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
 101 static int session_use(int delta)
 102 {
 103     int sessions;
 104     if (!g_session_mutex)
 105         yaz_mutex_create(&g_session_mutex);
 106     yaz_mutex_enter(g_session_mutex);
 107     no_sessions += delta;
 108     sessions = no_sessions;
 109     yaz_mutex_leave(g_session_mutex);
 110     yaz_log(YLOG_DEBUG, "%s sessions=%d", delta == 0 ? "" :
 111             (delta > 0 ? "INC" : "DEC"), no_sessions);
 112     return sessions;
 113 }
 114
 115 int sessions_count(void)
 116 {
 117     return session_use(0);
 118 }
 119
 120 static void log_xml_doc(xmlDoc *doc)
 121 {
 122     FILE *lf = yaz_log_file();
 123     xmlChar *result = 0;
 124     int len = 0;
 125 #if LIBXML_VERSION >= 20600
 126     xmlDocDumpFormatMemory(doc, &result, &len, 1);
 127 #else
 128     xmlDocDumpMemory(doc, &result, &len);
 129 #endif
 130     if (lf && len)
 131     {
 132         (void) fwrite(result, 1, len, lf);
 133         fprintf(lf, "\n");
 134     }
 135     xmlFree(result);
 136 }
 137
 138 static void session_enter(struct session *s, const char *caller)
 139 {
 140     if (caller)
 141         session_log(s, YLOG_DEBUG, "Session lock by %s", caller);
 142     yaz_mutex_enter(s->session_mutex);
 143 }
 144
 145 static void session_leave(struct session *s, const char *caller)
 146 {
 147     yaz_mutex_leave(s->session_mutex);
 148     if (caller)
 149         session_log(s, YLOG_DEBUG, "Session unlock by %s", caller);
 150 }
 151
 152 static void session_normalize_facet(struct session *s,
 153                                     const char *type, const char *value,
 154                                     WRBUF display_wrbuf, WRBUF facet_wrbuf)
 155 {
 156     struct conf_service *service = s->service;
 157     pp2_charset_token_t prt;
 158     const char *facet_component;
 159     int i;
 160     const char *icu_chain_id = 0;
 161
 162     for (i = 0; i < service->num_metadata; i++)
 163         if (!strcmp((service->metadata + i)->name, type))
 164             icu_chain_id = (service->metadata + i)->facetrule;
 165     if (!icu_chain_id)
 166         icu_chain_id = "facet";
 167     prt = pp2_charset_token_create(service->charsets, icu_chain_id);
 168     if (!prt)
 169     {
 170         session_log(s, YLOG_FATAL,
 171                     "Unknown ICU chain '%s' for facet of type '%s'",
 172                 icu_chain_id, type);
 173         wrbuf_destroy(facet_wrbuf);
 174         wrbuf_destroy(display_wrbuf);
 175         return;
 176     }
 177     pp2_charset_token_first(prt, value, 0);
 178     while ((facet_component = pp2_charset_token_next(prt)))
 179     {
 180         const char *display_component;
 181         if (*facet_component)
 182         {
 183             if (wrbuf_len(facet_wrbuf))
 184                 wrbuf_puts(facet_wrbuf, " ");
 185             wrbuf_puts(facet_wrbuf, facet_component);
 186         }
 187         display_component = pp2_get_display(prt);
 188         if (display_component)
 189         {
 190             if (wrbuf_len(display_wrbuf))
 191                 wrbuf_puts(display_wrbuf, " ");
 192             wrbuf_puts(display_wrbuf, display_component);
 193         }
 194     }
 195     pp2_charset_token_destroy(prt);
 196 }
 197
 198 void add_facet(struct session *s, const char *type, const char *value, int count)
 199 {
 200     WRBUF facet_wrbuf = wrbuf_alloc();
 201     WRBUF display_wrbuf = wrbuf_alloc();
 202
 203     session_normalize_facet(s, type, value, display_wrbuf, facet_wrbuf);
 204
 205     if (wrbuf_len(facet_wrbuf))
 206     {
 207         int i;
 208         for (i = 0; i < s->num_termlists; i++)
 209             if (!strcmp(s->termlists[i].name, type))
 210                 break;
 211         if (i == s->num_termlists)
 212         {
 213             if (i == SESSION_MAX_TERMLISTS)
 214             {
 215                 session_log(s, YLOG_FATAL, "Too many termlists");
 216                 wrbuf_destroy(facet_wrbuf);
 217                 wrbuf_destroy(display_wrbuf);
 218                 return;
 219             }
 220
 221             s->termlists[i].name = nmem_strdup(s->nmem, type);
 222             s->termlists[i].termlist = termlist_create(s->nmem);
 223             s->num_termlists = i + 1;
 224         }
 225
 226 #if 0
 227         session_log(s, YLOG_LOG, "Facets for %s: %s norm:%s (%d)", type, value, wrbuf_cstr(facet_wrbuf), count);
 228 #endif
 229         termlist_insert(s->termlists[i].termlist, wrbuf_cstr(display_wrbuf),
 230                         wrbuf_cstr(facet_wrbuf), count);
 231     }
 232     wrbuf_destroy(facet_wrbuf);
 233     wrbuf_destroy(display_wrbuf);
 234 }
 235
 236 static xmlDoc *record_to_xml(struct session *se,
 237                              struct session_database *sdb, const char *rec)
 238 {
 239     struct database *db = sdb->database;
 240     xmlDoc *rdoc = 0;
 241
 242     rdoc = xmlParseMemory(rec, strlen(rec));
 243
 244     if (!rdoc)
 245     {
 246         session_log(se, YLOG_WARN, "Non-wellformed XML");
 247         return 0;
 248     }
 249
 250     if (global_parameters.dump_records)
 251     {
 252         session_log(se, YLOG_LOG, "Un-normalized record from %s", db->id);
 253         log_xml_doc(rdoc);
 254     }
 255
 256     return rdoc;
 257 }
 258
 259 #define MAX_XSLT_ARGS 16
 260
 261 // Add static values from session database settings if applicable
 262 static void insert_settings_parameters(struct session_database *sdb,
 263                                        struct conf_service *service,
 264                                        char **parms,
 265                                        NMEM nmem)
 266 {
 267     int i;
 268     int nparms = 0;
 269     int offset = 0;
 270
 271     for (i = 0; i < service->num_metadata; i++)
 272     {
 273         struct conf_metadata *md = &service->metadata[i];
 274         int setting;
 275
 276         if (md->setting == Metadata_setting_parameter &&
 277             (setting = settings_lookup_offset(service, md->name)) >= 0)
 278         {
 279             const char *val = session_setting_oneval(sdb, setting);
 280             if (val && nparms < MAX_XSLT_ARGS)
 281             {
 282                 char *buf;
 283                 int len = strlen(val);
 284                 buf = nmem_malloc(nmem, len + 3);
 285                 buf[0] = '\'';
 286                 strcpy(buf + 1, val);
 287                 buf[len+1] = '\'';
 288                 buf[len+2] = '\0';
 289                 parms[offset++] = md->name;
 290                 parms[offset++] = buf;
 291                 nparms++;
 292             }
 293         }
 294     }
 295     parms[offset] = 0;
 296 }
 297
 298 // Add static values from session database settings if applicable
 299 static void insert_settings_values(struct session_database *sdb, xmlDoc *doc,
 300     struct conf_service *service)
 301 {
 302     int i;
 303
 304     for (i = 0; i < service->num_metadata; i++)
 305     {
 306         struct conf_metadata *md = &service->metadata[i];
 307         int offset;
 308
 309         if (md->setting == Metadata_setting_postproc &&
 310             (offset = settings_lookup_offset(service, md->name)) >= 0)
 311         {
 312             const char *val = session_setting_oneval(sdb, offset);
 313             if (val)
 314             {
 315                 xmlNode *r = xmlDocGetRootElement(doc);
 316                 xmlNode *n = xmlNewTextChild(r, 0, (xmlChar *) "metadata",
 317                                              (xmlChar *) val);
 318                 xmlSetProp(n, (xmlChar *) "type", (xmlChar *) md->name);
 319             }
 320         }
 321     }
 322 }
 323
 324 static xmlDoc *normalize_record(struct session *se,
 325                                 struct session_database *sdb,
 326                                 struct conf_service *service,
 327                                 const char *rec, NMEM nmem)
 328 {
 329     xmlDoc *rdoc = record_to_xml(se, sdb, rec);
 330
 331     if (rdoc)
 332     {
 333         char *parms[MAX_XSLT_ARGS*2+1];
 334
 335         insert_settings_parameters(sdb, service, parms, nmem);
 336
 337         if (normalize_record_transform(sdb->map, &rdoc, (const char **)parms))
 338         {
 339             session_log(se, YLOG_WARN, "Normalize failed");
 340         }
 341         else
 342         {
 343             insert_settings_values(sdb, rdoc, service);
 344
 345             if (global_parameters.dump_records)
 346             {
 347                 session_log(se, YLOG_LOG, "Normalized record from %s",
 348                             sdb->database->id);
 349                 log_xml_doc(rdoc);
 350             }
 351         }
 352     }
 353     return rdoc;
 354 }
 355
 356 void session_settings_dump(struct session *se,
 357                            struct session_database *db,
 358                            WRBUF w)
 359 {
 360     if (db->settings)
 361     {
 362         int i, num = db->num_settings;
 363         for (i = 0; i < num; i++)
 364         {
 365             struct setting *s = db->settings[i];
 366             for (;s ; s = s->next)
 367             {
 368                 wrbuf_puts(w, "<set name=\"");
 369                 wrbuf_xmlputs(w, s->name);
 370                 wrbuf_puts(w, "\" value=\"");
 371                 wrbuf_xmlputs(w, s->value);
 372                 wrbuf_puts(w, "\"/>");
 373             }
 374             if (db->settings[i])
 375                 wrbuf_puts(w, "\n");
 376         }
 377     }
 378 }
 379
 380 // Retrieve first defined value for 'name' for given database.
 381 // Will be extended to take into account user associated with session
 382 const char *session_setting_oneval(struct session_database *db, int offset)
 383 {
 384     if (offset >= db->num_settings || !db->settings[offset])
 385         return "";
 386     return db->settings[offset]->value;
 387 }
 388
 389 // Prepare XSLT stylesheets for record normalization
 390 // Structures are allocated on the session_wide nmem to avoid having
 391 // to recompute this for every search. This would lead
 392 // to leaking if a single session was to repeatedly change the PZ_XSLT
 393 // setting. However, this is not a realistic use scenario.
 394 static int prepare_map(struct session *se, struct session_database *sdb)
 395 {
 396     if (sdb->settings && !sdb->map)
 397     {
 398         const char *s;
 399
 400         if (sdb->settings[PZ_XSLT] &&
 401             (s = session_setting_oneval(sdb, PZ_XSLT)))
 402         {
 403             char auto_stylesheet[256];
 404
 405             if (!strcmp(s, "auto"))
 406             {
 407                 const char *request_syntax = session_setting_oneval(
 408                     sdb, PZ_REQUESTSYNTAX);
 409                 if (request_syntax)
 410                 {
 411                     char *cp;
 412                     yaz_snprintf(auto_stylesheet, sizeof(auto_stylesheet),
 413                                  "%s.xsl", request_syntax);
 414                     for (cp = auto_stylesheet; *cp; cp++)
 415                     {
 416                         /* deliberately only consider ASCII */
 417                         if (*cp > 32 && *cp < 127)
 418                             *cp = tolower(*cp);
 419                     }
 420                     s = auto_stylesheet;
 421                 }
 422                 else
 423                 {
 424                     session_log(se, YLOG_WARN,
 425                                 "No pz:requestsyntax for auto stylesheet");
 426                 }
 427             }
 428             sdb->map = normalize_cache_get(se->normalize_cache,
 429                                            se->service, s);
 430             if (!sdb->map)
 431                 return -1;
 432         }
 433     }
 434     return 0;
 435 }
 436
 437 // called if watch should be removed because http_channel is to be destroyed
 438 static void session_watch_cancel(void *data, struct http_channel *c,
 439                                  void *data2)
 440 {
 441     struct session_watchentry *ent = data;
 442
 443     ent->fun = 0;
 444     ent->data = 0;
 445     ent->obs = 0;
 446 }
 447
 448 // set watch. Returns 0=OK, -1 if watch is already set
 449 int session_set_watch(struct session *s, int what,
 450                       session_watchfun fun, void *data,
 451                       struct http_channel *chan)
 452 {
 453     int ret;
 454     session_enter(s, "session_set_watch");
 455     if (s->watchlist[what].fun)
 456         ret = -1;
 457     else
 458     {
 459
 460         s->watchlist[what].fun = fun;
 461         s->watchlist[what].data = data;
 462         s->watchlist[what].obs = http_add_observer(chan, &s->watchlist[what],
 463                                                    session_watch_cancel);
 464         ret = 0;
 465     }
 466     session_leave(s, "session_set_watch");
 467     return ret;
 468 }
 469
 470 void session_alert_watch(struct session *s, int what)
 471 {
 472     assert(s);
 473     session_enter(s, "session_alert_watch");
 474     if (s->watchlist[what].fun)
 475     {
 476         /* our watch is no longer associated with http_channel */
 477         void *data;
 478         session_watchfun fun;
 479
 480         http_remove_observer(s->watchlist[what].obs);
 481         fun  = s->watchlist[what].fun;
 482         data = s->watchlist[what].data;
 483
 484         /* reset watch before fun is invoked - in case fun wants to set
 485            it again */
 486         s->watchlist[what].fun = 0;
 487         s->watchlist[what].data = 0;
 488         s->watchlist[what].obs = 0;
 489
 490         session_leave(s, "session_alert_watch");
 491         session_log(s, YLOG_DEBUG,
 492                     "Alert Watch: %d calling function: %p", what, fun);
 493         fun(data);
 494     }
 495     else
 496         session_leave(s,"session_alert_watch");
 497 }
 498
 499 //callback for grep_databases
 500 static void select_targets_callback(struct session *se,
 501                                     struct session_database *db)
 502 {
 503     struct client *cl;
 504     struct client_list *l;
 505
 506     for (l = se->clients_cached; l; l = l->next)
 507         if (client_get_database(l->client) == db)
 508             break;
 509
 510     if (l)
 511         cl = l->client;
 512     else
 513     {
 514         cl = client_create(db->database->id);
 515         client_set_database(cl, db);
 516
 517         l = xmalloc(sizeof(*l));
 518         l->client = cl;
 519         l->next = se->clients_cached;
 520         se->clients_cached = l;
 521     }
 522     /* set session always. If may be 0 if client is not active */
 523     client_set_session(cl, se);
 524
 525     l = xmalloc(sizeof(*l));
 526     l->client = cl;
 527     l->next = se->clients_active;
 528     se->clients_active = l;
 529 }
 530
 531 static void session_reset_active_clients(struct session *se,
 532                                          struct client_list *new_list)
 533 {
 534     struct client_list *l;
 535
 536     session_enter(se, "session_reset_active_clients");
 537     l = se->clients_active;
 538     se->clients_active = new_list;
 539     session_leave(se, "session_reset_active_clients");
 540
 541     while (l)
 542     {
 543         struct client_list *l_next = l->next;
 544
 545         client_lock(l->client);
 546         client_set_session(l->client, 0); /* mark client inactive */
 547         client_unlock(l->client);
 548
 549         xfree(l);
 550         l = l_next;
 551     }
 552 }
 553
 554 static void session_remove_cached_clients(struct session *se)
 555 {
 556     struct client_list *l;
 557
 558     session_reset_active_clients(se, 0);
 559
 560     session_enter(se, "session_remove_cached_clients");
 561     l = se->clients_cached;
 562     se->clients_cached = 0;
 563     session_leave(se, "session_remove_cached_clients");
 564
 565     while (l)
 566     {
 567         struct client_list *l_next = l->next;
 568         client_lock(l->client);
 569         client_set_session(l->client, 0);
 570         client_set_database(l->client, 0);
 571         client_unlock(l->client);
 572         client_destroy(l->client);
 573         xfree(l);
 574         l = l_next;
 575     }
 576 }
 577
 578 // Associates a set of clients with a session;
 579 // Note: Session-databases represent databases with per-session
 580 // setting overrides
 581 static int select_targets(struct session *se, const char *filter)
 582 {
 583     return session_grep_databases(se, filter, select_targets_callback);
 584 }
 585
 586 int session_active_clients(struct session *s)
 587 {
 588     struct client_list *l;
 589     int res = 0;
 590
 591     for (l = s->clients_active; l; l = l->next)
 592         if (client_is_active(l->client))
 593             res++;
 594
 595     return res;
 596 }
 597
 598 int session_is_preferred_clients_ready(struct session *s)
 599 {
 600     struct client_list *l;
 601     int res = 0;
 602
 603     for (l = s->clients_active; l; l = l->next)
 604         if (client_is_active_preferred(l->client))
 605             res++;
 606     session_log(s, YLOG_DEBUG, "Has %d active preferred clients.", res);
 607     return res == 0;
 608 }
 609
 610 static void session_clear_set(struct session *se, struct reclist_sortparms *sp)
 611 {
 612     reclist_destroy(se->reclist);
 613     if (nmem_total(se->nmem))
 614         session_log(se, YLOG_DEBUG, "NMEN operation usage %zd",
 615                     nmem_total(se->nmem));
 616     nmem_reset(se->nmem);
 617     se->total_records = se->total_merged = 0;
 618     se->num_termlists = 0;
 619
 620     relevance_clear(se->relevance);
 621
 622     /* reset list of sorted results and clear to relevance search */
 623     se->sorted_results = nmem_malloc(se->nmem, sizeof(*se->sorted_results));
 624     se->sorted_results->name = nmem_strdup(se->nmem, sp->name);
 625     se->sorted_results->increasing = sp->increasing;
 626     se->sorted_results->type = sp->type;
 627     se->sorted_results->next = 0;
 628
 629     session_log(se, YLOG_DEBUG, "clear_set session_sort: field=%s increasing=%d type=%d configured",
 630             sp->name, sp->increasing, sp->type);
 631
 632     se->reclist = reclist_create(se->nmem);
 633 }
 634
 635 static void session_sort_unlocked(struct session *se,
 636                                   struct reclist_sortparms *sp,
 637                                   const char *mergekey,
 638                                   const char *rank)
 639 {
 640     struct client_list *l;
 641     const char *field = sp->name;
 642     int increasing = sp->increasing;
 643     int type  = sp->type;
 644     int clients_research = 0;
 645
 646     session_log(se, YLOG_DEBUG, "session_sort field=%s increasing=%d type=%d",
 647                 field, increasing, type);
 648
 649     if (rank && (!se->rank || strcmp(se->rank, rank)))
 650     {
 651         /* new rank must research/reingest anyway */
 652         assert(rank);
 653         xfree(se->rank);
 654         se->rank = *rank ? xstrdup(rank) : 0;
 655         clients_research = 1;
 656         session_log(se, YLOG_DEBUG, "session_sort: new rank = %s",
 657                     rank);
 658     }
 659     if (mergekey && (!se->mergekey || strcmp(se->mergekey, mergekey)))
 660     {
 661         /* new mergekey must research/reingest anyway */
 662         assert(mergekey);
 663         xfree(se->mergekey);
 664         se->mergekey = *mergekey ? xstrdup(mergekey) : 0;
 665         clients_research = 1;
 666         session_log(se, YLOG_DEBUG, "session_sort: new mergekey = %s",
 667                     mergekey);
 668     }
 669     if (clients_research == 0)
 670     {
 671         struct reclist_sortparms *sr;
 672         for (sr = se->sorted_results; sr; sr = sr->next)
 673             if (!reclist_sortparms_cmp(sr, sp))
 674                 break;
 675         if (sr)
 676         {
 677             session_log(se, YLOG_DEBUG, "session_sort: field=%s increasing=%d type=%d already fetched",
 678                         field, increasing, type);
 679             return;
 680         }
 681     }
 682     session_log(se, YLOG_DEBUG, "session_sort: field=%s increasing=%d type=%d must fetch",
 683                 field, increasing, type);
 684
 685     // We need to reset reclist on every sort that changes the records, not just for position
 686     // So if just one client requires new searching, we need to clear set.
 687     // Ask each of the client if sorting requires re-search due to native sort
 688     // If it does it will require us to
 689     for (l = se->clients_active; l; l = l->next)
 690     {
 691         struct client *cl = l->client;
 692         // Assume no re-search is required.
 693         client_parse_init(cl, 1);
 694         clients_research += client_parse_sort(cl, sp);
 695     }
 696     if (clients_research)
 697     {
 698         session_log(se, YLOG_DEBUG,
 699                     "session_sort: reset results due to %d clients researching",
 700                     clients_research);
 701         session_clear_set(se, sp);
 702     }
 703     else
 704     {
 705         // A new sorting based on same record set
 706         struct reclist_sortparms *sr = nmem_malloc(se->nmem, sizeof(*sr));
 707         sr->name = nmem_strdup(se->nmem, field);
 708         sr->increasing = increasing;
 709         sr->type = type;
 710         sr->next = se->sorted_results;
 711         se->sorted_results = sr;
 712         session_log(se, YLOG_DEBUG, "session_sort: no research/ingesting done");
 713         return ;
 714     }
 715     session_log(se, YLOG_DEBUG, "Re- search/ingesting for clients due to change in sort order");
 716
 717     for (l = se->clients_active; l; l = l->next)
 718     {
 719         struct client *cl = l->client;
 720         if (client_get_state(cl) == Client_Connecting ||
 721             client_get_state(cl) == Client_Idle ||
 722             client_get_state(cl) == Client_Working) {
 723             client_start_search(cl);
 724         }
 725         else
 726         {
 727             session_log(se, YLOG_DEBUG,
 728                         "session_sort: %s: No re-start/ingest in show. "
 729                         "Wrong client state: %d",
 730                         client_get_id(cl), client_get_state(cl));
 731         }
 732     }
 733 }
 734
 735 void session_sort(struct session *se, struct reclist_sortparms *sp,
 736                   const char *mergekey, const char *rank)
 737 {
 738     //session_enter(se, "session_sort");
 739     session_sort_unlocked(se, sp, mergekey, rank);
 740     //session_leave(se, "session_sort");
 741 }
 742
 743
 744 enum pazpar2_error_code session_search(struct session *se,
 745                                        const char *query,
 746                                        const char *startrecs,
 747                                        const char *maxrecs,
 748                                        const char *filter,
 749                                        const char *limit,
 750                                        const char **addinfo,
 751                                        struct reclist_sortparms *sp,
 752                                        const char *mergekey,
 753                                        const char *rank)
 754 {
 755     int live_channels = 0;
 756     int no_working = 0;
 757     int no_failed_query = 0;
 758     int no_failed_limit = 0;
 759     struct client_list *l, *l0;
 760
 761     session_log(se, YLOG_DEBUG, "Search");
 762
 763     *addinfo = 0;
 764
 765     if (se->settings_modified) {
 766         session_remove_cached_clients(se);
 767     }
 768     else
 769         session_reset_active_clients(se, 0);
 770
 771     session_enter(se, "session_search");
 772     se->settings_modified = 0;
 773
 774     if (mergekey)
 775     {
 776         xfree(se->mergekey);
 777         se->mergekey = *mergekey ? xstrdup(mergekey) : 0;
 778     }
 779     if (rank)
 780     {
 781         xfree(se->rank);
 782         se->rank = *rank ? xstrdup(rank) : 0;
 783     }
 784
 785     session_clear_set(se, sp);
 786     relevance_destroy(&se->relevance);
 787
 788     live_channels = select_targets(se, filter);
 789     if (!live_channels)
 790     {
 791         session_leave(se, "session_search");
 792         return PAZPAR2_NO_TARGETS;
 793     }
 794
 795     facet_limits_destroy(se->facet_limits);
 796     se->facet_limits = facet_limits_create(limit);
 797     if (!se->facet_limits)
 798     {
 799         *addinfo = "limit";
 800         session_leave(se, "session_search");
 801         return PAZPAR2_MALFORMED_PARAMETER_VALUE;
 802     }
 803
 804     l0 = se->clients_active;
 805     se->clients_active = 0;
 806     session_leave(se, "session_search");
 807
 808     for (l = l0; l; l = l->next)
 809     {
 810         int parse_ret;
 811         struct client *cl = l->client;
 812         client_parse_init(cl, 1);
 813         if (prepare_map(se, client_get_database(cl)) < 0)
 814             continue;
 815
 816         parse_ret = client_parse_query(cl, query, se->facet_limits);
 817         if (parse_ret == -1)
 818             no_failed_query++;
 819         else if (parse_ret == -2)
 820             no_failed_limit++;
 821         else if (parse_ret < 0)
 822             no_working++; /* other error, such as bad CCL map */
 823         else
 824         {
 825             client_parse_range(cl, startrecs, maxrecs);
 826             client_parse_sort(cl, sp);
 827             client_start_search(cl);
 828             no_working++;
 829         }
 830     }
 831     session_reset_active_clients(se, l0);
 832
 833     if (no_working == 0)
 834     {
 835         if (no_failed_query > 0)
 836         {
 837             *addinfo = "query";
 838             return PAZPAR2_MALFORMED_PARAMETER_VALUE;
 839         }
 840         else if (no_failed_limit > 0)
 841         {
 842             *addinfo = "limit";
 843             return PAZPAR2_MALFORMED_PARAMETER_VALUE;
 844         }
 845         else
 846             return PAZPAR2_NO_TARGETS;
 847     }
 848     session_log(se, YLOG_LOG, "session_start_search done");
 849     return PAZPAR2_NO_ERROR;
 850 }
 851
 852 // Creates a new session_database object for a database
 853 static void session_init_databases_fun(void *context, struct database *db)
 854 {
 855     struct session *se = (struct session *) context;
 856     struct session_database *new = nmem_malloc(se->session_nmem, sizeof(*new));
 857     int i;
 858
 859     new->database = db;
 860
 861     new->map = 0;
 862     assert(db->settings);
 863     new->settings = nmem_malloc(se->session_nmem,
 864                                 sizeof(struct settings *) * db->num_settings);
 865     new->num_settings = db->num_settings;
 866     for (i = 0; i < db->num_settings; i++)
 867     {
 868         struct setting *setting = db->settings[i];
 869         new->settings[i] = setting;
 870     }
 871     new->next = se->databases;
 872     se->databases = new;
 873 }
 874
 875 // Doesn't free memory associated with sdb -- nmem takes care of that
 876 static void session_database_destroy(struct session_database *sdb)
 877 {
 878     sdb->map = 0;
 879 }
 880
 881 // Initialize session_database list -- this represents this session's view
 882 // of the database list -- subject to modification by the settings ws command
 883 void session_init_databases(struct session *se)
 884 {
 885     se->databases = 0;
 886     predef_grep_databases(se, se->service, session_init_databases_fun);
 887 }
 888
 889 // Probably session_init_databases_fun should be refactored instead of
 890 // called here.
 891 static struct session_database *load_session_database(struct session *se,
 892                                                       char *id)
 893 {
 894     struct database *db = new_database_inherit_settings(id, se->session_nmem, se->service->settings);
 895     session_init_databases_fun((void*) se, db);
 896
 897     // New sdb is head of se->databases list
 898     return se->databases;
 899 }
 900
 901 // Find an existing session database. If not found, load it
 902 static struct session_database *find_session_database(struct session *se,
 903                                                       char *id)
 904 {
 905     struct session_database *sdb;
 906
 907     for (sdb = se->databases; sdb; sdb = sdb->next)
 908         if (!strcmp(sdb->database->id, id))
 909             return sdb;
 910     return load_session_database(se, id);
 911 }
 912
 913 // Apply a session override to a database
 914 void session_apply_setting(struct session *se, char *dbname, char *setting,
 915                            char *value)
 916 {
 917     struct session_database *sdb = find_session_database(se, dbname);
 918     struct conf_service *service = se->service;
 919     struct setting *new = nmem_malloc(se->session_nmem, sizeof(*new));
 920     int offset = settings_create_offset(service, setting);
 921
 922     expand_settings_array(&sdb->settings, &sdb->num_settings, offset,
 923                           se->session_nmem);
 924     new->precedence = 0;
 925     new->target = dbname;
 926     new->name = setting;
 927     new->value = value;
 928     new->next = sdb->settings[offset];
 929     sdb->settings[offset] = new;
 930
 931     se->settings_modified = 1;
 932
 933     // Force later recompute of settings-driven data structures
 934     // (happens when a search starts and client connections are prepared)
 935     switch (offset)
 936     {
 937     case PZ_XSLT:
 938         if (sdb->map)
 939         {
 940             sdb->map = 0;
 941         }
 942         break;
 943     }
 944 }
 945
 946 void session_destroy(struct session *se)
 947 {
 948     struct session_database *sdb;
 949     session_log(se, YLOG_DEBUG, "Destroying");
 950     session_use(-1);
 951     session_remove_cached_clients(se);
 952
 953     for (sdb = se->databases; sdb; sdb = sdb->next)
 954         session_database_destroy(sdb);
 955     normalize_cache_destroy(se->normalize_cache);
 956     relevance_destroy(&se->relevance);
 957     reclist_destroy(se->reclist);
 958     xfree(se->mergekey);
 959     xfree(se->rank);
 960     if (nmem_total(se->nmem))
 961         session_log(se, YLOG_DEBUG, "NMEN operation usage %zd", nmem_total(se->nmem));
 962     if (nmem_total(se->session_nmem))
 963         session_log(se, YLOG_DEBUG, "NMEN session usage %zd", nmem_total(se->session_nmem));
 964     facet_limits_destroy(se->facet_limits);
 965     nmem_destroy(se->nmem);
 966     service_destroy(se->service);
 967     yaz_mutex_destroy(&se->session_mutex);
 968 }
 969
 970 size_t session_get_memory_status(struct session *session) {
 971     size_t session_nmem;
 972     if (session == 0)
 973         return 0;
 974     session_enter(session, "session_get_memory_status");
 975     session_nmem = nmem_total(session->nmem);
 976     session_leave(session, "session_get_memory_status");
 977     return session_nmem;
 978 }
 979
 980
 981 struct session *new_session(NMEM nmem, struct conf_service *service,
 982                             unsigned session_id)
 983 {
 984     int i;
 985     struct session *session = nmem_malloc(nmem, sizeof(*session));
 986
 987     char tmp_str[50];
 988
 989     sprintf(tmp_str, "session#%u", session_id);
 990
 991     session->session_id = session_id;
 992     session_log(session, YLOG_DEBUG, "New");
 993     session->service = service;
 994     session->relevance = 0;
 995     session->total_records = 0;
 996     session->number_of_warnings_unknown_elements = 0;
 997     session->number_of_warnings_unknown_metadata = 0;
 998     session->num_termlists = 0;
 999     session->reclist = reclist_create(nmem);
1000     session->clients_active = 0;
1001     session->clients_cached = 0;
1002     session->settings_modified = 0;
1003     session->session_nmem = nmem;
1004     session->nmem = nmem_create();
1005     session->databases = 0;
1006     session->sorted_results = 0;
1007     session->facet_limits = 0;
1008     session->mergekey = 0;
1009     session->rank = 0;
1010
1011     for (i = 0; i <= SESSION_WATCH_MAX; i++)
1012     {
1013         session->watchlist[i].data = 0;
1014         session->watchlist[i].fun = 0;
1015     }
1016     session->normalize_cache = normalize_cache_create();
1017     session->session_mutex = 0;
1018     pazpar2_mutex_create(&session->session_mutex, tmp_str);
1019     session_use(1);
1020     return session;
1021 }
1022
1023 const char * client_get_suggestions_xml(struct client *cl, WRBUF wrbuf);
1024
1025 static struct hitsbytarget *hitsbytarget_nb(struct session *se,
1026                                             int *count, NMEM nmem)
1027 {
1028     struct hitsbytarget *res = 0;
1029     struct client_list *l;
1030     size_t sz = 0;
1031
1032     for (l = se->clients_active; l; l = l->next)
1033         sz++;
1034
1035     res = nmem_malloc(nmem, sizeof(*res) * sz);
1036     *count = 0;
1037     for (l = se->clients_active; l; l = l->next)
1038     {
1039         struct client *cl = l->client;
1040         WRBUF w = wrbuf_alloc();
1041         const char *name = session_setting_oneval(client_get_database(cl),
1042                                                   PZ_NAME);
1043
1044         res[*count].id = client_get_id(cl);
1045         res[*count].name = *name ? name : "Unknown";
1046         res[*count].hits = client_get_hits(cl);
1047         res[*count].approximation = client_get_approximation(cl);
1048         res[*count].records = client_get_num_records(cl);
1049         res[*count].filtered = client_get_num_records_filtered(cl);
1050         res[*count].diagnostic =
1051             client_get_diagnostic(cl, &res[*count].message,
1052                                   &res[*count].addinfo);
1053         res[*count].state = client_get_state_str(cl);
1054         res[*count].connected  = client_get_connection(cl) ? 1 : 0;
1055         session_settings_dump(se, client_get_database(cl), w);
1056         res[*count].settings_xml = nmem_strdup(nmem, wrbuf_cstr(w));
1057         wrbuf_rewind(w);
1058         wrbuf_puts(w, "");
1059         res[*count].suggestions_xml = nmem_strdup(nmem, client_get_suggestions_xml(cl, w));
1060         wrbuf_destroy(w);
1061         (*count)++;
1062     }
1063     return res;
1064 }
1065
1066 struct hitsbytarget *get_hitsbytarget(struct session *se, int *count, NMEM nmem)
1067 {
1068     struct hitsbytarget *p;
1069     session_enter(se, "get_hitsbytarget");
1070     p = hitsbytarget_nb(se, count, nmem);
1071     session_leave(se, "get_hitsbytarget");
1072     return p;
1073 }
1074
1075 // Compares two hitsbytarget nodes by hitcount
1076 static int cmp_ht(const void *p1, const void *p2)
1077 {
1078     const struct hitsbytarget *h1 = p1;
1079     const struct hitsbytarget *h2 = p2;
1080     return h2->hits - h1->hits;
1081 }
1082
1083 // Compares two hitsbytarget nodes by hitcount
1084 static int cmp_ht_approx(const void *p1, const void *p2)
1085 {
1086     const struct hitsbytarget *h1 = p1;
1087     const struct hitsbytarget *h2 = p2;
1088     return h2->approximation - h1->approximation;
1089 }
1090
1091 static int targets_termlist_nb(WRBUF wrbuf, struct session *se, int num,
1092                                NMEM nmem, int version)
1093 {
1094     struct hitsbytarget *ht;
1095     int count, i;
1096
1097     ht = hitsbytarget_nb(se, &count, nmem);
1098     if (version >= 2)
1099         qsort(ht, count, sizeof(struct hitsbytarget), cmp_ht_approx);
1100     else
1101         qsort(ht, count, sizeof(struct hitsbytarget), cmp_ht);
1102     for (i = 0; i < count && i < num && ht[i].hits > 0; i++)
1103     {
1104
1105         // do only print terms which have display names
1106
1107         wrbuf_puts(wrbuf, "<term>\n");
1108
1109         wrbuf_puts(wrbuf, "<id>");
1110         wrbuf_xmlputs(wrbuf, ht[i].id);
1111         wrbuf_puts(wrbuf, "</id>\n");
1112
1113         wrbuf_puts(wrbuf, "<name>");
1114         if (!ht[i].name || !ht[i].name[0])
1115             wrbuf_xmlputs(wrbuf, "NO TARGET NAME");
1116         else
1117             wrbuf_xmlputs(wrbuf, ht[i].name);
1118         wrbuf_puts(wrbuf, "</name>\n");
1119
1120         wrbuf_printf(wrbuf, "<frequency>" ODR_INT_PRINTF "</frequency>\n",
1121                      ht[i].hits);
1122
1123         if (version >= 2) {
1124             // Should not print if we know it isn't a approximation.
1125             wrbuf_printf(wrbuf, "<approximation>" ODR_INT_PRINTF "</approximation>\n", ht[i].approximation);
1126             wrbuf_printf(wrbuf, "<records>%d</records>\n", ht[i].records - ht[i].filtered);
1127             wrbuf_printf(wrbuf, "<filtered>%d</filtered>\n", ht[i].filtered);
1128         }
1129
1130         wrbuf_puts(wrbuf, "<state>");
1131         wrbuf_xmlputs(wrbuf, ht[i].state);
1132         wrbuf_puts(wrbuf, "</state>\n");
1133
1134         wrbuf_printf(wrbuf, "<diagnostic>%d</diagnostic>\n",
1135                      ht[i].diagnostic);
1136         wrbuf_puts(wrbuf, "</term>\n");
1137     }
1138     return count;
1139 }
1140
1141 void perform_termlist(struct http_channel *c, struct session *se,
1142                       const char *name, int num, int version)
1143 {
1144     int i, j;
1145     NMEM nmem_tmp = nmem_create();
1146     char **names;
1147     int num_names = 0;
1148
1149     if (!name)
1150         name = "*";
1151
1152     nmem_strsplit(nmem_tmp, ",", name, &names, &num_names);
1153
1154     session_enter(se, "perform_termlist");
1155
1156     for (j = 0; j < num_names; j++)
1157     {
1158         const char *tname;
1159         int must_generate_empty = 1; /* bug 5350 */
1160
1161         for (i = 0; i < se->num_termlists; i++)
1162         {
1163             tname = se->termlists[i].name;
1164             if (!strcmp(names[j], tname) || !strcmp(names[j], "*"))
1165             {
1166                 struct termlist_score **p = 0;
1167                 int len;
1168
1169                 wrbuf_puts(c->wrbuf, "<list name=\"");
1170                 wrbuf_xmlputs(c->wrbuf, tname);
1171                 wrbuf_puts(c->wrbuf, "\">\n");
1172                 must_generate_empty = 0;
1173
1174                 p = termlist_highscore(se->termlists[i].termlist, &len,
1175                                        nmem_tmp);
1176                 if (p)
1177                 {
1178                     int i;
1179                     for (i = 0; i < len && i < num; i++)
1180                     {
1181                         // prevent sending empty term elements
1182                         if (!p[i]->display_term || !p[i]->display_term[0])
1183                             continue;
1184
1185                         wrbuf_puts(c->wrbuf, "<term>");
1186                         wrbuf_puts(c->wrbuf, "<name>");
1187                         wrbuf_xmlputs(c->wrbuf, p[i]->display_term);
1188                         wrbuf_puts(c->wrbuf, "</name>");
1189
1190                         wrbuf_printf(c->wrbuf,
1191                                      "<frequency>%d</frequency>",
1192                                      p[i]->frequency);
1193                         wrbuf_puts(c->wrbuf, "</term>\n");
1194                     }
1195                 }
1196                 wrbuf_puts(c->wrbuf, "</list>\n");
1197             }
1198         }
1199         tname = "xtargets";
1200         if (!strcmp(names[j], tname) || !strcmp(names[j], "*"))
1201         {
1202             wrbuf_puts(c->wrbuf, "<list name=\"");
1203             wrbuf_xmlputs(c->wrbuf, tname);
1204             wrbuf_puts(c->wrbuf, "\">\n");
1205
1206             targets_termlist_nb(c->wrbuf, se, num, c->nmem, version);
1207             wrbuf_puts(c->wrbuf, "</list>\n");
1208             must_generate_empty = 0;
1209         }
1210         if (must_generate_empty)
1211         {
1212             wrbuf_puts(c->wrbuf, "<list name=\"");
1213             wrbuf_xmlputs(c->wrbuf, names[j]);
1214             wrbuf_puts(c->wrbuf, "\"/>\n");
1215         }
1216     }
1217     session_leave(se, "perform_termlist");
1218     nmem_destroy(nmem_tmp);
1219 }
1220
1221 #ifdef MISSING_HEADERS
1222 void report_nmem_stats(void)
1223 {
1224     size_t in_use, is_free;
1225
1226     nmem_get_memory_in_use(&in_use);
1227     nmem_get_memory_free(&is_free);
1228
1229     yaz_log(YLOG_LOG, "nmem stat: use=%ld free=%ld",
1230             (long) in_use, (long) is_free);
1231 }
1232 #endif
1233
1234 struct record_cluster *show_single_start(struct session *se, const char *id,
1235                                          struct record_cluster **prev_r,
1236                                          struct record_cluster **next_r)
1237 {
1238     struct record_cluster *r = 0;
1239
1240     session_enter(se, "show_single_start");
1241     *prev_r = 0;
1242     *next_r = 0;
1243     reclist_limit(se->reclist, se, 1);
1244
1245     reclist_enter(se->reclist);
1246     while ((r = reclist_read_record(se->reclist)))
1247     {
1248         if (!strcmp(r->recid, id))
1249         {
1250             *next_r = reclist_read_record(se->reclist);
1251             break;
1252         }
1253         *prev_r = r;
1254     }
1255     reclist_leave(se->reclist);
1256     if (!r)
1257         session_leave(se, "show_single_start");
1258     return r;
1259 }
1260
1261 void show_single_stop(struct session *se, struct record_cluster *rec)
1262 {
1263     session_leave(se, "show_single_stop");
1264 }
1265
1266
1267 int session_fetch_more(struct session *se)
1268 {
1269     struct client_list *l;
1270     int ret = 0;
1271
1272     for (l = se->clients_active; l; l = l->next)
1273     {
1274         struct client *cl = l->client;
1275         if (client_get_state(cl) == Client_Idle)
1276         {
1277             if (client_fetch_more(cl))
1278             {
1279                 session_log(se, YLOG_LOG, "%s: more to fetch",
1280                             client_get_id(cl));
1281                 ret = 1;
1282             }
1283             else
1284             {
1285                 session_log(se, YLOG_LOG, "%s: no more to fetch",
1286                             client_get_id(cl));
1287             }
1288         }
1289         else
1290         {
1291             session_log(se, YLOG_LOG, "%s: no fetch due to state=%s",
1292                         client_get_id(cl), client_get_state_str(cl));
1293         }
1294
1295     }
1296     return ret;
1297 }
1298
1299 struct record_cluster **show_range_start(struct session *se,
1300                                          struct reclist_sortparms *sp,
1301                                          int start, int *num, int *total,
1302                                          Odr_int *sumhits, Odr_int *approx_hits,
1303                                          void (*show_records_ready)(void *data),
1304                                          struct http_channel *chan)
1305 {
1306     struct record_cluster **recs = 0;
1307     struct reclist_sortparms *spp;
1308     struct client_list *l;
1309     int i;
1310 #if USE_TIMING
1311     yaz_timing_t t = yaz_timing_create();
1312 #endif
1313     session_enter(se, "show_range_start");
1314     *sumhits = 0;
1315     *approx_hits = 0;
1316     *total = 0;
1317     reclist_limit(se->reclist, se, 0);
1318     if (se->relevance)
1319     {
1320         for (spp = sp; spp; spp = spp->next)
1321             if (spp->type == Metadata_sortkey_relevance)
1322             {
1323                 relevance_prepare_read(se->relevance, se->reclist);
1324                 break;
1325             }
1326         for (l = se->clients_active; l; l = l->next) {
1327             *sumhits += client_get_hits(l->client);
1328             *approx_hits += client_get_approximation(l->client);
1329         }
1330     }
1331     reclist_sort(se->reclist, sp);
1332
1333     reclist_enter(se->reclist);
1334     *total = reclist_get_num_records(se->reclist);
1335
1336     for (l = se->clients_active; l; l = l->next)
1337         client_update_show_stat(l->client, 0);
1338
1339     for (i = 0; i < start; i++)
1340     {
1341         struct record_cluster *r = reclist_read_record(se->reclist);
1342         if (!r)
1343         {
1344             *num = 0;
1345             break;
1346         }
1347         else
1348         {
1349             struct record *rec = r->records;
1350             for (;rec; rec = rec->next)
1351                 client_update_show_stat(rec->client, 1);
1352         }
1353     }
1354     recs = nmem_malloc(se->nmem, (*num > 0 ? *num : 1) * sizeof(*recs));
1355     for (i = 0; i < *num; i++)
1356     {
1357         struct record_cluster *r = reclist_read_record(se->reclist);
1358         if (!r)
1359         {
1360             *num = i;
1361             break;
1362         }
1363         else
1364         {
1365             struct record *rec = r->records;
1366             for (;rec; rec = rec->next)
1367                 client_update_show_stat(rec->client, 1);
1368             recs[i] = r;
1369         }
1370     }
1371     reclist_leave(se->reclist);
1372 #if USE_TIMING
1373     yaz_timing_stop(t);
1374     session_log(se, YLOG_LOG, "show %6.5f %3.2f %3.2f",
1375             yaz_timing_get_real(t), yaz_timing_get_user(t),
1376             yaz_timing_get_sys(t));
1377     yaz_timing_destroy(&t);
1378 #endif
1379
1380     if (!session_fetch_more(se))
1381         session_log(se, YLOG_LOG, "can not fetch more");
1382     else
1383     {
1384         show_range_stop(se, recs);
1385         session_log(se, YLOG_LOG, "fetching more in progress");
1386         if (session_set_watch(se, SESSION_WATCH_SHOW,
1387                               show_records_ready, chan, chan))
1388         {
1389             session_log(se, YLOG_WARN, "Ignoring show block");
1390             session_enter(se, "show_range_start");
1391         }
1392         else
1393         {
1394             session_log(se, YLOG_LOG, "session watch OK");
1395             return 0;
1396         }
1397     }
1398     return recs;
1399 }
1400
1401 void show_range_stop(struct session *se, struct record_cluster **recs)
1402 {
1403     session_leave(se, "show_range_stop");
1404 }
1405
1406 void statistics(struct session *se, struct statistics *stat)
1407 {
1408     struct client_list *l;
1409     int count = 0;
1410
1411     memset(stat, 0, sizeof(*stat));
1412     stat->num_hits = 0;
1413     for (l = se->clients_active; l; l = l->next)
1414     {
1415         struct client *cl = l->client;
1416         if (!client_get_connection(cl))
1417             stat->num_no_connection++;
1418         stat->num_hits += client_get_hits(cl);
1419         switch (client_get_state(cl))
1420         {
1421         case Client_Connecting: stat->num_connecting++; break;
1422         case Client_Working: stat->num_working++; break;
1423         case Client_Idle: stat->num_idle++; break;
1424         case Client_Failed: stat->num_failed++; break;
1425         case Client_Error: stat->num_error++; break;
1426         default: break;
1427         }
1428         count++;
1429     }
1430     stat->num_records = se->total_records;
1431
1432     stat->num_clients = count;
1433 }
1434
1435 static struct record_metadata *record_metadata_init(
1436     NMEM nmem, const char *value, enum conf_metadata_type type,
1437     struct _xmlAttr *attr)
1438 {
1439     struct record_metadata *rec_md = record_metadata_create(nmem);
1440     struct record_metadata_attr **attrp = &rec_md->attributes;
1441
1442     for (; attr; attr = attr->next)
1443     {
1444         if (attr->children && attr->children->content)
1445         {
1446             if (strcmp((const char *) attr->name, "type")
1447                 && strcmp((const char *) attr->name, "empty"))
1448             {  /* skip the "type" + "empty" attribute..
1449                   The "Type" is already part of the element in output
1450                   (md-%s) and so repeating it here is redundant */
1451                 *attrp = nmem_malloc(nmem, sizeof(**attrp));
1452                 (*attrp)->name =
1453                     nmem_strdup(nmem, (const char *) attr->name);
1454                 (*attrp)->value =
1455                     nmem_strdup(nmem, (const char *) attr->children->content);
1456                 attrp = &(*attrp)->next;
1457             }
1458         }
1459     }
1460     *attrp = 0;
1461
1462     if (type == Metadata_type_generic)
1463     {
1464         char *p = nmem_strdup(nmem, value);
1465
1466         p = normalize7bit_generic(p, " ,/.:([");
1467
1468         rec_md->data.text.disp = p;
1469         rec_md->data.text.sort = 0;
1470         rec_md->data.text.snippet = 0;
1471     }
1472     else if (type == Metadata_type_year || type == Metadata_type_date)
1473     {
1474         int first, last;
1475         int longdate = 0;
1476
1477         if (type == Metadata_type_date)
1478             longdate = 1;
1479         if (extract7bit_dates((char *) value, &first, &last, longdate) < 0)
1480             return 0;
1481
1482         rec_md->data.number.min = first;
1483         rec_md->data.number.max = last;
1484     }
1485     else
1486         return 0;
1487     return rec_md;
1488 }
1489
1490 static void mergekey_norm_wr(pp2_charset_fact_t charsets,
1491                              WRBUF norm_wr, const char *value)
1492 {
1493     const char *norm_str;
1494     pp2_charset_token_t prt =
1495         pp2_charset_token_create(charsets, "mergekey");
1496
1497     pp2_charset_token_first(prt, value, 0);
1498     while ((norm_str = pp2_charset_token_next(prt)))
1499     {
1500         if (*norm_str)
1501         {
1502             if (wrbuf_len(norm_wr))
1503                 wrbuf_puts(norm_wr, " ");
1504             wrbuf_puts(norm_wr, norm_str);
1505         }
1506     }
1507     pp2_charset_token_destroy(prt);
1508 }
1509
1510 static int get_mergekey_from_doc(xmlDoc *doc, xmlNode *root, const char *name,
1511                                  struct conf_service *service, WRBUF norm_wr)
1512 {
1513     xmlNode *n;
1514     int no_found = 0;
1515     for (n = root->children; n; n = n->next)
1516     {
1517         if (n->type != XML_ELEMENT_NODE)
1518             continue;
1519         if (!strcmp((const char *) n->name, "metadata"))
1520         {
1521             xmlChar *type = xmlGetProp(n, (xmlChar *) "type");
1522             if (type == NULL) {
1523                 yaz_log(YLOG_FATAL, "Missing type attribute on metadata element. Skipping!");
1524             }
1525             else if (!strcmp(name, (const char *) type))
1526             {
1527                 xmlChar *value = xmlNodeListGetString(doc, n->children, 1);
1528                 if (value && *value)
1529                 {
1530                     if (wrbuf_len(norm_wr) > 0)
1531                         wrbuf_puts(norm_wr, " ");
1532                     wrbuf_puts(norm_wr, name);
1533                     mergekey_norm_wr(service->charsets, norm_wr,
1534                                      (const char *) value);
1535                     no_found++;
1536                 }
1537                 if (value)
1538                     xmlFree(value);
1539             }
1540             xmlFree(type);
1541         }
1542     }
1543     return no_found;
1544 }
1545
1546 static const char *get_mergekey(xmlDoc *doc, struct client *cl, int record_no,
1547                                 struct conf_service *service, NMEM nmem,
1548                                 const char *session_mergekey)
1549 {
1550     char *mergekey_norm = 0;
1551     xmlNode *root = xmlDocGetRootElement(doc);
1552     WRBUF norm_wr = wrbuf_alloc();
1553     xmlChar *mergekey;
1554
1555     if (session_mergekey)
1556     {
1557         int i, num = 0;
1558         char **values = 0;
1559         nmem_strsplit_escape2(nmem, ",", session_mergekey, &values,
1560                               &num, 1, '\\', 1);
1561
1562         for (i = 0; i < num; i++)
1563             get_mergekey_from_doc(doc, root, values[i], service, norm_wr);
1564     }
1565     else if ((mergekey = xmlGetProp(root, (xmlChar *) "mergekey")))
1566     {
1567         mergekey_norm_wr(service->charsets, norm_wr, (const char *) mergekey);
1568         xmlFree(mergekey);
1569     }
1570     else
1571     {
1572         /* no mergekey defined in XSL. Look for mergekey metadata instead */
1573         int field_id;
1574         for (field_id = 0; field_id < service->num_metadata; field_id++)
1575         {
1576             struct conf_metadata *ser_md = &service->metadata[field_id];
1577             if (ser_md->mergekey != Metadata_mergekey_no)
1578             {
1579                 int r = get_mergekey_from_doc(doc, root, ser_md->name,
1580                                               service, norm_wr);
1581                 if (r == 0 && ser_md->mergekey == Metadata_mergekey_required)
1582                 {
1583                     /* no mergekey on this one and it is required..
1584                        Generate unique key instead */
1585                     wrbuf_rewind(norm_wr);
1586                     break;
1587                 }
1588             }
1589         }
1590     }
1591
1592     /* generate unique key if none is not generated already or is empty */
1593     if (wrbuf_len(norm_wr) == 0)
1594     {
1595         wrbuf_printf(norm_wr, "position: %s-%d",
1596                      client_get_id(cl), record_no);
1597     }
1598     else
1599     {
1600         const char *lead = "content: ";
1601         wrbuf_insert(norm_wr, 0, lead, strlen(lead));
1602     }
1603     if (wrbuf_len(norm_wr) > 0)
1604         mergekey_norm = nmem_strdup(nmem, wrbuf_cstr(norm_wr));
1605     wrbuf_destroy(norm_wr);
1606     return mergekey_norm;
1607 }
1608
1609 /** \brief see if metadata for pz:recordfilter exists
1610     \param root xml root element of normalized record
1611     \param sdb session database for client
1612     \retval 0 if there is no metadata for pz:recordfilter
1613     \retval 1 if there is metadata for pz:recordfilter
1614
1615     If there is no pz:recordfilter defined, this function returns 1
1616     as well.
1617 */
1618
1619 static int check_record_filter(xmlNode *root, struct session_database *sdb)
1620 {
1621     int match = 0;
1622     xmlNode *n;
1623     const char *s;
1624     s = session_setting_oneval(sdb, PZ_RECORDFILTER);
1625
1626     if (!s || !*s)
1627         return 1;
1628
1629     for (n = root->children; n; n = n->next)
1630     {
1631         if (n->type != XML_ELEMENT_NODE)
1632             continue;
1633         if (!strcmp((const char *) n->name, "metadata"))
1634         {
1635             xmlChar *type = xmlGetProp(n, (xmlChar *) "type");
1636             if (type)
1637             {
1638                 size_t len;
1639                 int substring;
1640                 const char *eq;
1641
1642                 if ((eq = strchr(s, '=')))
1643                     substring = 0;
1644                 else if ((eq = strchr(s, '~')))
1645                     substring = 1;
1646                 if (eq)
1647                     len = eq - s;
1648                 else
1649                     len = strlen(s);
1650                 if (len == strlen((const char *)type) &&
1651                     !memcmp((const char *) type, s, len))
1652                 {
1653                     xmlChar *value = xmlNodeGetContent(n);
1654                     if (value && *value)
1655                     {
1656                         if (!eq ||
1657                             (substring && strstr((const char *) value, eq+1)) ||
1658                             (!substring && !strcmp((const char *) value, eq + 1)))
1659                             match = 1;
1660                     }
1661                     xmlFree(value);
1662                 }
1663                 xmlFree(type);
1664             }
1665         }
1666     }
1667     return match;
1668 }
1669
1670
1671 static int ingest_to_cluster(struct client *cl,
1672                              xmlDoc *xdoc,
1673                              xmlNode *root,
1674                              int record_no,
1675                              const char *mergekey_norm);
1676
1677 /** \brief ingest XML record
1678     \param cl client holds the result set for record
1679     \param rec record buffer (0 terminated)
1680     \param record_no record position (1, 2, ..)
1681     \param nmem working NMEM
1682     \retval 0 OK
1683     \retval -1 failure
1684     \retval -2 Filtered
1685 */
1686 int ingest_record(struct client *cl, const char *rec,
1687                   int record_no, NMEM nmem)
1688 {
1689     struct session *se = client_get_session(cl);
1690     int ret = 0;
1691     struct session_database *sdb = client_get_database(cl);
1692     struct conf_service *service = se->service;
1693     xmlDoc *xdoc = normalize_record(se, sdb, service, rec, nmem);
1694     xmlNode *root;
1695     const char *mergekey_norm;
1696
1697     if (!xdoc)
1698         return -1;
1699
1700     root = xmlDocGetRootElement(xdoc);
1701
1702     if (!check_record_filter(root, sdb))
1703     {
1704         session_log(se, YLOG_LOG, "Filtered out record no %d from %s", record_no, sdb->database->id);
1705         xmlFreeDoc(xdoc);
1706         return -2;
1707     }
1708
1709     mergekey_norm = get_mergekey(xdoc, cl, record_no, service, nmem,
1710         se->mergekey);
1711     if (!mergekey_norm)
1712     {
1713         session_log(se, YLOG_WARN, "Got no mergekey");
1714         xmlFreeDoc(xdoc);
1715         return -1;
1716     }
1717     session_enter(se, "ingest_record");
1718     if (client_get_session(cl) == se && se->relevance)
1719         ret = ingest_to_cluster(cl, xdoc, root, record_no, mergekey_norm);
1720     session_leave(se, "ingest_record");
1721
1722     xmlFreeDoc(xdoc);
1723     return ret;
1724 }
1725
1726 //    struct conf_metadata *ser_md = &service->metadata[md_field_id];
1727 //    struct record_metadata *rec_md = record->metadata[md_field_id];
1728 static int match_metadata_local(struct conf_service *service,
1729                                 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     WRBUF val_wr = 0;
1736     WRBUF text_wr = wrbuf_alloc();
1737     for (i = 0; i < num_v; )
1738     {
1739         if (rec_md)
1740         {
1741             if (ser_md->type == Metadata_type_year
1742                 || ser_md->type == Metadata_type_date)
1743             {
1744                 int y = atoi(values[i]);
1745                 if (y >= rec_md->data.number.min
1746                     && y <= rec_md->data.number.max)
1747                     break;
1748             }
1749             else
1750             {
1751                 if (!val_wr)
1752                 {
1753                     val_wr = wrbuf_alloc();
1754                     mergekey_norm_wr(service->charsets, val_wr, values[i]);
1755                 }
1756                 wrbuf_rewind(text_wr);
1757                 mergekey_norm_wr(service->charsets, text_wr,
1758                                  rec_md->data.text.disp);
1759                 if (!strcmp(wrbuf_cstr(val_wr), wrbuf_cstr(text_wr)))
1760                     break;
1761             }
1762             rec_md = rec_md->next;
1763         }
1764         else
1765         {
1766             rec_md = rec_md0;
1767             wrbuf_destroy(val_wr);
1768             val_wr = 0;
1769             i++;
1770         }
1771     }
1772     wrbuf_destroy(val_wr);
1773     wrbuf_destroy(text_wr);
1774     return i < num_v ? 1 : 0;
1775 }
1776
1777 int session_check_cluster_limit(struct session *se, struct record_cluster *rec)
1778 {
1779     int i;
1780     struct conf_service *service = se->service;
1781     int ret = 1;
1782     const char *name;
1783     const char *value;
1784     NMEM nmem_tmp = nmem_create();
1785
1786     for (i = 0; (name = facet_limits_get(se->facet_limits, i, &value)); i++)
1787     {
1788         int j;
1789         for (j = 0; j < service->num_metadata; j++)
1790         {
1791             struct conf_metadata *md = service->metadata + j;
1792             if (!strcmp(md->name, name) && md->limitcluster)
1793             {
1794                 char **values = 0;
1795                 int num = 0;
1796                 int md_field_id =
1797                     conf_service_metadata_field_id(service,
1798                                                    md->limitcluster);
1799
1800                 if (md_field_id < 0)
1801                 {
1802                     ret = 0;
1803                     break;
1804                 }
1805
1806                 nmem_strsplit_escape2(nmem_tmp, "|", value, &values,
1807                                       &num, 1, '\\', 1);
1808
1809                 if (!match_metadata_local(service,
1810                                           &service->metadata[md_field_id],
1811                                           rec->metadata[md_field_id],
1812                                           values, num))
1813                 {
1814                     ret = 0;
1815                     break;
1816                 }
1817             }
1818         }
1819     }
1820     nmem_destroy(nmem_tmp);
1821     return ret;
1822 }
1823
1824 // Skip record on non-zero
1825 static int check_limit_local(struct client *cl,
1826                              struct record *record,
1827                              int record_no)
1828 {
1829     int skip_record = 0;
1830     struct session *se = client_get_session(cl);
1831     struct conf_service *service = se->service;
1832     NMEM nmem_tmp = nmem_create();
1833     struct session_database *sdb = client_get_database(cl);
1834     int l = 0;
1835     while (!skip_record)
1836     {
1837         int md_field_id;
1838         char **values = 0;
1839         int num_v = 0;
1840         const char *name =
1841             client_get_facet_limit_local(cl, sdb, &l, nmem_tmp,
1842                                          &num_v, &values);
1843         if (!name)
1844             break;
1845
1846         if (!strcmp(name, "*"))
1847         {
1848             for (md_field_id = 0; md_field_id < service->num_metadata;
1849                  md_field_id++)
1850             {
1851                 if (match_metadata_local(
1852                         service,
1853                         &service->metadata[md_field_id],
1854                         record->metadata[md_field_id],
1855                         values, num_v))
1856                     break;
1857             }
1858             if (md_field_id == service->num_metadata)
1859                 skip_record = 1;
1860         }
1861         else
1862         {
1863             md_field_id = conf_service_metadata_field_id(service, name);
1864             if (md_field_id < 0)
1865             {
1866                 skip_record = 1;
1867                 break;
1868             }
1869             if (!match_metadata_local(
1870                     service,
1871                     &service->metadata[md_field_id],
1872                     record->metadata[md_field_id],
1873                     values, num_v))
1874             {
1875                 skip_record = 1;
1876             }
1877         }
1878     }
1879     nmem_destroy(nmem_tmp);
1880     return skip_record;
1881 }
1882
1883 static int ingest_to_cluster(struct client *cl,
1884                              xmlDoc *xdoc,
1885                              xmlNode *root,
1886                              int record_no,
1887                              const char *mergekey_norm)
1888 {
1889     xmlNode *n;
1890     xmlChar *type = 0;
1891     xmlChar *value = 0;
1892     struct session *se = client_get_session(cl);
1893     struct conf_service *service = se->service;
1894     int term_factor = 1;
1895     struct record_cluster *cluster;
1896     struct record_metadata **metadata0;
1897     struct session_database *sdb = client_get_database(cl);
1898     NMEM ingest_nmem = 0;
1899     char **rank_values = 0;
1900     int rank_num = 0;
1901     struct record *record = record_create(se->nmem,
1902                                           service->num_metadata,
1903                                           service->num_sortkeys, cl,
1904                                           record_no);
1905
1906     for (n = root->children; n; n = n->next)
1907     {
1908         if (type)
1909             xmlFree(type);
1910         if (value)
1911             xmlFree(value);
1912         type = value = 0;
1913
1914         if (n->type != XML_ELEMENT_NODE)
1915             continue;
1916         if (!strcmp((const char *) n->name, "metadata"))
1917         {
1918             struct conf_metadata *ser_md = 0;
1919             struct record_metadata **wheretoput = 0;
1920             struct record_metadata *rec_md = 0;
1921             int md_field_id = -1;
1922
1923             type = xmlGetProp(n, (xmlChar *) "type");
1924             value = xmlNodeListGetString(xdoc, n->children, 1);
1925             if (!type)
1926                 continue;
1927             if (!value || !*value)
1928             {
1929                 xmlChar *empty = xmlGetProp(n, (xmlChar *) "empty");
1930                 if (!empty)
1931                     continue;
1932                 if (value)
1933                     xmlFree(value);
1934                 value = empty;
1935             }
1936             md_field_id
1937                 = conf_service_metadata_field_id(service, (const char *) type);
1938             if (md_field_id < 0)
1939             {
1940                 if (se->number_of_warnings_unknown_metadata == 0)
1941                 {
1942                     session_log(se, YLOG_WARN,
1943                             "Ignoring unknown metadata element: %s", type);
1944                 }
1945                 se->number_of_warnings_unknown_metadata++;
1946                 continue;
1947             }
1948
1949             ser_md = &service->metadata[md_field_id];
1950
1951             // non-merged metadata
1952             rec_md = record_metadata_init(se->nmem, (const char *) value,
1953                                           ser_md->type, n->properties);
1954             if (!rec_md)
1955             {
1956                 session_log(se, YLOG_WARN, "bad metadata data '%s' "
1957                             "for element '%s'", value, type);
1958                 continue;
1959             }
1960
1961             if (ser_md->type == Metadata_type_generic)
1962             {
1963                 WRBUF w = wrbuf_alloc();
1964                 if (relevance_snippet(se->relevance,
1965                                       (char*) value, ser_md->name, w))
1966                     rec_md->data.text.snippet = nmem_strdup(se->nmem,
1967                                                             wrbuf_cstr(w));
1968                 wrbuf_destroy(w);
1969             }
1970
1971
1972             wheretoput = &record->metadata[md_field_id];
1973             while (*wheretoput)
1974                 wheretoput = &(*wheretoput)->next;
1975             *wheretoput = rec_md;
1976         }
1977     }
1978
1979     if (check_limit_local(cl, record, record_no))
1980     {
1981         session_log(se, YLOG_LOG, "Facet filtered out record no %d from %s",
1982                     record_no, sdb->database->id);
1983         if (type)
1984             xmlFree(type);
1985         if (value)
1986             xmlFree(value);
1987         return -2;
1988     }
1989     cluster = reclist_insert(se->reclist, service, record,
1990                              mergekey_norm, &se->total_merged);
1991     if (!cluster)
1992         return 0; // complete match with existing record
1993
1994     {
1995         const char *use_term_factor_str =
1996             session_setting_oneval(sdb, PZ_TERMLIST_TERM_FACTOR);
1997         if (use_term_factor_str && use_term_factor_str[0] == '1')
1998         {
1999             int maxrecs = client_get_maxrecs(cl);
2000             int hits = (int) client_get_hits(cl);
2001             term_factor = MAX(hits, maxrecs) /  MAX(1, maxrecs);
2002             assert(term_factor >= 1);
2003             session_log(se, YLOG_DEBUG, "Using term factor: %d (%d / %d)",
2004                         term_factor, MAX(hits, maxrecs), MAX(1, maxrecs));
2005         }
2006     }
2007
2008     if (global_parameters.dump_records)
2009         session_log(se, YLOG_LOG, "Cluster id %s from %s (#%d)", cluster->recid,
2010                     sdb->database->id, record_no);
2011
2012
2013     relevance_newrec(se->relevance, cluster);
2014
2015     // original metadata, to check if first existence of a field
2016     metadata0 = xmalloc(sizeof(*metadata0) * service->num_metadata);
2017     memcpy(metadata0, cluster->metadata,
2018            sizeof(*metadata0) * service->num_metadata);
2019
2020     ingest_nmem = nmem_create();
2021     if (se->rank)
2022     {
2023         yaz_log(YLOG_LOG, "local in sort : %s", se->rank);
2024         nmem_strsplit_escape2(ingest_nmem, ",", se->rank, &rank_values,
2025                               &rank_num, 1, '\\', 1);
2026     }
2027
2028     // now parsing XML record and adding data to cluster or record metadata
2029     for (n = root->children; n; n = n->next)
2030     {
2031         pp2_charset_token_t prt;
2032         if (type)
2033             xmlFree(type);
2034         if (value)
2035             xmlFree(value);
2036         type = value = 0;
2037
2038         if (n->type != XML_ELEMENT_NODE)
2039             continue;
2040         if (!strcmp((const char *) n->name, "metadata"))
2041         {
2042             struct conf_metadata *ser_md = 0;
2043             struct conf_sortkey *ser_sk = 0;
2044             struct record_metadata **wheretoput = 0;
2045             struct record_metadata *rec_md = 0;
2046             int md_field_id = -1;
2047             int sk_field_id = -1;
2048             const char *rank = 0;
2049             xmlChar *xml_rank = 0;
2050
2051             type = xmlGetProp(n, (xmlChar *) "type");
2052             value = xmlNodeListGetString(xdoc, n->children, 1);
2053
2054             if (!type || !value || !*value)
2055                 continue;
2056
2057             md_field_id
2058                 = conf_service_metadata_field_id(service, (const char *) type);
2059             if (md_field_id < 0)
2060                 continue;
2061
2062             ser_md = &service->metadata[md_field_id];
2063
2064             if (ser_md->sortkey_offset >= 0)
2065             {
2066                 sk_field_id = ser_md->sortkey_offset;
2067                 ser_sk = &service->sortkeys[sk_field_id];
2068             }
2069
2070             // merged metadata
2071             rec_md = record_metadata_init(se->nmem, (const char *) value,
2072                                           ser_md->type, 0);
2073
2074             // see if the field was not in cluster already (from beginning)
2075
2076             if (!rec_md)
2077                 continue;
2078
2079             if (rank_num)
2080             {
2081                 int i;
2082                 for (i = 0; i < rank_num; i++)
2083                 {
2084                     const char *val = rank_values[i];
2085                     const char *cp = strchr(val, '=');
2086                     if (!cp)
2087                         continue;
2088                     if ((cp - val) == strlen((const char *) type)
2089                         && !memcmp(val, type, cp - val))
2090                     {
2091                         rank = cp + 1;
2092                         break;
2093                     }
2094                 }
2095             }
2096             else
2097             {
2098                 xml_rank = xmlGetProp(n, (xmlChar *) "rank");
2099                 rank = xml_rank ? (const char *) xml_rank : ser_md->rank;
2100             }
2101
2102             wheretoput = &cluster->metadata[md_field_id];
2103
2104             if (ser_md->merge == Metadata_merge_first)
2105             {
2106                 if (!metadata0[md_field_id])
2107                 {
2108                     while (*wheretoput)
2109                         wheretoput = &(*wheretoput)->next;
2110                     *wheretoput = rec_md;
2111                 }
2112             }
2113             else if (ser_md->merge == Metadata_merge_unique)
2114             {
2115                 while (*wheretoput)
2116                 {
2117                     if (!strcmp((const char *) (*wheretoput)->data.text.disp,
2118                                 rec_md->data.text.disp))
2119                         break;
2120                     wheretoput = &(*wheretoput)->next;
2121                 }
2122                 if (!*wheretoput)
2123                     *wheretoput = rec_md;
2124             }
2125             else if (ser_md->merge == Metadata_merge_longest)
2126             {
2127                 if (!*wheretoput
2128                     || strlen(rec_md->data.text.disp)
2129                     > strlen((*wheretoput)->data.text.disp))
2130                 {
2131                     *wheretoput = rec_md;
2132                     if (ser_sk)
2133                     {
2134                         const char *sort_str = 0;
2135                         int skip_article =
2136                             ser_sk->type == Metadata_sortkey_skiparticle;
2137
2138                         if (!cluster->sortkeys[sk_field_id])
2139                             cluster->sortkeys[sk_field_id] =
2140                                 nmem_malloc(se->nmem,
2141                                             sizeof(union data_types));
2142
2143                         prt =
2144                             pp2_charset_token_create(service->charsets, "sort");
2145
2146                         pp2_charset_token_first(prt, rec_md->data.text.disp,
2147                                                 skip_article);
2148
2149                         pp2_charset_token_next(prt);
2150
2151                         sort_str = pp2_get_sort(prt);
2152
2153                         cluster->sortkeys[sk_field_id]->text.disp =
2154                             rec_md->data.text.disp;
2155                         if (!sort_str)
2156                         {
2157                             sort_str = rec_md->data.text.disp;
2158                             session_log(se, YLOG_WARN,
2159                                     "Could not make sortkey. Bug #1858");
2160                         }
2161                         cluster->sortkeys[sk_field_id]->text.sort =
2162                             nmem_strdup(se->nmem, sort_str);
2163                         pp2_charset_token_destroy(prt);
2164                     }
2165                 }
2166             }
2167             else if (ser_md->merge == Metadata_merge_all)
2168             {
2169                 while (*wheretoput)
2170                     wheretoput = &(*wheretoput)->next;
2171                 *wheretoput = rec_md;
2172             }
2173             else if (ser_md->merge == Metadata_merge_range)
2174             {
2175                 if (!*wheretoput)
2176                 {
2177                     *wheretoput = rec_md;
2178                     if (ser_sk)
2179                         cluster->sortkeys[sk_field_id]
2180                             = &rec_md->data;
2181                 }
2182                 else
2183                 {
2184                     int this_min = rec_md->data.number.min;
2185                     int this_max = rec_md->data.number.max;
2186                     if (this_min < (*wheretoput)->data.number.min)
2187                         (*wheretoput)->data.number.min = this_min;
2188                     if (this_max > (*wheretoput)->data.number.max)
2189                         (*wheretoput)->data.number.max = this_max;
2190                 }
2191             }
2192
2193             // ranking of _all_ fields enabled ...
2194             if (rank)
2195             {
2196                 relevance_countwords(se->relevance, cluster,
2197                                      (char *) value, rank, ser_md->name);
2198             }
2199             // construct facets ... unless the client already has reported them
2200             if (ser_md->termlist && !client_has_facet(cl, (char *) type))
2201             {
2202                 if (ser_md->type == Metadata_type_year)
2203                 {
2204                     char year[64];
2205                     sprintf(year, "%d", rec_md->data.number.max);
2206
2207                     add_facet(se, (char *) type, year, term_factor);
2208                     if (rec_md->data.number.max != rec_md->data.number.min)
2209                     {
2210                         sprintf(year, "%d", rec_md->data.number.min);
2211                         add_facet(se, (char *) type, year, term_factor);
2212                     }
2213                 }
2214                 else
2215                     add_facet(se, (char *) type, (char *) value, term_factor);
2216             }
2217
2218             // cleaning up
2219             if (xml_rank)
2220                 xmlFree(xml_rank);
2221             xmlFree(type);
2222             xmlFree(value);
2223             type = value = 0;
2224         }
2225         else
2226         {
2227             if (se->number_of_warnings_unknown_elements == 0)
2228                 session_log(se, YLOG_WARN,
2229                         "Unexpected element in internal record: %s", n->name);
2230             se->number_of_warnings_unknown_elements++;
2231         }
2232     }
2233     if (type)
2234         xmlFree(type);
2235     if (value)
2236         xmlFree(value);
2237
2238     nmem_destroy(ingest_nmem);
2239     xfree(metadata0);
2240     relevance_donerecord(se->relevance, cluster);
2241     se->total_records++;
2242
2243     return 0;
2244 }
2245
2246 void session_log(struct session *s, int level, const char *fmt, ...)
2247 {
2248     char buf[1024];
2249     va_list ap;
2250     va_start(ap, fmt);
2251
2252     yaz_vsnprintf(buf, sizeof(buf)-30, fmt, ap);
2253     yaz_log(level, "Session %u: %s", s ? s->session_id : 0, buf);
2254
2255     va_end(ap);
2256 }
2257
2258 /*
2259  * Local variables:
2260  * c-basic-offset: 4
2261  * c-file-style: "Stroustrup"
2262  * indent-tabs-mode: nil
2263  * End:
2264  * vim: shiftwidth=4 tabstop=8 expandtab
2265  */
2266