_records_ready)(void *data),
1299                                          struct http_channel *chan)
1300 {
1301     struct record_cluster **recs = 0;
1302     struct reclist_sortparms *spp;
1303     struct client_list *l;
1304     int i;
1305 #if USE_TIMING
1306     yaz_timing_t t = yaz_timing_create();
1307 #endif
1308     session_enter(se, "show_range_start");
1309     *sumhits = 0;
1310     *approx_hits = 0;
1311     *total = 0;
1312     reclist_limit(se->reclist, se);
1313     if (se->relevance)
1314     {
1315         for (spp = sp; spp; spp = spp->next)
1316             if (spp->type == Metadata_sortkey_relevance)
1317             {
1318                 relevance_prepare_read(se->relevance, se->reclist);
1319                 break;
1320             }
1321         for (l = se->clients_active; l; l = l->next) {
1322             *sumhits += client_get_hits(l->client);
1323             *approx_hits += client_get_approximation(l->client);
1324         }
1325     }
1326     reclist_sort(se->reclist, sp);
1327
1328     reclist_enter(se->reclist);
1329     *total = reclist_get_num_records(se->reclist);
1330
1331     for (l = se->clients_active; l; l = l->next)
1332         client_update_show_stat(l->client, 0);
1333
1334     for (i = 0; i < start; i++)
1335     {
1336         struct record_cluster *r = reclist_read_record(se->reclist);
1337         if (!r)
1338         {
1339             *num = 0;
1340             break;
1341         }
1342         else
1343         {
1344             struct record *rec = r->records;
1345             for (;rec; rec = rec->next)
1346                 client_update_show_stat(rec->client, 1);
1347         }
1348     }
1349     recs = nmem_malloc(se->nmem, (*num > 0 ? *num : 1) * sizeof(*recs));
1350     for (i = 0; i < *num; i++)
1351     {
1352         struct record_cluster *r = reclist_read_record(se->reclist);
1353         if (!r)
1354         {
1355             *num = i;
1356             break;
1357         }
1358         else
1359         {
1360             struct record *rec = r->records;
1361             for (;rec; rec = rec->next)
1362                 client_update_show_stat(rec->client, 1);
1363             recs[i] = r;
1364         }
1365     }
1366     reclist_leave(se->reclist);
1367 #if USE_TIMING
1368     yaz_timing_stop(t);
1369     session_log(se, YLOG_LOG, "show %6.5f %3.2f %3.2f",
1370             yaz_timing_get_real(t), yaz_timing_get_user(t),
1371             yaz_timing_get_sys(t));
1372     yaz_timing_destroy(&t);
1373 #endif
1374
1375     if (!session_fetch_more(se))
1376         session_log(se, YLOG_LOG, "can not fetch more");
1377     else
1378     {
1379         show_range_stop(se, recs);
1380         session_log(se, YLOG_LOG, "fetching more in progress");
1381         if (session_set_watch(se, SESSION_WATCH_SHOW,
1382                               show_records_ready, chan, chan))
1383         {
1384             session_log(se, YLOG_WARN, "Ignoring show block");
1385             session_enter(se, "show_range_start");
1386         }
1387         else
1388         {
1389             session_log(se, YLOG_LOG, "session watch OK");
1390             return 0;
1391         }
1392     }
1393     return recs;
1394 }
1395
1396 void show_range_stop(struct session *se, struct record_cluster **recs)
1397 {
1398     session_leave(se, "show_range_stop");
1399 }
1400
1401 void statistics(struct session *se, struct statistics *stat)
1402 {
1403     struct client_list *l;
1404     int count = 0;
1405
1406     memset(stat, 0, sizeof(*stat));
1407     stat->num_hits = 0;
1408     for (l = se->clients_active; l; l = l->next)
1409     {
1410         struct client *cl = l->client;
1411         if (!client_get_connection(cl))
1412             stat->num_no_connection++;
1413         stat->num_hits += client_get_hits(cl);
1414         switch (client_get_state(cl))
1415         {
1416         case Client_Connecting: stat->num_connecting++; break;
1417         case Client_Working: stat->num_working++; break;
1418         case Client_Idle: stat->num_idle++; break;
1419         case Client_Failed: stat->num_failed++; break;
1420         case Client_Error: stat->num_error++; break;
1421         default: break;
1422         }
1423         count++;
1424     }
1425     stat->num_records = se->total_records;
1426
1427     stat->num_clients = count;
1428 }
1429
1430 static struct record_metadata *record_metadata_init(
1431     NMEM nmem, const char *value, enum conf_metadata_type type,
1432     struct _xmlAttr *attr)
1433 {
1434     struct record_metadata *rec_md = record_metadata_create(nmem);
1435     struct record_metadata_attr **attrp = &rec_md->attributes;
1436
1437     for (; attr; attr = attr->next)
1438     {
1439         if (attr->children && attr->children->content)
1440         {
1441             if (strcmp((const char *) attr->name, "type")
1442                 && strcmp((const char *) attr->name, "empty"))
1443             {  /* skip the "type" + "empty" attribute..
1444                   The "Type" is already part of the element in output
1445                   (md-%s) and so repeating it here is redundant */
1446                 *attrp = nmem_malloc(nmem, sizeof(**attrp));
1447                 (*attrp)->name =
1448                     nmem_strdup(nmem, (const char *) attr->name);
1449                 (*attrp)->value =
1450                     nmem_strdup(nmem, (const char *) attr->children->content);
1451                 attrp = &(*attrp)->next;
1452             }
1453         }
1454     }
1455     *attrp = 0;
1456
1457     if (type == Metadata_type_generic)
1458     {
1459         char *p = nmem_strdup(nmem, value);
1460
1461         p = normalize7bit_generic(p, " ,/.:([");
1462
1463         rec_md->data.text.disp = p;
1464         rec_md->data.text.sort = 0;
1465     }
1466     else if (type == Metadata_type_year || type == Metadata_type_date)
1467     {
1468         int first, last;
1469         int longdate = 0;
1470
1471         if (type == Metadata_type_date)
1472             longdate = 1;
1473         if (extract7bit_dates((char *) value, &first, &last, longdate) < 0)
1474             return 0;
1475
1476         rec_md->data.number.min = first;
1477         rec_md->data.number.max = last;
1478     }
1479     else
1480         return 0;
1481     return rec_md;
1482 }
1483
1484 static int get_mergekey_from_doc(xmlDoc *doc, xmlNode *root, const char *name,
1485                                  struct conf_service *service, WRBUF norm_wr)
1486 {
1487     xmlNode *n;
1488     int no_found = 0;
1489     for (n = root->children; n; n = n->next)
1490     {
1491         if (n->type != XML_ELEMENT_NODE)
1492             continue;
1493         if (!strcmp((const char *) n->name, "metadata"))
1494         {
1495             xmlChar *type = xmlGetProp(n, (xmlChar *) "type");
1496             if (type == NULL) {
1497                 yaz_log(YLOG_FATAL, "Missing type attribute on metadata element. Skipping!");
1498             }
1499             else if (!strcmp(name, (const char *) type))
1500             {
1501                 xmlChar *value = xmlNodeListGetString(doc, n->children, 1);
1502                 if (value && *value)
1503                 {
1504                     const char *norm_str;
1505                     pp2_charset_token_t prt =
1506                         pp2_charset_token_create(service->charsets, "mergekey");
1507
1508                     pp2_charset_token_first(prt, (const char *) value, 0);
1509                     if (wrbuf_len(norm_wr) > 0)
1510                         wrbuf_puts(norm_wr, " ");
1511                     wrbuf_puts(norm_wr, name);
1512                     while ((norm_str =
1513                             pp2_charset_token_next(prt)))
1514                     {
1515                         if (*norm_str)
1516                         {
1517                             wrbuf_puts(norm_wr, " ");
1518                             wrbuf_puts(norm_wr, norm_str);
1519                         }
1520                     }
1521                     pp2_charset_token_destroy(prt);
1522                     no_found++;
1523                 }
1524                 if (value)
1525                     xmlFree(value);
1526             }
1527             xmlFree(type);
1528         }
1529     }
1530     return no_found;
1531 }
1532
1533 static const char *get_mergekey(xmlDoc *doc, struct client *cl, int record_no,
1534                                 struct conf_service *service, NMEM nmem,
1535                                 const char *session_mergekey)
1536 {
1537     char *mergekey_norm = 0;
1538     xmlNode *root = xmlDocGetRootElement(doc);
1539     WRBUF norm_wr = wrbuf_alloc();
1540     xmlChar *mergekey;
1541
1542     if (session_mergekey)
1543     {
1544         int i, num = 0;
1545         char **values = 0;
1546         nmem_strsplit_escape2(nmem, ",", session_mergekey, &values,
1547                               &num, 1, '\\', 1);
1548
1549         for (i = 0; i < num; i++)
1550             get_mergekey_from_doc(doc, root, values[i], service, norm_wr);
1551     }
1552     else if ((mergekey = xmlGetProp(root, (xmlChar *) "mergekey")))
1553     {
1554         const char *norm_str;
1555         pp2_charset_token_t prt =
1556             pp2_charset_token_create(service->charsets, "mergekey");
1557
1558         pp2_charset_token_first(prt, (const char *) mergekey, 0);
1559         while ((norm_str = pp2_charset_token_next(prt)))
1560         {
1561             if (*norm_str)
1562             {
1563                 if (wrbuf_len(norm_wr))
1564                     wrbuf_puts(norm_wr, " ");
1565                 wrbuf_puts(norm_wr, norm_str);
1566             }
1567         }
1568         pp2_charset_token_destroy(prt);
1569         xmlFree(mergekey);
1570     }
1571     else
1572     {
1573         /* no mergekey defined in XSL. Look for mergekey metadata instead */
1574         int field_id;
1575         for (field_id = 0; field_id < service->num_metadata; field_id++)
1576         {
1577             struct conf_metadata *ser_md = &service->metadata[field_id];
1578             if (ser_md->mergekey != Metadata_mergekey_no)
1579             {
1580                 int r = get_mergekey_from_doc(doc, root, ser_md->name,
1581                                               service, norm_wr);
1582                 if (r == 0 && ser_md->mergekey == Metadata_mergekey_required)
1583                 {
1584                     /* no mergekey on this one and it is required..
1585                        Generate unique key instead */
1586                     wrbuf_rewind(norm_wr);
1587                     break;
1588                 }
1589             }
1590         }
1591     }
1592
1593     /* generate unique key if none is not generated already or is empty */
1594     if (wrbuf_len(norm_wr) == 0)
1595     {
1596         wrbuf_printf(norm_wr, "position: %s-%d",
1597                      client_get_id(cl), record_no);
1598     }
1599     else
1600     {
1601         const char *lead = "content: ";
1602         wrbuf_insert(norm_wr, 0, lead, strlen(lead));
1603     }
1604     if (wrbuf_len(norm_wr) > 0)
1605         mergekey_norm = nmem_strdup(nmem, wrbuf_cstr(norm_wr));
1606     wrbuf_destroy(norm_wr);
1607     return mergekey_norm;
1608 }
1609
1610 /** \brief see if metadata for pz:recordfilter exists
1611     \param root xml root element of normalized record
1612     \param sdb session database for client
1613     \retval 0 if there is no metadata for pz:recordfilter
1614     \retval 1 if there is metadata for pz:recordfilter
1615
1616     If there is no pz:recordfilter defined, this function returns 1
1617     as well.
1618 */
1619
1620 static int check_record_filter(xmlNode *root, struct session_database *sdb)
1621 {
1622     int match = 0;
1623     xmlNode *n;
1624     const char *s;
1625     s = session_setting_oneval(sdb, PZ_RECORDFILTER);
1626
1627     if (!s || !*s)
1628         return 1;
1629
1630     for (n = root->children; n; n = n->next)
1631     {
1632         if (n->type != XML_ELEMENT_NODE)
1633             continue;
1634         if (!strcmp((const char *) n->name, "metadata"))
1635         {
1636             xmlChar *type = xmlGetProp(n, (xmlChar *) "type");
1637             if (type)
1638             {
1639                 size_t len;
1640                 int substring;
1641                 const char *eq;
1642
1643                 if ((eq = strchr(s, '=')))
1644                     substring = 0;
1645                 else if ((eq = strchr(s, '~')))
1646                     substring = 1;
1647                 if (eq)
1648                     len = eq - s;
1649                 else
1650                     len = strlen(s);
1651                 if (len == strlen((const char *)type) &&
1652                     !memcmp((const char *) type, s, len))
1653                 {
1654                     xmlChar *value = xmlNodeGetContent(n);
1655                     if (value && *value)
1656                     {
1657                         if (!eq ||
1658                             (substring && strstr((const char *) value, eq+1)) ||
1659                             (!substring && !strcmp((const char *) value, eq + 1)))
1660                             match = 1;
1661                     }
1662                     xmlFree(value);
1663                 }
1664                 xmlFree(type);
1665             }
1666         }
1667     }
1668     return match;
1669 }
1670
1671
1672 static int ingest_to_cluster(struct client *cl,
1673                              xmlDoc *xdoc,
1674                              xmlNode *root,
1675                              int record_no,
1676                              const char *mergekey_norm);
1677
1678 /** \brief ingest XML record
1679     \param cl client holds the result set for record
1680     \param rec record buffer (0 terminated)
1681     \param record_no record position (1, 2, ..)
1682     \param nmem working NMEM
1683     \retval 0 OK
1684     \retval -1 failure
1685     \retval -2 Filtered
1686 */
1687 int ingest_record(struct client *cl, const char *rec,
1688                   int record_no, NMEM nmem)
1689 {
1690     struct session *se = client_get_session(cl);
1691     int ret = 0;
1692     struct session_database *sdb = client_get_database(cl);
1693     struct conf_service *service = se->service;
1694     xmlDoc *xdoc = normalize_record(se, sdb, service, rec, nmem);
1695     xmlNode *root;
1696     const char *mergekey_norm;
1697
1698     if (!xdoc)
1699         return -1;
1700
1701     root = xmlDocGetRootElement(xdoc);
1702
1703     if (!check_record_filter(root, sdb))
1704     {
1705         session_log(se, YLOG_LOG, "Filtered out record no %d from %s", record_no, sdb->database->id);
1706         xmlFreeDoc(xdoc);
1707         return -2;
1708     }
1709
1710     mergekey_norm = get_mergekey(xdoc, cl, record_no, service, nmem,
1711         se->mergekey);
1712     if (!mergekey_norm)
1713     {
1714         session_log(se, YLOG_WARN, "Got no mergekey");
1715         xmlFreeDoc(xdoc);
1716         return -1;
1717     }
1718     session_enter(se, "ingest_record");
1719     if (client_get_session(cl) == se && se->relevance)
1720         ret = ingest_to_cluster(cl, xdoc, root, record_no, mergekey_norm);
1721     session_leave(se, "ingest_record");
1722
1723     xmlFreeDoc(xdoc);
1724     return ret;
1725 }
1726
1727 //    struct conf_metadata *ser_md = &service->metadata[md_field_id];
1728 //    struct record_metadata *rec_md = record->metadata[md_field_id];
1729 static int match_metadata_local(struct conf_metadata *ser_md,
1730                                 struct record_metadata *rec_md0,
1731                                 char **values, int num_v)
1732 {
1733     int i;
1734     struct record_metadata *rec_md = rec_md0;
1735     for (i = 0; i < num_v; )
1736     {
1737         if (rec_md)
1738         {
1739             if (ser_md->type == Metadata_type_year
1740                 || ser_md->type == Metadata_type_date)
1741             {
1742                 int y = atoi(values[i]);
1743                 if (y >= rec_md->data.number.min
1744                     && y <= rec_md->data.number.max)
1745                     break;
1746             }
1747             else
1748             {
1749                 yaz_log(YLOG_DEBUG, "cmp: '%s' '%s'", rec_md->data.text.disp, values[i]);
1750                 if (!strcmp(rec_md->data.text.disp, values[i]))
1751                 {
1752                     // Value equals, should not be filtered.
1753                     break;
1754                 }
1755             }
1756             rec_md = rec_md->next;
1757         }
1758         else
1759         {
1760             rec_md = rec_md0;
1761             i++;
1762         }
1763     }
1764     return i < num_v ? 1 : 0;
1765 }
1766
1767 int session_check_cluster_limit(struct session *se, struct record_cluster *rec)
1768 {
1769     int i;
1770     struct conf_service *service = se->service;
1771     int ret = 1;
1772     const char *name;
1773     const char *value;
1774     NMEM nmem_tmp = nmem_create();
1775
1776     for (i = 0; (name = facet_limits_get(se->facet_limits, i, &value)); i++)
1777     {
1778         int j;
1779         for (j = 0; j < service->num_metadata; j++)
1780         {
1781             struct conf_metadata *md = service->metadata + j;
1782             if (!strcmp(md->name, name) && md->limitcluster)
1783             {
1784                 char **values = 0;
1785                 int num = 0;
1786                 int md_field_id =
1787                     conf_service_metadata_field_id(service,
1788                                                    md->limitcluster);
1789
1790                 if (md_field_id < 0)
1791                 {
1792                     ret = 0;
1793                     break;
1794                 }
1795
1796                 nmem_strsplit_escape2(nmem_tmp, "|", value, &values,
1797                                       &num, 1, '\\', 1);
1798
1799                 if (!match_metadata_local(&service->metadata[md_field_id],
1800                                           rec->metadata[md_field_id],
1801                                           values, num))
1802                 {
1803                     ret = 0;
1804                     break;
1805                 }
1806             }
1807         }
1808     }
1809     nmem_destroy(nmem_tmp);
1810     return ret;
1811 }
1812
1813 // Skip record on non-zero
1814 static int check_limit_local(struct client *cl,
1815                              struct record *record,
1816                              int record_no)
1817 {
1818     int skip_record = 0;
1819     struct session *se = client_get_session(cl);
1820     struct conf_service *service = se->service;
1821     NMEM nmem_tmp = nmem_create();
1822     struct session_database *sdb = client_get_database(cl);
1823     int l = 0;
1824     while (!skip_record)
1825     {
1826         int md_field_id;
1827         char **values = 0;
1828         int num_v = 0;
1829         const char *name =
1830             client_get_facet_limit_local(cl, sdb, &l, nmem_tmp,
1831                                          &num_v, &values);
1832         if (!name)
1833             break;
1834
1835         if (!strcmp(name, "*"))
1836         {
1837             for (md_field_id = 0; md_field_id < service->num_metadata;
1838                  md_field_id++)
1839             {
1840                 if (match_metadata_local(
1841                         &service->metadata[md_field_id],
1842                         record->metadata[md_field_id],
1843                         values, num_v))
1844                     break;
1845             }
1846             if (md_field_id == service->num_metadata)
1847                 skip_record = 1;
1848         }
1849         else
1850         {
1851             md_field_id = conf_service_metadata_field_id(service, name);
1852             if (md_field_id < 0)
1853             {
1854                 skip_record = 1;
1855                 break;
1856             }
1857             if (!match_metadata_local(
1858                     &service->metadata[md_field_id],
1859                     record->metadata[md_field_id],
1860                     values, num_v))
1861             {
1862                 skip_record = 1;
1863             }
1864         }
1865     }
1866     nmem_destroy(nmem_tmp);
1867     return skip_record;
1868 }
1869
1870 static int ingest_to_cluster(struct client *cl,
1871                              xmlDoc *xdoc,
1872                              xmlNode *root,
1873                              int record_no,
1874                              const char *mergekey_norm)
1875 {
1876     xmlNode *n;
1877     xmlChar *type = 0;
1878     xmlChar *value = 0;
1879     struct session *se = client_get_session(cl);
1880     struct conf_service *service = se->service;
1881     int term_factor = 1;
1882     struct record_cluster *cluster;
1883     struct record_metadata **metadata0;
1884     struct session_database *sdb = client_get_database(cl);
1885     struct record *record = record_create(se->nmem,
1886                                           service->num_metadata,
1887                                           service->num_sortkeys, cl,
1888                                           record_no);
1889
1890     for (n = root->children; n; n = n->next)
1891     {
1892         if (type)
1893             xmlFree(type);
1894         if (value)
1895             xmlFree(value);
1896         type = value = 0;
1897
1898         if (n->type != XML_ELEMENT_NODE)
1899             continue;
1900         if (!strcmp((const char *) n->name, "metadata"))
1901         {
1902             struct conf_metadata *ser_md = 0;
1903             struct record_metadata **wheretoput = 0;
1904             struct record_metadata *rec_md = 0;
1905             int md_field_id = -1;
1906
1907             type = xmlGetProp(n, (xmlChar *) "type");
1908             value = xmlNodeListGetString(xdoc, n->children, 1);
1909             if (!type)
1910                 continue;
1911             if (!value || !*value)
1912             {
1913                 xmlChar *empty = xmlGetProp(n, (xmlChar *) "empty");
1914                 if (!empty)
1915                     continue;
1916                 if (value)
1917                     xmlFree(value);
1918                 value = empty;
1919             }
1920             md_field_id
1921                 = conf_service_metadata_field_id(service, (const char *) type);
1922             if (md_field_id < 0)
1923             {
1924                 if (se->number_of_warnings_unknown_metadata == 0)
1925                 {
1926                     session_log(se, YLOG_WARN,
1927                             "Ignoring unknown metadata element: %s", type);
1928                 }
1929                 se->number_of_warnings_unknown_metadata++;
1930                 continue;
1931             }
1932
1933             ser_md = &service->metadata[md_field_id];
1934
1935             // non-merged metadata
1936             rec_md = record_metadata_init(se->nmem, (const char *) value,
1937                                           ser_md->type, n->properties);
1938             if (!rec_md)
1939             {
1940                 session_log(se, YLOG_WARN, "bad metadata data '%s' "
1941                             "for element '%s'", value, type);
1942                 continue;
1943             }
1944             wheretoput = &record->metadata[md_field_id];
1945             while (*wheretoput)
1946                 wheretoput = &(*wheretoput)->next;
1947             *wheretoput = rec_md;
1948         }
1949     }
1950
1951     if (check_limit_local(cl, record, record_no))
1952     {
1953         session_log(se, YLOG_LOG, "Facet filtered out record no %d from %s",
1954                     record_no, sdb->database->id);
1955         if (type)
1956             xmlFree(type);
1957         if (value)
1958             xmlFree(value);
1959         return -2;
1960     }
1961     cluster = reclist_insert(se->reclist, service, record,
1962                              mergekey_norm, &se->total_merged);
1963     if (!cluster)
1964         return 0; // complete match with existing record
1965
1966     {
1967         const char *use_term_factor_str =
1968             session_setting_oneval(sdb, PZ_TERMLIST_TERM_FACTOR);
1969         if (use_term_factor_str && use_term_factor_str[0] == '1')
1970         {
1971             int maxrecs = client_get_maxrecs(cl);
1972             int hits = (int) client_get_hits(cl);
1973             term_factor = MAX(hits, maxrecs) /  MAX(1, maxrecs);
1974             assert(term_factor >= 1);
1975             session_log(se, YLOG_DEBUG, "Using term factor: %d (%d / %d)",
1976                         term_factor, MAX(hits, maxrecs), MAX(1, maxrecs));
1977         }
1978     }
1979
1980     if (global_parameters.dump_records)
1981         session_log(se, YLOG_LOG, "Cluster id %s from %s (#%d)", cluster->recid,
1982                     sdb->database->id, record_no);
1983
1984
1985     relevance_newrec(se->relevance, cluster);
1986
1987     // original metadata, to check if first existence of a field
1988     metadata0 = xmalloc(sizeof(*metadata0) * service->num_metadata);
1989     memcpy(metadata0, cluster->metadata,
1990            sizeof(*metadata0) * service->num_metadata);
1991
1992     // now parsing XML record and adding data to cluster or record metadata
1993     for (n = root->children; n; n = n->next)
1994     {
1995         pp2_charset_token_t prt;
1996         if (type)
1997             xmlFree(type);
1998         if (value)
1999             xmlFree(value);
2000         type = value = 0;
2001
2002         if (n->type != XML_ELEMENT_NODE)
2003             continue;
2004         if (!strcmp((const char *) n->name, "metadata"))
2005         {
2006             struct conf_metadata *ser_md = 0;
2007             struct conf_sortkey *ser_sk = 0;
2008             struct record_metadata **wheretoput = 0;
2009             struct record_metadata *rec_md = 0;
2010             int md_field_id = -1;
2011             int sk_field_id = -1;
2012             const char *rank;
2013             xmlChar *xml_rank;
2014
2015             type = xmlGetProp(n, (xmlChar *) "type");
2016             value = xmlNodeListGetString(xdoc, n->children, 1);
2017
2018             if (!type || !value || !*value)
2019                 continue;
2020
2021             md_field_id
2022                 = conf_service_metadata_field_id(service, (const char *) type);
2023             if (md_field_id < 0)
2024                 continue;
2025
2026             ser_md = &service->metadata[md_field_id];
2027
2028             if (ser_md->sortkey_offset >= 0)
2029             {
2030                 sk_field_id = ser_md->sortkey_offset;
2031                 ser_sk = &service->sortkeys[sk_field_id];
2032             }
2033
2034             // merged metadata
2035             rec_md = record_metadata_init(se->nmem, (const char *) value,
2036                                           ser_md->type, 0);
2037
2038             // see if the field was not in cluster already (from beginning)
2039
2040             if (!rec_md)
2041                 continue;
2042
2043             xml_rank = xmlGetProp(n, (xmlChar *) "rank");
2044             rank = xml_rank ? (const char *) xml_rank : ser_md->rank;
2045
2046             wheretoput = &cluster->metadata[md_field_id];
2047
2048             if (ser_md->merge == Metadata_merge_first)
2049             {
2050                 if (!metadata0[md_field_id])
2051                 {
2052                     while (*wheretoput)
2053                         wheretoput = &(*wheretoput)->next;
2054                     *wheretoput = rec_md;
2055                 }
2056             }
2057             else if (ser_md->merge == Metadata_merge_unique)
2058             {
2059                 while (*wheretoput)
2060                 {
2061                     if (!strcmp((const char *) (*wheretoput)->data.text.disp,
2062                                 rec_md->data.text.disp))
2063                         break;
2064                     wheretoput = &(*wheretoput)->next;
2065                 }
2066                 if (!*wheretoput)
2067                     *wheretoput = rec_md;
2068             }
2069             else if (ser_md->merge == Metadata_merge_longest)
2070             {
2071                 if (!*wheretoput
2072                     || strlen(rec_md->data.text.disp)
2073                     > strlen((*wheretoput)->data.text.disp))
2074                 {
2075                     *wheretoput = rec_md;
2076                     if (ser_sk)
2077                     {
2078                         const char *sort_str = 0;
2079                         int skip_article =
2080                             ser_sk->type == Metadata_sortkey_skiparticle;
2081
2082                         if (!cluster->sortkeys[sk_field_id])
2083                             cluster->sortkeys[sk_field_id] =
2084                                 nmem_malloc(se->nmem,
2085                                             sizeof(union data_types));
2086
2087                         prt =
2088                             pp2_charset_token_create(service->charsets, "sort");
2089
2090                         pp2_charset_token_first(prt, rec_md->data.text.disp,
2091                                                 skip_article);
2092
2093                         pp2_charset_token_next(prt);
2094
2095                         sort_str = pp2_get_sort(prt);
2096
2097                         cluster->sortkeys[sk_field_id]->text.disp =
2098                             rec_md->data.text.disp;
2099                         if (!sort_str)
2100                         {
2101                             sort_str = rec_md->data.text.disp;
2102                             session_log(se, YLOG_WARN,
2103                                     "Could not make sortkey. Bug #1858");
2104                         }
2105                         cluster->sortkeys[sk_field_id]->text.sort =
2106                             nmem_strdup(se->nmem, sort_str);
2107                         pp2_charset_token_destroy(prt);
2108                     }
2109                 }
2110             }
2111             else if (ser_md->merge == Metadata_merge_all)
2112             {
2113                 while (*wheretoput)
2114                     wheretoput = &(*wheretoput)->next;
2115                 *wheretoput = rec_md;
2116             }
2117             else if (ser_md->merge == Metadata_merge_range)
2118             {
2119                 if (!*wheretoput)
2120                 {
2121                     *wheretoput = rec_md;
2122                     if (ser_sk)
2123                         cluster->sortkeys[sk_field_id]
2124                             = &rec_md->data;
2125                 }
2126                 else
2127                 {
2128                     int this_min = rec_md->data.number.min;
2129                     int this_max = rec_md->data.number.max;
2130                     if (this_min < (*wheretoput)->data.number.min)
2131                         (*wheretoput)->data.number.min = this_min;
2132                     if (this_max > (*wheretoput)->data.number.max)
2133                         (*wheretoput)->data.number.max = this_max;
2134                 }
2135             }
2136
2137             // ranking of _all_ fields enabled ...
2138             if (rank)
2139             {
2140                 relevance_countwords(se->relevance, cluster,
2141                                      (char *) value, rank, ser_md->name);
2142             }
2143
2144             // construct facets ... unless the client already has reported them
2145             if (ser_md->termlist && !client_has_facet(cl, (char *) type))
2146             {
2147                 if (ser_md->type == Metadata_type_year)
2148                 {
2149                     char year[64];
2150                     sprintf(year, "%d", rec_md->data.number.max);
2151
2152                     add_facet(se, (char *) type, year, term_factor);
2153                     if (rec_md->data.number.max != rec_md->data.number.min)
2154                     {
2155                         sprintf(year, "%d", rec_md->data.number.min);
2156                         add_facet(se, (char *) type, year, term_factor);
2157                     }
2158                 }
2159                 else
2160                     add_facet(se, (char *) type, (char *) value, term_factor);
2161             }
2162
2163             // cleaning up
2164             if (xml_rank)
2165                 xmlFree(xml_rank);
2166             xmlFree(type);
2167             xmlFree(value);
2168             type = value = 0;
2169         }
2170         else
2171         {
2172             if (se->number_of_warnings_unknown_elements == 0)
2173                 session_log(se, YLOG_WARN,
2174                         "Unexpected element in internal record: %s", n->name);
2175             se->number_of_warnings_unknown_elements++;
2176         }
2177     }
2178     if (type)
2179         xmlFree(type);
2180     if (value)
2181         xmlFree(value);
2182
2183     xfree(metadata0);
2184     relevance_donerecord(se->relevance, cluster);
2185     se->total_records++;
2186
2187     return 0;
2188 }
2189
2190 void session_log(struct session *s, int level, const char *fmt, ...)
2191 {
2192     char buf[1024];
2193     va_list ap;
2194     va_start(ap, fmt);
2195
2196     yaz_vsnprintf(buf, sizeof(buf)-30, fmt, ap);
2197     yaz_log(level, "Session %u: %s", s ? s->session_id : 0, buf);
2198
2199     va_end(ap);
2200 }
2201
2202 /*
2203  * Local variables:
2204  * c-basic-offset: 4
2205  * c-file-style: "Stroustrup"
2206  * indent-tabs-mode: nil
2207  * End:
2208  * vim: shiftwidth=4 tabstop=8 expandtab
2209  */
2210