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                                   const char *rank)
 653 {
 654     struct client_list *l;
 655     const char *field = sp->name;
 656     int increasing = sp->increasing;
 657     int type  = sp->type;
 658     int clients_research = 0;
 659
 660     session_log(se, YLOG_DEBUG, "session_sort field=%s increasing=%d type=%d",
 661                 field, increasing, type);
 662
 663     if (rank && (!se->rank || strcmp(se->rank, rank)))
 664     {
 665         /* new rank must research/reingest anyway */
 666         assert(rank);
 667         xfree(se->rank);
 668         se->rank = *rank ? xstrdup(rank) : 0;
 669         clients_research = 1;
 670         session_log(se, YLOG_DEBUG, "session_sort: new rank = %s",
 671                     rank);
 672     }
 673     if (mergekey && (!se->mergekey || strcmp(se->mergekey, mergekey)))
 674     {
 675         /* new mergekey must research/reingest anyway */
 676         assert(mergekey);
 677         xfree(se->mergekey);
 678         se->mergekey = *mergekey ? xstrdup(mergekey) : 0;
 679         clients_research = 1;
 680         session_log(se, YLOG_DEBUG, "session_sort: new mergekey = %s",
 681                     mergekey);
 682     }
 683     if (clients_research == 0)
 684     {
 685         struct reclist_sortparms *sr;
 686         for (sr = se->sorted_results; sr; sr = sr->next)
 687             if (!reclist_sortparms_cmp(sr, sp))
 688                 break;
 689         if (sr)
 690         {
 691             session_log(se, YLOG_DEBUG, "session_sort: field=%s increasing=%d type=%d already fetched",
 692                         field, increasing, type);
 693             return;
 694         }
 695     }
 696     session_log(se, YLOG_DEBUG, "session_sort: field=%s increasing=%d type=%d must fetch",
 697                 field, increasing, type);
 698
 699     // We need to reset reclist on every sort that changes the records, not just for position
 700     // So if just one client requires new searching, we need to clear set.
 701     // Ask each of the client if sorting requires re-search due to native sort
 702     // If it does it will require us to
 703     for (l = se->clients_active; l; l = l->next)
 704     {
 705         struct client *cl = l->client;
 706         // Assume no re-search is required.
 707         client_parse_init(cl, 1);
 708         clients_research += client_parse_sort(cl, sp);
 709     }
 710     if (clients_research)
 711     {
 712         session_log(se, YLOG_DEBUG,
 713                     "session_sort: reset results due to %d clients researching",
 714                     clients_research);
 715         session_clear_set(se, sp);
 716     }
 717     else
 718     {
 719         // A new sorting based on same record set
 720         struct reclist_sortparms *sr = nmem_malloc(se->nmem, sizeof(*sr));
 721         sr->name = nmem_strdup(se->nmem, field);
 722         sr->increasing = increasing;
 723         sr->type = type;
 724         sr->next = se->sorted_results;
 725         se->sorted_results = sr;
 726         session_log(se, YLOG_DEBUG, "session_sort: no research/ingesting done");
 727         return ;
 728     }
 729     session_log(se, YLOG_DEBUG, "Re- search/ingesting for clients due to change in sort order");
 730
 731     for (l = se->clients_active; l; l = l->next)
 732     {
 733         struct client *cl = l->client;
 734         if (client_get_state(cl) == Client_Connecting ||
 735             client_get_state(cl) == Client_Idle ||
 736             client_get_state(cl) == Client_Working) {
 737             client_start_search(cl);
 738         }
 739         else
 740         {
 741             session_log(se, YLOG_DEBUG,
 742                         "session_sort: %s: No re-start/ingest in show. "
 743                         "Wrong client state: %d",
 744                         client_get_id(cl), client_get_state(cl));
 745         }
 746     }
 747 }
 748
 749 void session_sort(struct session *se, struct reclist_sortparms *sp,
 750                   const char *mergekey, const char *rank)
 751 {
 752     //session_enter(se, "session_sort");
 753     session_sort_unlocked(se, sp, mergekey, rank);
 754     //session_leave(se, "session_sort");
 755 }
 756
 757
 758 enum pazpar2_error_code session_search(struct session *se,
 759                                        const char *query,
 760                                        const char *startrecs,
 761                                        const char *maxrecs,
 762                                        const char *filter,
 763                                        const char *limit,
 764                                        const char **addinfo,
 765                                        struct reclist_sortparms *sp,
 766                                        const char *mergekey,
 767                                        const char *rank)
 768 {
 769     int live_channels = 0;
 770     int no_working = 0;
 771     int no_failed_query = 0;
 772     int no_failed_limit = 0;
 773     struct client_list *l, *l0;
 774
 775     session_log(se, YLOG_DEBUG, "Search");
 776
 777     *addinfo = 0;
 778
 779     if (se->settings_modified) {
 780         session_remove_cached_clients(se);
 781     }
 782     else
 783         session_reset_active_clients(se, 0);
 784
 785     session_enter(se, "session_search");
 786     se->settings_modified = 0;
 787
 788     if (mergekey)
 789     {
 790         xfree(se->mergekey);
 791         se->mergekey = *mergekey ? xstrdup(mergekey) : 0;
 792     }
 793     if (rank)
 794     {
 795         xfree(se->rank);
 796         se->rank = *rank ? xstrdup(rank) : 0;
 797     }
 798
 799     session_clear_set(se, sp);
 800     relevance_destroy(&se->relevance);
 801
 802     live_channels = select_targets(se, filter);
 803     if (!live_channels)
 804     {
 805         session_leave(se, "session_search");
 806         return PAZPAR2_NO_TARGETS;
 807     }
 808
 809     facet_limits_destroy(se->facet_limits);
 810     se->facet_limits = facet_limits_create(limit);
 811     if (!se->facet_limits)
 812     {
 813         *addinfo = "limit";
 814         session_leave(se, "session_search");
 815         return PAZPAR2_MALFORMED_PARAMETER_VALUE;
 816     }
 817
 818     l0 = se->clients_active;
 819     se->clients_active = 0;
 820     session_leave(se, "session_search");
 821
 822     for (l = l0; l; l = l->next)
 823     {
 824         int parse_ret;
 825         struct client *cl = l->client;
 826         client_parse_init(cl, 1);
 827         if (prepare_map(se, client_get_database(cl)) < 0)
 828             continue;
 829
 830         parse_ret = client_parse_query(cl, query, se->facet_limits);
 831         if (parse_ret == -1)
 832             no_failed_query++;
 833         else if (parse_ret == -2)
 834             no_failed_limit++;
 835         else if (parse_ret < 0)
 836             no_working++; /* other error, such as bad CCL map */
 837         else
 838         {
 839             client_parse_range(cl, startrecs, maxrecs);
 840             client_parse_sort(cl, sp);
 841             client_start_search(cl);
 842             no_working++;
 843         }
 844     }
 845     session_reset_active_clients(se, l0);
 846
 847     if (no_working == 0)
 848     {
 849         if (no_failed_query > 0)
 850         {
 851             *addinfo = "query";
 852             return PAZPAR2_MALFORMED_PARAMETER_VALUE;
 853         }
 854         else if (no_failed_limit > 0)
 855         {
 856             *addinfo = "limit";
 857             return PAZPAR2_MALFORMED_PARAMETER_VALUE;
 858         }
 859         else
 860             return PAZPAR2_NO_TARGETS;
 861     }
 862     session_log(se, YLOG_LOG, "session_start_search done");
 863     return PAZPAR2_NO_ERROR;
 864 }
 865
 866 // Creates a new session_database object for a database
 867 static void session_init_databases_fun(void *context, struct database *db)
 868 {
 869     struct session *se = (struct session *) context;
 870     struct session_database *new = nmem_malloc(se->session_nmem, sizeof(*new));
 871     int i;
 872
 873     new->database = db;
 874
 875     new->map = 0;
 876     assert(db->settings);
 877     new->settings = nmem_malloc(se->session_nmem,
 878                                 sizeof(struct settings *) * db->num_settings);
 879     new->num_settings = db->num_settings;
 880     for (i = 0; i < db->num_settings; i++)
 881     {
 882         struct setting *setting = db->settings[i];
 883         new->settings[i] = setting;
 884     }
 885     new->next = se->databases;
 886     se->databases = new;
 887 }
 888
 889 // Doesn't free memory associated with sdb -- nmem takes care of that
 890 static void session_database_destroy(struct session_database *sdb)
 891 {
 892     sdb->map = 0;
 893 }
 894
 895 // Initialize session_database list -- this represents this session's view
 896 // of the database list -- subject to modification by the settings ws command
 897 void session_init_databases(struct session *se)
 898 {
 899     se->databases = 0;
 900     predef_grep_databases(se, se->service, session_init_databases_fun);
 901 }
 902
 903 // Probably session_init_databases_fun should be refactored instead of
 904 // called here.
 905 static struct session_database *load_session_database(struct session *se,
 906                                                       char *id)
 907 {
 908     struct database *db = new_database_inherit_settings(id, se->session_nmem, se->service->settings);
 909     session_init_databases_fun((void*) se, db);
 910
 911     // New sdb is head of se->databases list
 912     return se->databases;
 913 }
 914
 915 // Find an existing session database. If not found, load it
 916 static struct session_database *find_session_database(struct session *se,
 917                                                       char *id)
 918 {
 919     struct session_database *sdb;
 920
 921     for (sdb = se->databases; sdb; sdb = sdb->next)
 922         if (!strcmp(sdb->database->id, id))
 923             return sdb;
 924     return load_session_database(se, id);
 925 }
 926
 927 // Apply a session override to a database
 928 void session_apply_setting(struct session *se, char *dbname, char *setting,
 929                            char *value)
 930 {
 931     struct session_database *sdb = find_session_database(se, dbname);
 932     struct conf_service *service = se->service;
 933     struct setting *new = nmem_malloc(se->session_nmem, sizeof(*new));
 934     int offset = settings_create_offset(service, setting);
 935
 936     expand_settings_array(&sdb->settings, &sdb->num_settings, offset,
 937                           se->session_nmem);
 938     new->precedence = 0;
 939     new->target = dbname;
 940     new->name = setting;
 941     new->value = value;
 942     new->next = sdb->settings[offset];
 943     sdb->settings[offset] = new;
 944
 945     se->settings_modified = 1;
 946
 947     // Force later recompute of settings-driven data structures
 948     // (happens when a search starts and client connections are prepared)
 949     switch (offset)
 950     {
 951     case PZ_XSLT:
 952         if (sdb->map)
 953         {
 954             sdb->map = 0;
 955         }
 956         break;
 957     }
 958 }
 959
 960 void session_destroy(struct session *se)
 961 {
 962     struct session_database *sdb;
 963     session_log(se, YLOG_DEBUG, "Destroying");
 964     session_use(-1);
 965     session_remove_cached_clients(se);
 966
 967     for (sdb = se->databases; sdb; sdb = sdb->next)
 968         session_database_destroy(sdb);
 969     normalize_cache_destroy(se->normalize_cache);
 970     relevance_destroy(&se->relevance);
 971     reclist_destroy(se->reclist);
 972     xfree(se->mergekey);
 973     xfree(se->rank);
 974     if (nmem_total(se->nmem))
 975         session_log(se, YLOG_DEBUG, "NMEN operation usage %zd", nmem_total(se->nmem));
 976     if (nmem_total(se->session_nmem))
 977         session_log(se, YLOG_DEBUG, "NMEN session usage %zd", nmem_total(se->session_nmem));
 978     facet_limits_destroy(se->facet_limits);
 979     nmem_destroy(se->nmem);
 980     service_destroy(se->service);
 981     yaz_mutex_destroy(&se->session_mutex);
 982 }
 983
 984 size_t session_get_memory_status(struct session *session) {
 985     size_t session_nmem;
 986     if (session == 0)
 987         return 0;
 988     session_enter(session, "session_get_memory_status");
 989     session_nmem = nmem_total(session->nmem);
 990     session_leave(session, "session_get_memory_status");
 991     return session_nmem;
 992 }
 993
 994
 995 struct session *new_session(NMEM nmem, struct conf_service *service,
 996                             unsigned session_id)
 997 {
 998     int i;
 999     struct session *session = nmem_malloc(nmem, sizeof(*session));
1000
1001     char tmp_str[50];
1002
1003     sprintf(tmp_str, "session#%u", session_id);
1004
1005     session->session_id = session_id;
1006     session_log(session, YLOG_DEBUG, "New");
1007     session->service = service;
1008     session->relevance = 0;
1009     session->total_records = 0;
1010     session->number_of_warnings_unknown_elements = 0;
1011     session->number_of_warnings_unknown_metadata = 0;
1012     session->num_termlists = 0;
1013     session->reclist = reclist_create(nmem);
1014     session->clients_active = 0;
1015     session->clients_cached = 0;
1016     session->settings_modified = 0;
1017     session->session_nmem = nmem;
1018     session->nmem = nmem_create();
1019     session->databases = 0;
1020     session->sorted_results = 0;
1021     session->facet_limits = 0;
1022     session->mergekey = 0;
1023     session->rank = 0;
1024
1025     for (i = 0; i <= SESSION_WATCH_MAX; i++)
1026     {
1027         session->watchlist[i].data = 0;
1028         session->watchlist[i].fun = 0;
1029     }
1030     session->normalize_cache = normalize_cache_create();
1031     session->session_mutex = 0;
1032     pazpar2_mutex_create(&session->session_mutex, tmp_str);
1033     session_use(1);
1034     return session;
1035 }
1036
1037 const char * client_get_suggestions_xml(struct client *cl, WRBUF wrbuf);
1038
1039 static struct hitsbytarget *hitsbytarget_nb(struct session *se,
1040                                             int *count, NMEM nmem)
1041 {
1042     struct hitsbytarget *res = 0;
1043     struct client_list *l;
1044     size_t sz = 0;
1045
1046     for (l = se->clients_active; l; l = l->next)
1047         sz++;
1048
1049     res = nmem_malloc(nmem, sizeof(*res) * sz);
1050     *count = 0;
1051     for (l = se->clients_active; l; l = l->next)
1052     {
1053         struct client *cl = l->client;
1054         WRBUF w = wrbuf_alloc();
1055         const char *name = session_setting_oneval(client_get_database(cl),
1056                                                   PZ_NAME);
1057
1058         res[*count].id = client_get_id(cl);
1059         res[*count].name = *name ? name : "Unknown";
1060         res[*count].hits = client_get_hits(cl);
1061         res[*count].approximation = client_get_approximation(cl);
1062         res[*count].records = client_get_num_records(cl);
1063         res[*count].filtered = client_get_num_records_filtered(cl);
1064         res[*count].diagnostic =
1065             client_get_diagnostic(cl, &res[*count].message,
1066                                   &res[*count].addinfo);
1067         res[*count].state = client_get_state_str(cl);
1068         res[*count].connected  = client_get_connection(cl) ? 1 : 0;
1069         session_settings_dump(se, client_get_database(cl), w);
1070         res[*count].settings_xml = nmem_strdup(nmem, wrbuf_cstr(w));
1071         wrbuf_rewind(w);
1072         wrbuf_puts(w, "");
1073         res[*count].suggestions_xml = nmem_strdup(nmem, client_get_suggestions_xml(cl, w));
1074         wrbuf_destroy(w);
1075         (*count)++;
1076     }
1077     return res;
1078 }
1079
1080 struct hitsbytarget *get_hitsbytarget(struct session *se, int *count, NMEM nmem)
1081 {
1082     struct hitsbytarget *p;
1083     session_enter(se, "get_hitsbytarget");
1084     p = hitsbytarget_nb(se, count, nmem);
1085     session_leave(se, "get_hitsbytarget");
1086     return p;
1087 }
1088
1089 // Compares two hitsbytarget nodes by hitcount
1090 static int cmp_ht(const void *p1, const void *p2)
1091 {
1092     const struct hitsbytarget *h1 = p1;
1093     const struct hitsbytarget *h2 = p2;
1094     return h2->hits - h1->hits;
1095 }
1096
1097 // Compares two hitsbytarget nodes by hitcount
1098 static int cmp_ht_approx(const void *p1, const void *p2)
1099 {
1100     const struct hitsbytarget *h1 = p1;
1101     const struct hitsbytarget *h2 = p2;
1102     return h2->approximation - h1->approximation;
1103 }
1104
1105 static int targets_termlist_nb(WRBUF wrbuf, struct session *se, int num,
1106                                NMEM nmem, int version)
1107 {
1108     struct hitsbytarget *ht;
1109     int count, i;
1110
1111     ht = hitsbytarget_nb(se, &count, nmem);
1112     if (version >= 2)
1113         qsort(ht, count, sizeof(struct hitsbytarget), cmp_ht_approx);
1114     else
1115         qsort(ht, count, sizeof(struct hitsbytarget), cmp_ht);
1116     for (i = 0; i < count && i < num && ht[i].hits > 0; i++)
1117     {
1118
1119         // do only print terms which have display names
1120
1121         wrbuf_puts(wrbuf, "<term>\n");
1122
1123         wrbuf_puts(wrbuf, "<id>");
1124         wrbuf_xmlputs(wrbuf, ht[i].id);
1125         wrbuf_puts(wrbuf, "</id>\n");
1126
1127         wrbuf_puts(wrbuf, "<name>");
1128         if (!ht[i].name || !ht[i].name[0])
1129             wrbuf_xmlputs(wrbuf, "NO TARGET NAME");
1130         else
1131             wrbuf_xmlputs(wrbuf, ht[i].name);
1132         wrbuf_puts(wrbuf, "</name>\n");
1133
1134         wrbuf_printf(wrbuf, "<frequency>" ODR_INT_PRINTF "</frequency>\n",
1135                      ht[i].hits);
1136
1137         if (version >= 2) {
1138             // Should not print if we know it isn't a approximation.
1139             wrbuf_printf(wrbuf, "<approximation>" ODR_INT_PRINTF "</approximation>\n", ht[i].approximation);
1140             wrbuf_printf(wrbuf, "<records>%d</records>\n", ht[i].records - ht[i].filtered);
1141             wrbuf_printf(wrbuf, "<filtered>%d</filtered>\n", ht[i].filtered);
1142         }
1143
1144         wrbuf_puts(wrbuf, "<state>");
1145         wrbuf_xmlputs(wrbuf, ht[i].state);
1146         wrbuf_puts(wrbuf, "</state>\n");
1147
1148         wrbuf_printf(wrbuf, "<diagnostic>%d</diagnostic>\n",
1149                      ht[i].diagnostic);
1150         wrbuf_puts(wrbuf, "</term>\n");
1151     }
1152     return count;
1153 }
1154
1155 void perform_termlist(struct http_channel *c, struct session *se,
1156                       const char *name, int num, int version)
1157 {
1158     int i, j;
1159     NMEM nmem_tmp = nmem_create();
1160     char **names;
1161     int num_names = 0;
1162
1163     if (!name)
1164         name = "*";
1165
1166     nmem_strsplit(nmem_tmp, ",", name, &names, &num_names);
1167
1168     session_enter(se, "perform_termlist");
1169
1170     for (j = 0; j < num_names; j++)
1171     {
1172         const char *tname;
1173         int must_generate_empty = 1; /* bug 5350 */
1174
1175         for (i = 0; i < se->num_termlists; i++)
1176         {
1177             tname = se->termlists[i].name;
1178             if (!strcmp(names[j], tname) || !strcmp(names[j], "*"))
1179             {
1180                 struct termlist_score **p = 0;
1181                 int len;
1182
1183                 wrbuf_puts(c->wrbuf, "<list name=\"");
1184                 wrbuf_xmlputs(c->wrbuf, tname);
1185                 wrbuf_puts(c->wrbuf, "\">\n");
1186                 must_generate_empty = 0;
1187
1188                 p = termlist_highscore(se->termlists[i].termlist, &len,
1189                                        nmem_tmp);
1190                 if (p)
1191                 {
1192                     int i;
1193                     for (i = 0; i < len && i < num; i++)
1194                     {
1195                         // prevent sending empty term elements
1196                         if (!p[i]->display_term || !p[i]->display_term[0])
1197                             continue;
1198
1199                         wrbuf_puts(c->wrbuf, "<term>");
1200                         wrbuf_puts(c->wrbuf, "<name>");
1201                         wrbuf_xmlputs(c->wrbuf, p[i]->display_term);
1202                         wrbuf_puts(c->wrbuf, "</name>");
1203
1204                         wrbuf_printf(c->wrbuf,
1205                                      "<frequency>%d</frequency>",
1206                                      p[i]->frequency);
1207                         wrbuf_puts(c->wrbuf, "</term>\n");
1208                     }
1209                 }
1210                 wrbuf_puts(c->wrbuf, "</list>\n");
1211             }
1212         }
1213         tname = "xtargets";
1214         if (!strcmp(names[j], tname) || !strcmp(names[j], "*"))
1215         {
1216             wrbuf_puts(c->wrbuf, "<list name=\"");
1217             wrbuf_xmlputs(c->wrbuf, tname);
1218             wrbuf_puts(c->wrbuf, "\">\n");
1219
1220             targets_termlist_nb(c->wrbuf, se, num, c->nmem, version);
1221             wrbuf_puts(c->wrbuf, "</list>\n");
1222             must_generate_empty = 0;
1223         }
1224         if (must_generate_empty)
1225         {
1226             wrbuf_puts(c->wrbuf, "<list name=\"");
1227             wrbuf_xmlputs(c->wrbuf, names[j]);
1228             wrbuf_puts(c->wrbuf, "\"/>\n");
1229         }
1230     }
1231     session_leave(se, "perform_termlist");
1232     nmem_destroy(nmem_tmp);
1233 }
1234
1235 #ifdef MISSING_HEADERS
1236 void report_nmem_stats(void)
1237 {
1238     size_t in_use, is_free;
1239
1240     nmem_get_memory_in_use(&in_use);
1241     nmem_get_memory_free(&is_free);
1242
1243     yaz_log(YLOG_LOG, "nmem stat: use=%ld free=%ld",
1244             (long) in_use, (long) is_free);
1245 }
1246 #endif
1247
1248 struct record_cluster *show_single_start(struct session *se, const char *id,
1249                                          struct record_cluster **prev_r,
1250                                          struct record_cluster **next_r)
1251 {
1252     struct record_cluster *r = 0;
1253
1254     session_enter(se, "show_single_start");
1255     *prev_r = 0;
1256     *next_r = 0;
1257     reclist_limit(se->reclist, se);
1258
1259     reclist_enter(se->reclist);
1260     while ((r = reclist_read_record(se->reclist)))
1261     {
1262         if (!strcmp(r->recid, id))
1263         {
1264             *next_r = reclist_read_record(se->reclist);
1265             break;
1266         }
1267         *prev_r = r;
1268     }
1269     reclist_leave(se->reclist);
1270     if (!r)
1271         session_leave(se, "show_single_start");
1272     return r;
1273 }
1274
1275 void show_single_stop(struct session *se, struct record_cluster *rec)
1276 {
1277     session_leave(se, "show_single_stop");
1278 }
1279
1280
1281 int session_fetch_more(struct session *se)
1282 {
1283     struct client_list *l;
1284     int ret = 0;
1285
1286     for (l = se->clients_active; l; l = l->next)
1287     {
1288         struct client *cl = l->client;
1289         if (client_get_state(cl) == Client_Idle)
1290         {
1291             if (client_fetch_more(cl))
1292             {
1293                 session_log(se, YLOG_LOG, "%s: more to fetch",
1294                             client_get_id(cl));
1295                 ret = 1;
1296             }
1297             else
1298             {
1299                 session_log(se, YLOG_LOG, "%s: no more to fetch",
1300                             client_get_id(cl));
1301             }
1302         }
1303         else
1304         {
1305             session_log(se, YLOG_LOG, "%s: no fetch due to state=%s",
1306                         client_get_id(cl), client_get_state_str(cl));
1307         }
1308
1309     }
1310     return ret;
1311 }
1312
1313 struct record_cluster **show_range_start(struct session *se,
1314                                          struct reclist_sortparms *sp,
1315                                          int start, int *num, int *total,
1316                                          Odr_int *sumhits, Odr_int *approx_hits,
1317                                          void (*show_records_ready)(void *data),
1318                                          struct http_channel *chan)
1319 {
1320     struct record_cluster **recs = 0;
1321     struct reclist_sortparms *spp;
1322     struct client_list *l;
1323     int i;
1324 #if USE_TIMING
1325     yaz_timing_t t = yaz_timing_create();
1326 #endif
1327     session_enter(se, "show_range_start");
1328     *sumhits = 0;
1329     *approx_hits = 0;
1330     *total = 0;
1331     reclist_limit(se->reclist, se);
1332     if (se->relevance)
1333     {
1334         for (spp = sp; spp; spp = spp->next)
1335             if (spp->type == Metadata_sortkey_relevance)
1336             {
1337                 relevance_prepare_read(se->relevance, se->reclist);
1338                 break;
1339             }
1340         for (l = se->clients_active; l; l = l->next) {
1341             *sumhits += client_get_hits(l->client);
1342             *approx_hits += client_get_approximation(l->client);
1343         }
1344     }
1345     reclist_sort(se->reclist, sp);
1346
1347     reclist_enter(se->reclist);
1348     *total = reclist_get_num_records(se->reclist);
1349
1350     for (l = se->clients_active; l; l = l->next)
1351         client_update_show_stat(l->client, 0);
1352
1353     for (i = 0; i < start; i++)
1354     {
1355         struct record_cluster *r = reclist_read_record(se->reclist);
1356         if (!r)
1357         {
1358             *num = 0;
1359             break;
1360         }
1361         else
1362         {
1363             struct record *rec = r->records;
1364             for (;rec; rec = rec->next)
1365                 client_update_show_stat(rec->client, 1);
1366         }
1367     }
1368     recs = nmem_malloc(se->nmem, (*num > 0 ? *num : 1) * sizeof(*recs));
1369     for (i = 0; i < *num; i++)
1370     {
1371         struct record_cluster *r = reclist_read_record(se->reclist);
1372         if (!r)
1373         {
1374             *num = i;
1375             break;
1376         }
1377         else
1378         {
1379             struct record *rec = r->records;
1380             for (;rec; rec = rec->next)
1381                 client_update_show_stat(rec->client, 1);
1382             recs[i] = r;
1383         }
1384     }
1385     reclist_leave(se->reclist);
1386 #if USE_TIMING
1387     yaz_timing_stop(t);
1388     session_log(se, YLOG_LOG, "show %6.5f %3.2f %3.2f",
1389             yaz_timing_get_real(t), yaz_timing_get_user(t),
1390             yaz_timing_get_sys(t));
1391     yaz_timing_destroy(&t);
1392 #endif
1393
1394     if (!session_fetch_more(se))
1395         session_log(se, YLOG_LOG, "can not fetch more");
1396     else
1397     {
1398         show_range_stop(se, recs);
1399         session_log(se, YLOG_LOG, "fetching more in progress");
1400         if (session_set_watch(se, SESSION_WATCH_SHOW,
1401                               show_records_ready, chan, chan))
1402         {
1403             session_log(se, YLOG_WARN, "Ignoring show block");
1404             session_enter(se, "show_range_start");
1405         }
1406         else
1407         {
1408             session_log(se, YLOG_LOG, "session watch OK");
1409             return 0;
1410         }
1411     }
1412     return recs;
1413 }
1414
1415 void show_range_stop(struct session *se, struct record_cluster **recs)
1416 {
1417     session_leave(se, "show_range_stop");
1418 }
1419
1420 void statistics(struct session *se, struct statistics *stat)
1421 {
1422     struct client_list *l;
1423     int count = 0;
1424
1425     memset(stat, 0, sizeof(*stat));
1426     stat->num_hits = 0;
1427     for (l = se->clients_active; l; l = l->next)
1428     {
1429         struct client *cl = l->client;
1430         if (!client_get_connection(cl))
1431             stat->num_no_connection++;
1432         stat->num_hits += client_get_hits(cl);
1433         switch (client_get_state(cl))
1434         {
1435         case Client_Connecting: stat->num_connecting++; break;
1436         case Client_Working: stat->num_working++; break;
1437         case Client_Idle: stat->num_idle++; break;
1438         case Client_Failed: stat->num_failed++; break;
1439         case Client_Error: stat->num_error++; break;
1440         default: break;
1441         }
1442         count++;
1443     }
1444     stat->num_records = se->total_records;
1445
1446     stat->num_clients = count;
1447 }
1448
1449 static struct record_metadata *record_metadata_init(
1450     NMEM nmem, const char *value, enum conf_metadata_type type,
1451     struct _xmlAttr *attr)
1452 {
1453     struct record_metadata *rec_md = record_metadata_create(nmem);
1454     struct record_metadata_attr **attrp = &rec_md->attributes;
1455
1456     for (; attr; attr = attr->next)
1457     {
1458         if (attr->children && attr->children->content)
1459         {
1460             if (strcmp((const char *) attr->name, "type")
1461                 && strcmp((const char *) attr->name, "empty"))
1462             {  /* skip the "type" + "empty" attribute..
1463                   The "Type" is already part of the element in output
1464                   (md-%s) and so repeating it here is redundant */
1465                 *attrp = nmem_malloc(nmem, sizeof(**attrp));
1466                 (*attrp)->name =
1467                     nmem_strdup(nmem, (const char *) attr->name);
1468                 (*attrp)->value =
1469                     nmem_strdup(nmem, (const char *) attr->children->content);
1470                 attrp = &(*attrp)->next;
1471             }
1472         }
1473     }
1474     *attrp = 0;
1475
1476     if (type == Metadata_type_generic)
1477     {
1478         char *p = nmem_strdup(nmem, value);
1479
1480         p = normalize7bit_generic(p, " ,/.:([");
1481
1482         rec_md->data.text.disp = p;
1483         rec_md->data.text.sort = 0;
1484     }
1485     else if (type == Metadata_type_year || type == Metadata_type_date)
1486     {
1487         int first, last;
1488         int longdate = 0;
1489
1490         if (type == Metadata_type_date)
1491             longdate = 1;
1492         if (extract7bit_dates((char *) value, &first, &last, longdate) < 0)
1493             return 0;
1494
1495         rec_md->data.number.min = first;
1496         rec_md->data.number.max = last;
1497     }
1498     else
1499         return 0;
1500     return rec_md;
1501 }
1502
1503 static int get_mergekey_from_doc(xmlDoc *doc, xmlNode *root, const char *name,
1504                                  struct conf_service *service, WRBUF norm_wr)
1505 {
1506     xmlNode *n;
1507     int no_found = 0;
1508     for (n = root->children; n; n = n->next)
1509     {
1510         if (n->type != XML_ELEMENT_NODE)
1511             continue;
1512         if (!strcmp((const char *) n->name, "metadata"))
1513         {
1514             xmlChar *type = xmlGetProp(n, (xmlChar *) "type");
1515             if (type == NULL) {
1516                 yaz_log(YLOG_FATAL, "Missing type attribute on metadata element. Skipping!");
1517             }
1518             else if (!strcmp(name, (const char *) type))
1519             {
1520                 xmlChar *value = xmlNodeListGetString(doc, n->children, 1);
1521                 if (value && *value)
1522                 {
1523                     const char *norm_str;
1524                     pp2_charset_token_t prt =
1525                         pp2_charset_token_create(service->charsets, "mergekey");
1526
1527                     pp2_charset_token_first(prt, (const char *) value, 0);
1528                     if (wrbuf_len(norm_wr) > 0)
1529                         wrbuf_puts(norm_wr, " ");
1530                     wrbuf_puts(norm_wr, name);
1531                     while ((norm_str =
1532                             pp2_charset_token_next(prt)))
1533                     {
1534                         if (*norm_str)
1535                         {
1536                             wrbuf_puts(norm_wr, " ");
1537                             wrbuf_puts(norm_wr, norm_str);
1538                         }
1539                     }
1540                     pp2_charset_token_destroy(prt);
1541                     no_found++;
1542                 }
1543                 if (value)
1544                     xmlFree(value);
1545             }
1546             xmlFree(type);
1547         }
1548     }
1549     return no_found;
1550 }
1551
1552 static const char *get_mergekey(xmlDoc *doc, struct client *cl, int record_no,
1553                                 struct conf_service *service, NMEM nmem,
1554                                 const char *session_mergekey)
1555 {
1556     char *mergekey_norm = 0;
1557     xmlNode *root = xmlDocGetRootElement(doc);
1558     WRBUF norm_wr = wrbuf_alloc();
1559     xmlChar *mergekey;
1560
1561     if (session_mergekey)
1562     {
1563         int i, num = 0;
1564         char **values = 0;
1565         nmem_strsplit_escape2(nmem, ",", session_mergekey, &values,
1566                               &num, 1, '\\', 1);
1567
1568         for (i = 0; i < num; i++)
1569             get_mergekey_from_doc(doc, root, values[i], service, norm_wr);
1570     }
1571     else if ((mergekey = xmlGetProp(root, (xmlChar *) "mergekey")))
1572     {
1573         const char *norm_str;
1574         pp2_charset_token_t prt =
1575             pp2_charset_token_create(service->charsets, "mergekey");
1576
1577         pp2_charset_token_first(prt, (const char *) mergekey, 0);
1578         while ((norm_str = pp2_charset_token_next(prt)))
1579         {
1580             if (*norm_str)
1581             {
1582                 if (wrbuf_len(norm_wr))
1583                     wrbuf_puts(norm_wr, " ");
1584                 wrbuf_puts(norm_wr, norm_str);
1585             }
1586         }
1587         pp2_charset_token_destroy(prt);
1588         xmlFree(mergekey);
1589     }
1590     else
1591     {
1592         /* no mergekey defined in XSL. Look for mergekey metadata instead */
1593         int field_id;
1594         for (field_id = 0; field_id < service->num_metadata; field_id++)
1595         {
1596             struct conf_metadata *ser_md = &service->metadata[field_id];
1597             if (ser_md->mergekey != Metadata_mergekey_no)
1598             {
1599                 int r = get_mergekey_from_doc(doc, root, ser_md->name,
1600                                               service, norm_wr);
1601                 if (r == 0 && ser_md->mergekey == Metadata_mergekey_required)
1602                 {
1603                     /* no mergekey on this one and it is required..
1604                        Generate unique key instead */
1605                     wrbuf_rewind(norm_wr);
1606                     break;
1607                 }
1608             }
1609         }
1610     }
1611
1612     /* generate unique key if none is not generated already or is empty */
1613     if (wrbuf_len(norm_wr) == 0)
1614     {
1615         wrbuf_printf(norm_wr, "position: %s-%d",
1616                      client_get_id(cl), record_no);
1617     }
1618     else
1619     {
1620         const char *lead = "content: ";
1621         wrbuf_insert(norm_wr, 0, lead, strlen(lead));
1622     }
1623     if (wrbuf_len(norm_wr) > 0)
1624         mergekey_norm = nmem_strdup(nmem, wrbuf_cstr(norm_wr));
1625     wrbuf_destroy(norm_wr);
1626     return mergekey_norm;
1627 }
1628
1629 /** \brief see if metadata for pz:recordfilter exists
1630     \param root xml root element of normalized record
1631     \param sdb session database for client
1632     \retval 0 if there is no metadata for pz:recordfilter
1633     \retval 1 if there is metadata for pz:recordfilter
1634
1635     If there is no pz:recordfilter defined, this function returns 1
1636     as well.
1637 */
1638
1639 static int check_record_filter(xmlNode *root, struct session_database *sdb)
1640 {
1641     int match = 0;
1642     xmlNode *n;
1643     const char *s;
1644     s = session_setting_oneval(sdb, PZ_RECORDFILTER);
1645
1646     if (!s || !*s)
1647         return 1;
1648
1649     for (n = root->children; n; n = n->next)
1650     {
1651         if (n->type != XML_ELEMENT_NODE)
1652             continue;
1653         if (!strcmp((const char *) n->name, "metadata"))
1654         {
1655             xmlChar *type = xmlGetProp(n, (xmlChar *) "type");
1656             if (type)
1657             {
1658                 size_t len;
1659                 int substring;
1660                 const char *eq;
1661
1662                 if ((eq = strchr(s, '=')))
1663                     substring = 0;
1664                 else if ((eq = strchr(s, '~')))
1665                     substring = 1;
1666                 if (eq)
1667                     len = eq - s;
1668                 else
1669                     len = strlen(s);
1670                 if (len == strlen((const char *)type) &&
1671                     !memcmp((const char *) type, s, len))
1672                 {
1673                     xmlChar *value = xmlNodeGetContent(n);
1674                     if (value && *value)
1675                     {
1676                         if (!eq ||
1677                             (substring && strstr((const char *) value, eq+1)) ||
1678                             (!substring && !strcmp((const char *) value, eq + 1)))
1679                             match = 1;
1680                     }
1681                     xmlFree(value);
1682                 }
1683                 xmlFree(type);
1684             }
1685         }
1686     }
1687     return match;
1688 }
1689
1690
1691 static int ingest_to_cluster(struct client *cl,
1692                              xmlDoc *xdoc,
1693                              xmlNode *root,
1694                              int record_no,
1695                              const char *mergekey_norm);
1696
1697 /** \brief ingest XML record
1698     \param cl client holds the result set for record
1699     \param rec record buffer (0 terminated)
1700     \param record_no record position (1, 2, ..)
1701     \param nmem working NMEM
1702     \retval 0 OK
1703     \retval -1 failure
1704     \retval -2 Filtered
1705 */
1706 int ingest_record(struct client *cl, const char *rec,
1707                   int record_no, NMEM nmem)
1708 {
1709     struct session *se = client_get_session(cl);
1710     int ret = 0;
1711     struct session_database *sdb = client_get_database(cl);
1712     struct conf_service *service = se->service;
1713     xmlDoc *xdoc = normalize_record(se, sdb, service, rec, nmem);
1714     xmlNode *root;
1715     const char *mergekey_norm;
1716
1717     if (!xdoc)
1718         return -1;
1719
1720     root = xmlDocGetRootElement(xdoc);
1721
1722     if (!check_record_filter(root, sdb))
1723     {
1724         session_log(se, YLOG_LOG, "Filtered out record no %d from %s", record_no, sdb->database->id);
1725         xmlFreeDoc(xdoc);
1726         return -2;
1727     }
1728
1729     mergekey_norm = get_mergekey(xdoc, cl, record_no, service, nmem,
1730         se->mergekey);
1731     if (!mergekey_norm)
1732     {
1733         session_log(se, YLOG_WARN, "Got no mergekey");
1734         xmlFreeDoc(xdoc);
1735         return -1;
1736     }
1737     session_enter(se, "ingest_record");
1738     if (client_get_session(cl) == se && se->relevance)
1739         ret = ingest_to_cluster(cl, xdoc, root, record_no, mergekey_norm);
1740     session_leave(se, "ingest_record");
1741
1742     xmlFreeDoc(xdoc);
1743     return ret;
1744 }
1745
1746 //    struct conf_metadata *ser_md = &service->metadata[md_field_id];
1747 //    struct record_metadata *rec_md = record->metadata[md_field_id];
1748 static int match_metadata_local(struct conf_metadata *ser_md,
1749                                 struct record_metadata *rec_md0,
1750                                 char **values, int num_v)
1751 {
1752     int i;
1753     struct record_metadata *rec_md = rec_md0;
1754     for (i = 0; i < num_v; )
1755     {
1756         if (rec_md)
1757         {
1758             if (ser_md->type == Metadata_type_year
1759                 || ser_md->type == Metadata_type_date)
1760             {
1761                 int y = atoi(values[i]);
1762                 if (y >= rec_md->data.number.min
1763                     && y <= rec_md->data.number.max)
1764                     break;
1765             }
1766             else
1767             {
1768                 yaz_log(YLOG_DEBUG, "cmp: '%s' '%s'", rec_md->data.text.disp, values[i]);
1769                 if (!strcmp(rec_md->data.text.disp, values[i]))
1770                 {
1771                     // Value equals, should not be filtered.
1772                     break;
1773                 }
1774             }
1775             rec_md = rec_md->next;
1776         }
1777         else
1778         {
1779             rec_md = rec_md0;
1780             i++;
1781         }
1782     }
1783     return i < num_v ? 1 : 0;
1784 }
1785
1786 int session_check_cluster_limit(struct session *se, struct record_cluster *rec)
1787 {
1788     int i;
1789     struct conf_service *service = se->service;
1790     int ret = 1;
1791     const char *name;
1792     const char *value;
1793     NMEM nmem_tmp = nmem_create();
1794
1795     for (i = 0; (name = facet_limits_get(se->facet_limits, i, &value)); i++)
1796     {
1797         int j;
1798         for (j = 0; j < service->num_metadata; j++)
1799         {
1800             struct conf_metadata *md = service->metadata + j;
1801             if (!strcmp(md->name, name) && md->limitcluster)
1802             {
1803                 char **values = 0;
1804                 int num = 0;
1805                 int md_field_id =
1806                     conf_service_metadata_field_id(service,
1807                                                    md->limitcluster);
1808
1809                 if (md_field_id < 0)
1810                 {
1811                     ret = 0;
1812                     break;
1813                 }
1814
1815                 nmem_strsplit_escape2(nmem_tmp, "|", value, &values,
1816                                       &num, 1, '\\', 1);
1817
1818                 if (!match_metadata_local(&service->metadata[md_field_id],
1819                                           rec->metadata[md_field_id],
1820                                           values, num))
1821                 {
1822                     ret = 0;
1823                     break;
1824                 }
1825             }
1826         }
1827     }
1828     nmem_destroy(nmem_tmp);
1829     return ret;
1830 }
1831
1832 // Skip record on non-zero
1833 static int check_limit_local(struct client *cl,
1834                              struct record *record,
1835                              int record_no)
1836 {
1837     int skip_record = 0;
1838     struct session *se = client_get_session(cl);
1839     struct conf_service *service = se->service;
1840     NMEM nmem_tmp = nmem_create();
1841     struct session_database *sdb = client_get_database(cl);
1842     int l = 0;
1843     while (!skip_record)
1844     {
1845         int md_field_id;
1846         char **values = 0;
1847         int num_v = 0;
1848         const char *name =
1849             client_get_facet_limit_local(cl, sdb, &l, nmem_tmp,
1850                                          &num_v, &values);
1851         if (!name)
1852             break;
1853
1854         if (!strcmp(name, "*"))
1855         {
1856             for (md_field_id = 0; md_field_id < service->num_metadata;
1857                  md_field_id++)
1858             {
1859                 if (match_metadata_local(
1860                         &service->metadata[md_field_id],
1861                         record->metadata[md_field_id],
1862                         values, num_v))
1863                     break;
1864             }
1865             if (md_field_id == service->num_metadata)
1866                 skip_record = 1;
1867         }
1868         else
1869         {
1870             md_field_id = conf_service_metadata_field_id(service, name);
1871             if (md_field_id < 0)
1872             {
1873                 skip_record = 1;
1874                 break;
1875             }
1876             if (!match_metadata_local(
1877                     &service->metadata[md_field_id],
1878                     record->metadata[md_field_id],
1879                     values, num_v))
1880             {
1881                 skip_record = 1;
1882             }
1883         }
1884     }
1885     nmem_destroy(nmem_tmp);
1886     return skip_record;
1887 }
1888
1889 static int ingest_to_cluster(struct client *cl,
1890                              xmlDoc *xdoc,
1891                              xmlNode *root,
1892                              int record_no,
1893                              const char *mergekey_norm)
1894 {
1895     xmlNode *n;
1896     xmlChar *type = 0;
1897     xmlChar *value = 0;
1898     struct session *se = client_get_session(cl);
1899     struct conf_service *service = se->service;
1900     int term_factor = 1;
1901     struct record_cluster *cluster;
1902     struct record_metadata **metadata0;
1903     struct session_database *sdb = client_get_database(cl);
1904     NMEM ingest_nmem = 0;
1905     char **rank_values = 0;
1906     int rank_num = 0;
1907     struct record *record = record_create(se->nmem,
1908                                           service->num_metadata,
1909                                           service->num_sortkeys, cl,
1910                                           record_no);
1911
1912     for (n = root->children; n; n = n->next)
1913     {
1914         if (type)
1915             xmlFree(type);
1916         if (value)
1917             xmlFree(value);
1918         type = value = 0;
1919
1920         if (n->type != XML_ELEMENT_NODE)
1921             continue;
1922         if (!strcmp((const char *) n->name, "metadata"))
1923         {
1924             struct conf_metadata *ser_md = 0;
1925             struct record_metadata **wheretoput = 0;
1926             struct record_metadata *rec_md = 0;
1927             int md_field_id = -1;
1928
1929             type = xmlGetProp(n, (xmlChar *) "type");
1930             value = xmlNodeListGetString(xdoc, n->children, 1);
1931             if (!type)
1932                 continue;
1933             if (!value || !*value)
1934             {
1935                 xmlChar *empty = xmlGetProp(n, (xmlChar *) "empty");
1936                 if (!empty)
1937                     continue;
1938                 if (value)
1939                     xmlFree(value);
1940                 value = empty;
1941             }
1942             md_field_id
1943                 = conf_service_metadata_field_id(service, (const char *) type);
1944             if (md_field_id < 0)
1945             {
1946                 if (se->number_of_warnings_unknown_metadata == 0)
1947                 {
1948                     session_log(se, YLOG_WARN,
1949                             "Ignoring unknown metadata element: %s", type);
1950                 }
1951                 se->number_of_warnings_unknown_metadata++;
1952                 continue;
1953             }
1954
1955             ser_md = &service->metadata[md_field_id];
1956
1957             // non-merged metadata
1958             rec_md = record_metadata_init(se->nmem, (const char *) value,
1959                                           ser_md->type, n->properties);
1960             if (!rec_md)
1961             {
1962                 session_log(se, YLOG_WARN, "bad metadata data '%s' "
1963                             "for element '%s'", value, type);
1964                 continue;
1965             }
1966             wheretoput = &record->metadata[md_field_id];
1967             while (*wheretoput)
1968                 wheretoput = &(*wheretoput)->next;
1969             *wheretoput = rec_md;
1970         }
1971     }
1972
1973     if (check_limit_local(cl, record, record_no))
1974     {
1975         session_log(se, YLOG_LOG, "Facet filtered out record no %d from %s",
1976                     record_no, sdb->database->id);
1977         if (type)
1978             xmlFree(type);
1979         if (value)
1980             xmlFree(value);
1981         return -2;
1982     }
1983     cluster = reclist_insert(se->reclist, service, record,
1984                              mergekey_norm, &se->total_merged);
1985     if (!cluster)
1986         return 0; // complete match with existing record
1987
1988     {
1989         const char *use_term_factor_str =
1990             session_setting_oneval(sdb, PZ_TERMLIST_TERM_FACTOR);
1991         if (use_term_factor_str && use_term_factor_str[0] == '1')
1992         {
1993             int maxrecs = client_get_maxrecs(cl);
1994             int hits = (int) client_get_hits(cl);
1995             term_factor = MAX(hits, maxrecs) /  MAX(1, maxrecs);
1996             assert(term_factor >= 1);
1997             session_log(se, YLOG_DEBUG, "Using term factor: %d (%d / %d)",
1998                         term_factor, MAX(hits, maxrecs), MAX(1, maxrecs));
1999         }
2000     }
2001
2002     if (global_parameters.dump_records)
2003         session_log(se, YLOG_LOG, "Cluster id %s from %s (#%d)", cluster->recid,
2004                     sdb->database->id, record_no);
2005
2006
2007     relevance_newrec(se->relevance, cluster);
2008
2009     // original metadata, to check if first existence of a field
2010     metadata0 = xmalloc(sizeof(*metadata0) * service->num_metadata);
2011     memcpy(metadata0, cluster->metadata,
2012            sizeof(*metadata0) * service->num_metadata);
2013
2014     ingest_nmem = nmem_create();
2015     if (se->rank)
2016     {
2017         yaz_log(YLOG_LOG, "local in sort : %s", se->rank);
2018         nmem_strsplit_escape2(ingest_nmem, ",", se->rank, &rank_values,
2019                               &rank_num, 1, '\\', 1);
2020     }
2021
2022     // now parsing XML record and adding data to cluster or record metadata
2023     for (n = root->children; n; n = n->next)
2024     {
2025         pp2_charset_token_t prt;
2026         if (type)
2027             xmlFree(type);
2028         if (value)
2029             xmlFree(value);
2030         type = value = 0;
2031
2032         if (n->type != XML_ELEMENT_NODE)
2033             continue;
2034         if (!strcmp((const char *) n->name, "metadata"))
2035         {
2036             struct conf_metadata *ser_md = 0;
2037             struct conf_sortkey *ser_sk = 0;
2038             struct record_metadata **wheretoput = 0;
2039             struct record_metadata *rec_md = 0;
2040             int md_field_id = -1;
2041             int sk_field_id = -1;
2042             const char *rank = 0;
2043             xmlChar *xml_rank = 0;
2044
2045             type = xmlGetProp(n, (xmlChar *) "type");
2046             value = xmlNodeListGetString(xdoc, n->children, 1);
2047
2048             if (!type || !value || !*value)
2049                 continue;
2050
2051             md_field_id
2052                 = conf_service_metadata_field_id(service, (const char *) type);
2053             if (md_field_id < 0)
2054                 continue;
2055
2056             ser_md = &service->metadata[md_field_id];
2057
2058             if (ser_md->sortkey_offset >= 0)
2059             {
2060                 sk_field_id = ser_md->sortkey_offset;
2061                 ser_sk = &service->sortkeys[sk_field_id];
2062             }
2063
2064             // merged metadata
2065             rec_md = record_metadata_init(se->nmem, (const char *) value,
2066                                           ser_md->type, 0);
2067
2068             // see if the field was not in cluster already (from beginning)
2069
2070             if (!rec_md)
2071                 continue;
2072
2073             if (rank_num)
2074             {
2075                 int i;
2076                 for (i = 0; i < rank_num; i++)
2077                 {
2078                     const char *val = rank_values[i];
2079                     const char *cp = strchr(val, '=');
2080                     if (!cp)
2081                         continue;
2082                     if ((cp - val) == strlen((const char *) type)
2083                         && !memcmp(val, type, cp - val))
2084                     {
2085                         rank = cp + 1;
2086                         break;
2087                     }
2088                 }
2089             }
2090             else
2091             {
2092                 xml_rank = xmlGetProp(n, (xmlChar *) "rank");
2093                 rank = xml_rank ? (const char *) xml_rank : ser_md->rank;
2094             }
2095
2096             wheretoput = &cluster->metadata[md_field_id];
2097
2098             if (ser_md->merge == Metadata_merge_first)
2099             {
2100                 if (!metadata0[md_field_id])
2101                 {
2102                     while (*wheretoput)
2103                         wheretoput = &(*wheretoput)->next;
2104                     *wheretoput = rec_md;
2105                 }
2106             }
2107             else if (ser_md->merge == Metadata_merge_unique)
2108             {
2109                 while (*wheretoput)
2110                 {
2111                     if (!strcmp((const char *) (*wheretoput)->data.text.disp,
2112                                 rec_md->data.text.disp))
2113                         break;
2114                     wheretoput = &(*wheretoput)->next;
2115                 }
2116                 if (!*wheretoput)
2117                     *wheretoput = rec_md;
2118             }
2119             else if (ser_md->merge == Metadata_merge_longest)
2120             {
2121                 if (!*wheretoput
2122                     || strlen(rec_md->data.text.disp)
2123                     > strlen((*wheretoput)->data.text.disp))
2124                 {
2125                     *wheretoput = rec_md;
2126                     if (ser_sk)
2127                     {
2128                         const char *sort_str = 0;
2129                         int skip_article =
2130                             ser_sk->type == Metadata_sortkey_skiparticle;
2131
2132                         if (!cluster->sortkeys[sk_field_id])
2133                             cluster->sortkeys[sk_field_id] =
2134                                 nmem_malloc(se->nmem,
2135                                             sizeof(union data_types));
2136
2137                         prt =
2138                             pp2_charset_token_create(service->charsets, "sort");
2139
2140                         pp2_charset_token_first(prt, rec_md->data.text.disp,
2141                                                 skip_article);
2142
2143                         pp2_charset_token_next(prt);
2144
2145                         sort_str = pp2_get_sort(prt);
2146
2147                         cluster->sortkeys[sk_field_id]->text.disp =
2148                             rec_md->data.text.disp;
2149                         if (!sort_str)
2150                         {
2151                             sort_str = rec_md->data.text.disp;
2152                             session_log(se, YLOG_WARN,
2153                                     "Could not make sortkey. Bug #1858");
2154                         }
2155                         cluster->sortkeys[sk_field_id]->text.sort =
2156                             nmem_strdup(se->nmem, sort_str);
2157                         pp2_charset_token_destroy(prt);
2158                     }
2159                 }
2160             }
2161             else if (ser_md->merge == Metadata_merge_all)
2162             {
2163                 while (*wheretoput)
2164                     wheretoput = &(*wheretoput)->next;
2165                 *wheretoput = rec_md;
2166             }
2167             else if (ser_md->merge == Metadata_merge_range)
2168             {
2169                 if (!*wheretoput)
2170                 {
2171                     *wheretoput = rec_md;
2172                     if (ser_sk)
2173                         cluster->sortkeys[sk_field_id]
2174                             = &rec_md->data;
2175                 }
2176                 else
2177                 {
2178                     int this_min = rec_md->data.number.min;
2179                     int this_max = rec_md->data.number.max;
2180                     if (this_min < (*wheretoput)->data.number.min)
2181                         (*wheretoput)->data.number.min = this_min;
2182                     if (this_max > (*wheretoput)->data.number.max)
2183                         (*wheretoput)->data.number.max = this_max;
2184                 }
2185             }
2186
2187             // ranking of _all_ fields enabled ...
2188             if (rank)
2189             {
2190                 relevance_countwords(se->relevance, cluster,
2191                                      (char *) value, rank, ser_md->name);
2192             }
2193
2194             // construct facets ... unless the client already has reported them
2195             if (ser_md->termlist && !client_has_facet(cl, (char *) type))
2196             {
2197                 if (ser_md->type == Metadata_type_year)
2198                 {
2199                     char year[64];
2200                     sprintf(year, "%d", rec_md->data.number.max);
2201
2202                     add_facet(se, (char *) type, year, term_factor);
2203                     if (rec_md->data.number.max != rec_md->data.number.min)
2204                     {
2205                         sprintf(year, "%d", rec_md->data.number.min);
2206                         add_facet(se, (char *) type, year, term_factor);
2207                     }
2208                 }
2209                 else
2210                     add_facet(se, (char *) type, (char *) value, term_factor);
2211             }
2212
2213             // cleaning up
2214             if (xml_rank)
2215                 xmlFree(xml_rank);
2216             xmlFree(type);
2217             xmlFree(value);
2218             type = value = 0;
2219         }
2220         else
2221         {
2222             if (se->number_of_warnings_unknown_elements == 0)
2223                 session_log(se, YLOG_WARN,
2224                         "Unexpected element in internal record: %s", n->name);
2225             se->number_of_warnings_unknown_elements++;
2226         }
2227     }
2228     if (type)
2229         xmlFree(type);
2230     if (value)
2231         xmlFree(value);
2232
2233     nmem_destroy(ingest_nmem);
2234     xfree(metadata0);
2235     relevance_donerecord(se->relevance, cluster);
2236     se->total_records++;
2237
2238     return 0;
2239 }
2240
2241 void session_log(struct session *s, int level, const char *fmt, ...)
2242 {
2243     char buf[1024];
2244     va_list ap;
2245     va_start(ap, fmt);
2246
2247     yaz_vsnprintf(buf, sizeof(buf)-30, fmt, ap);
2248     yaz_log(level, "Session %u: %s", s ? s->session_id : 0, buf);
2249
2250     va_end(ap);
2251 }
2252
2253 /*
2254  * Local variables:
2255  * c-basic-offset: 4
2256  * c-file-style: "Stroustrup"
2257  * indent-tabs-mode: nil
2258  * End:
2259  * vim: shiftwidth=4 tabstop=8 expandtab
2260  */
2261