src/session.c

   1 /* This file is part of Pazpar2.
   2    Copyright (C) 2006-2011 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 #include <yaz/gettimeofday.h>
  60
  61 #define USE_TIMING 0
  62 #if USE_TIMING
  63 #include <yaz/timing.h>
  64 #endif
  65
  66 #include "ppmutex.h"
  67 #include "parameters.h"
  68 #include "session.h"
  69 #include "eventl.h"
  70 #include "http.h"
  71 #include "termlists.h"
  72 #include "reclists.h"
  73 #include "relevance.h"
  74 #include "database.h"
  75 #include "client.h"
  76 #include "settings.h"
  77 #include "normalize7bit.h"
  78
  79 #define TERMLIST_HIGH_SCORE 25
  80
  81 #define MAX_CHUNK 15
  82
  83 #define MAX(a,b) ((a)>(b)?(a):(b))
  84
  85 // Note: Some things in this structure will eventually move to configuration
  86 struct parameters global_parameters =
  87 {
  88     0,   // dump_records
  89     0,   // debug_mode
  90     0,   // predictable sessions
  91 };
  92
  93 struct client_list {
  94     struct client *client;
  95     struct client_list *next;
  96 };
  97
  98 /* session counting (1) , disable client counting (0) */
  99 static YAZ_MUTEX g_session_mutex = 0;
 100 static int no_sessions = 0;
 101 static int no_session_total = 0;
 102
 103 static int session_use(int delta)
 104 {
 105     int sessions;
 106     if (!g_session_mutex)
 107         yaz_mutex_create(&g_session_mutex);
 108     yaz_mutex_enter(g_session_mutex);
 109     no_sessions += delta;
 110     if (delta > 0)
 111         no_session_total += delta;
 112     sessions = no_sessions;
 113     yaz_mutex_leave(g_session_mutex);
 114     yaz_log(YLOG_DEBUG, "%s sessions=%d", delta == 0 ? "" : (delta > 0 ? "INC" : "DEC"), no_sessions);
 115     return sessions;
 116 }
 117
 118 int sessions_count(void)
 119 {
 120     return session_use(0);
 121 }
 122
 123 int session_count_total(void)
 124 {
 125     int total = 0;
 126     if (!g_session_mutex)
 127         return 0;
 128     yaz_mutex_enter(g_session_mutex);
 129     total = no_session_total;
 130     yaz_mutex_leave(g_session_mutex);
 131     return total;
 132 }
 133
 134 static void log_xml_doc(xmlDoc *doc)
 135 {
 136     FILE *lf = yaz_log_file();
 137     xmlChar *result = 0;
 138     int len = 0;
 139 #if LIBXML_VERSION >= 20600
 140     xmlDocDumpFormatMemory(doc, &result, &len, 1);
 141 #else
 142     xmlDocDumpMemory(doc, &result, &len);
 143 #endif
 144     if (lf && len)
 145     {
 146         (void) fwrite(result, 1, len, lf);
 147         fprintf(lf, "\n");
 148     }
 149     xmlFree(result);
 150 }
 151
 152 static void session_enter(struct session *s)
 153 {
 154     yaz_mutex_enter(s->session_mutex);
 155 }
 156
 157 static void session_leave(struct session *s)
 158 {
 159     yaz_mutex_leave(s->session_mutex);
 160 }
 161
 162 void add_facet(struct session *s, const char *type, const char *value, int count)
 163 {
 164     struct conf_service *service = s->service;
 165     pp2_charset_token_t prt;
 166     const char *facet_component;
 167     WRBUF facet_wrbuf = wrbuf_alloc();
 168     WRBUF display_wrbuf = wrbuf_alloc();
 169     int i;
 170     const char *icu_chain_id = 0;
 171
 172     for (i = 0; i < service->num_metadata; i++)
 173         if (!strcmp((service->metadata + i)->name, type))
 174             icu_chain_id = (service->metadata + i)->facetrule;
 175     if (!icu_chain_id)
 176         icu_chain_id = "facet";
 177     prt = pp2_charset_token_create(service->charsets, icu_chain_id);
 178     if (!prt)
 179     {
 180         yaz_log(YLOG_FATAL, "Unknown ICU chain '%s' for facet of type '%s'",
 181                 icu_chain_id, type);
 182         wrbuf_destroy(facet_wrbuf);
 183         wrbuf_destroy(display_wrbuf);
 184         return;
 185     }
 186     pp2_charset_token_first(prt, value, 0);
 187     while ((facet_component = pp2_charset_token_next(prt)))
 188     {
 189         const char *display_component;
 190         if (*facet_component)
 191         {
 192             if (wrbuf_len(facet_wrbuf))
 193                 wrbuf_puts(facet_wrbuf, " ");
 194             wrbuf_puts(facet_wrbuf, facet_component);
 195         }
 196         display_component = pp2_get_display(prt);
 197         if (display_component)
 198         {
 199             if (wrbuf_len(display_wrbuf))
 200                 wrbuf_puts(display_wrbuf, " ");
 201             wrbuf_puts(display_wrbuf, display_component);
 202         }
 203     }
 204     pp2_charset_token_destroy(prt);
 205
 206     if (wrbuf_len(facet_wrbuf))
 207     {
 208         int i;
 209         for (i = 0; i < s->num_termlists; i++)
 210             if (!strcmp(s->termlists[i].name, type))
 211                 break;
 212         if (i == s->num_termlists)
 213         {
 214             if (i == SESSION_MAX_TERMLISTS)
 215             {
 216                 session_log(s, YLOG_FATAL, "Too many termlists");
 217                 wrbuf_destroy(facet_wrbuf);
 218                 wrbuf_destroy(display_wrbuf);
 219                 return;
 220             }
 221
 222             s->termlists[i].name = nmem_strdup(s->nmem, type);
 223             s->termlists[i].termlist
 224                 = termlist_create(s->nmem, TERMLIST_HIGH_SCORE);
 225             s->num_termlists = i + 1;
 226         }
 227
 228 #if 0
 229         session_log(s, YLOG_DEBUG, "Facets for %s: %s norm:%s (%d)", type, value, wrbuf_cstr(facet_wrbuf), count);
 230 #endif
 231         termlist_insert(s->termlists[i].termlist, wrbuf_cstr(display_wrbuf),
 232                         wrbuf_cstr(facet_wrbuf), count);
 233     }
 234     wrbuf_destroy(facet_wrbuf);
 235     wrbuf_destroy(display_wrbuf);
 236 }
 237
 238 static xmlDoc *record_to_xml(struct session *se,
 239                              struct session_database *sdb, const char *rec)
 240 {
 241     struct database *db = sdb->database;
 242     xmlDoc *rdoc = 0;
 243
 244     rdoc = xmlParseMemory(rec, strlen(rec));
 245
 246     if (!rdoc)
 247     {
 248         session_log(se, YLOG_FATAL, "Non-wellformed XML received from %s",
 249                     db->url);
 250         return 0;
 251     }
 252
 253     if (global_parameters.dump_records)
 254     {
 255         session_log(se, YLOG_LOG, "Un-normalized record from %s", db->url);
 256         log_xml_doc(rdoc);
 257     }
 258
 259     return rdoc;
 260 }
 261
 262 #define MAX_XSLT_ARGS 16
 263
 264 // Add static values from session database settings if applicable
 265 static void insert_settings_parameters(struct session_database *sdb,
 266                                        struct conf_service *service,
 267                                        char **parms,
 268                                        NMEM nmem)
 269 {
 270     int i;
 271     int nparms = 0;
 272     int offset = 0;
 273
 274     for (i = 0; i < service->num_metadata; i++)
 275     {
 276         struct conf_metadata *md = &service->metadata[i];
 277         int setting;
 278
 279         if (md->setting == Metadata_setting_parameter &&
 280             (setting = settings_lookup_offset(service, md->name)) >= 0)
 281         {
 282             const char *val = session_setting_oneval(sdb, setting);
 283             if (val && nparms < MAX_XSLT_ARGS)
 284             {
 285                 char *buf;
 286                 int len = strlen(val);
 287                 buf = nmem_malloc(nmem, len + 3);
 288                 buf[0] = '\'';
 289                 strcpy(buf + 1, val);
 290                 buf[len+1] = '\'';
 291                 buf[len+2] = '\0';
 292                 parms[offset++] = md->name;
 293                 parms[offset++] = buf;
 294                 nparms++;
 295             }
 296         }
 297     }
 298     parms[offset] = 0;
 299 }
 300
 301 // Add static values from session database settings if applicable
 302 static void insert_settings_values(struct session_database *sdb, xmlDoc *doc,
 303     struct conf_service *service)
 304 {
 305     int i;
 306
 307     for (i = 0; i < service->num_metadata; i++)
 308     {
 309         struct conf_metadata *md = &service->metadata[i];
 310         int offset;
 311
 312         if (md->setting == Metadata_setting_postproc &&
 313             (offset = settings_lookup_offset(service, md->name)) >= 0)
 314         {
 315             const char *val = session_setting_oneval(sdb, offset);
 316             if (val)
 317             {
 318                 xmlNode *r = xmlDocGetRootElement(doc);
 319                 xmlNode *n = xmlNewTextChild(r, 0, (xmlChar *) "metadata",
 320                                              (xmlChar *) val);
 321                 xmlSetProp(n, (xmlChar *) "type", (xmlChar *) md->name);
 322             }
 323         }
 324     }
 325 }
 326
 327 static xmlDoc *normalize_record(struct session *se,
 328                                 struct session_database *sdb,
 329                                 struct conf_service *service,
 330                                 const char *rec, NMEM nmem)
 331 {
 332     xmlDoc *rdoc = record_to_xml(se, sdb, rec);
 333
 334     if (rdoc)
 335     {
 336         char *parms[MAX_XSLT_ARGS*2+1];
 337
 338         insert_settings_parameters(sdb, service, parms, nmem);
 339
 340         if (normalize_record_transform(sdb->map, &rdoc, (const char **)parms))
 341         {
 342             session_log(se, YLOG_WARN, "Normalize failed from %s",
 343                         sdb->database->url);
 344         }
 345         else
 346         {
 347             insert_settings_values(sdb, rdoc, service);
 348
 349             if (global_parameters.dump_records)
 350             {
 351                 session_log(se, YLOG_LOG, "Normalized record from %s",
 352                             sdb->database->url);
 353                 log_xml_doc(rdoc);
 354             }
 355         }
 356     }
 357     return rdoc;
 358 }
 359
 360 void session_settings_dump(struct session *se,
 361                            struct session_database *db,
 362                            WRBUF w)
 363 {
 364     if (db->settings)
 365     {
 366         int i, num = db->num_settings;
 367         for (i = 0; i < num; i++)
 368         {
 369             struct setting *s = db->settings[i];
 370             for (;s ; s = s->next)
 371             {
 372                 wrbuf_puts(w, "<set name=\"");
 373                 wrbuf_xmlputs(w, s->name);
 374                 wrbuf_puts(w, "\" value=\"");
 375                 wrbuf_xmlputs(w, s->value);
 376                 wrbuf_puts(w, "\"/>");
 377             }
 378             if (db->settings[i])
 379                 wrbuf_puts(w, "\n");
 380         }
 381     }
 382 }
 383
 384 // Retrieve first defined value for 'name' for given database.
 385 // Will be extended to take into account user associated with session
 386 const char *session_setting_oneval(struct session_database *db, int offset)
 387 {
 388     if (offset >= db->num_settings || !db->settings[offset])
 389         return "";
 390     return db->settings[offset]->value;
 391 }
 392
 393 // Prepare XSLT stylesheets for record normalization
 394 // Structures are allocated on the session_wide nmem to avoid having
 395 // to recompute this for every search. This would lead
 396 // to leaking if a single session was to repeatedly change the PZ_XSLT
 397 // setting. However, this is not a realistic use scenario.
 398 static int prepare_map(struct session *se, struct session_database *sdb)
 399 {
 400     const char *s;
 401
 402     if (!sdb->settings)
 403     {
 404         session_log(se, YLOG_WARN, "No settings on %s", sdb->database->url);
 405         return -1;
 406     }
 407     if ((s = session_setting_oneval(sdb, PZ_XSLT)))
 408     {
 409         char auto_stylesheet[256];
 410
 411         if (!strcmp(s, "auto"))
 412         {
 413             const char *request_syntax = session_setting_oneval(
 414                 sdb, PZ_REQUESTSYNTAX);
 415             if (request_syntax)
 416             {
 417                 char *cp;
 418                 yaz_snprintf(auto_stylesheet, sizeof(auto_stylesheet),
 419                              "%s.xsl", request_syntax);
 420                 for (cp = auto_stylesheet; *cp; cp++)
 421                 {
 422                     /* deliberately only consider ASCII */
 423                     if (*cp > 32 && *cp < 127)
 424                         *cp = tolower(*cp);
 425                 }
 426                 s = auto_stylesheet;
 427             }
 428             else
 429             {
 430                 session_log(se, YLOG_WARN,
 431                             "No pz:requestsyntax for auto stylesheet");
 432             }
 433         }
 434         sdb->map = normalize_cache_get(se->normalize_cache,
 435                                        se->service->server->config, s);
 436         if (!sdb->map)
 437             return -1;
 438     }
 439     return 0;
 440 }
 441
 442 // This analyzes settings and recomputes any supporting data structures
 443 // if necessary.
 444 static int prepare_session_database(struct session *se,
 445                                     struct session_database *sdb)
 446 {
 447     if (!sdb->settings)
 448     {
 449         session_log(se, YLOG_WARN,
 450                 "No settings associated with %s", sdb->database->url);
 451         return -1;
 452     }
 453     if (sdb->settings[PZ_XSLT] && !sdb->map)
 454     {
 455         if (prepare_map(se, sdb) < 0)
 456             return -1;
 457     }
 458     return 0;
 459 }
 460
 461 // called if watch should be removed because http_channel is to be destroyed
 462 static void session_watch_cancel(void *data, struct http_channel *c,
 463                                  void *data2)
 464 {
 465     struct session_watchentry *ent = data;
 466
 467     ent->fun = 0;
 468     ent->data = 0;
 469     ent->obs = 0;
 470 }
 471
 472 // set watch. Returns 0=OK, -1 if watch is already set
 473 int session_set_watch(struct session *s, int what,
 474                       session_watchfun fun, void *data,
 475                       struct http_channel *chan)
 476 {
 477     int ret;
 478     session_enter(s);
 479     if (s->watchlist[what].fun)
 480         ret = -1;
 481     else
 482     {
 483
 484         s->watchlist[what].fun = fun;
 485         s->watchlist[what].data = data;
 486         s->watchlist[what].obs = http_add_observer(chan, &s->watchlist[what],
 487                                                    session_watch_cancel);
 488         ret = 0;
 489     }
 490     session_leave(s);
 491     return ret;
 492 }
 493
 494 void session_alert_watch(struct session *s, int what)
 495 {
 496     assert(s);
 497     session_enter(s);
 498     if (s->watchlist[what].fun)
 499     {
 500         /* our watch is no longer associated with http_channel */
 501         void *data;
 502         session_watchfun fun;
 503
 504         http_remove_observer(s->watchlist[what].obs);
 505         fun  = s->watchlist[what].fun;
 506         data = s->watchlist[what].data;
 507
 508         /* reset watch before fun is invoked - in case fun wants to set
 509            it again */
 510         s->watchlist[what].fun = 0;
 511         s->watchlist[what].data = 0;
 512         s->watchlist[what].obs = 0;
 513
 514         session_leave(s);
 515         session_log(s, YLOG_DEBUG,
 516                     "Alert Watch: %d calling function: %p", what, fun);
 517         fun(data);
 518     }
 519     else
 520         session_leave(s);
 521 }
 522
 523 //callback for grep_databases
 524 static void select_targets_callback(struct session *se,
 525                                     struct session_database *db)
 526 {
 527     struct client *cl = client_create(db->database->url);
 528     struct client_list *l;
 529     const char *url = session_setting_oneval(db, PZ_URL);
 530
 531     if (!url || !*url)
 532         url = db->database->url;
 533
 534     client_set_database(cl, db);
 535
 536     client_set_session(cl, se);
 537
 538     l = xmalloc(sizeof(*l));
 539     l->client = cl;
 540     l->next = se->clients;
 541     se->clients = l;
 542 }
 543
 544 static void session_remove_clients(struct session *se)
 545 {
 546     struct client_list *l;
 547
 548     session_enter(se);
 549     l = se->clients;
 550     se->clients = 0;
 551     session_leave(se);
 552
 553     while (l)
 554     {
 555         struct client_list *l_next = l->next;
 556         client_lock(l->client);
 557         client_set_session(l->client, 0);
 558         client_set_database(l->client, 0);
 559         client_unlock(l->client);
 560         client_destroy(l->client);
 561         xfree(l);
 562         l = l_next;
 563     }
 564 }
 565
 566 // Associates a set of clients with a session;
 567 // Note: Session-databases represent databases with per-session
 568 // setting overrides
 569 static int select_targets(struct session *se, const char *filter)
 570 {
 571     return session_grep_databases(se, filter, select_targets_callback);
 572 }
 573
 574 int session_active_clients(struct session *s)
 575 {
 576     struct client_list *l;
 577     int res = 0;
 578
 579     for (l = s->clients; l; l = l->next)
 580         if (client_is_active(l->client))
 581             res++;
 582
 583     return res;
 584 }
 585
 586 int session_is_preferred_clients_ready(struct session *s)
 587 {
 588     struct client_list *l;
 589     int res = 0;
 590
 591     for (l = s->clients; l; l = l->next)
 592         if (client_is_active_preferred(l->client))
 593             res++;
 594     session_log(s, YLOG_DEBUG, "Has %d active preferred clients.", res);
 595     return res == 0;
 596 }
 597
 598 enum pazpar2_error_code search(struct session *se,
 599                                const char *query,
 600                                const char *startrecs, const char *maxrecs,
 601                                const char *filter,
 602                                const char *limit,
 603                                const char **addinfo)
 604 {
 605     int live_channels = 0;
 606     int no_working = 0;
 607     int no_failed = 0;
 608     struct client_list *l;
 609     struct timeval tval;
 610     facet_limits_t facet_limits;
 611
 612     session_log(se, YLOG_DEBUG, "Search");
 613
 614     *addinfo = 0;
 615
 616     session_remove_clients(se);
 617
 618     session_enter(se);
 619     reclist_destroy(se->reclist);
 620     se->reclist = 0;
 621     relevance_destroy(&se->relevance);
 622     nmem_reset(se->nmem);
 623     se->total_records = se->total_hits = se->total_merged = 0;
 624     se->num_termlists = 0;
 625     live_channels = select_targets(se, filter);
 626     if (!live_channels)
 627     {
 628         session_leave(se);
 629         return PAZPAR2_NO_TARGETS;
 630     }
 631     se->reclist = reclist_create(se->nmem);
 632
 633     yaz_gettimeofday(&tval);
 634
 635     tval.tv_sec += 5;
 636
 637     facet_limits = facet_limits_create(limit);
 638     if (!facet_limits)
 639     {
 640         *addinfo = "limit";
 641         session_leave(se);
 642         return PAZPAR2_MALFORMED_PARAMETER_VALUE;
 643     }
 644     for (l = se->clients; l; l = l->next)
 645     {
 646         struct client *cl = l->client;
 647
 648         if (maxrecs)
 649             client_set_maxrecs(cl, atoi(maxrecs));
 650         if (startrecs)
 651             client_set_startrecs(cl, atoi(startrecs));
 652         if (prepare_session_database(se, client_get_database(cl)) < 0)
 653             ;
 654         else if (client_parse_query(cl, query, facet_limits) < 0)
 655             no_failed++;
 656         else
 657         {
 658             no_working++;
 659             if (client_prep_connection(cl, se->service->z3950_operation_timeout,
 660                                        se->service->z3950_session_timeout,
 661                                        se->service->server->iochan_man,
 662                                        &tval))
 663                 client_start_search(cl);
 664         }
 665     }
 666     facet_limits_destroy(facet_limits);
 667     session_leave(se);
 668     if (no_working == 0)
 669     {
 670         if (no_failed > 0)
 671         {
 672             *addinfo = "query";
 673             return PAZPAR2_MALFORMED_PARAMETER_VALUE;
 674         }
 675         else
 676             return PAZPAR2_NO_TARGETS;
 677     }
 678     return PAZPAR2_NO_ERROR;
 679 }
 680
 681 // Creates a new session_database object for a database
 682 static void session_init_databases_fun(void *context, struct database *db)
 683 {
 684     struct session *se = (struct session *) context;
 685     struct session_database *new = nmem_malloc(se->session_nmem, sizeof(*new));
 686     int i;
 687
 688     new->database = db;
 689
 690     new->map = 0;
 691     assert(db->settings);
 692     new->settings = nmem_malloc(se->session_nmem,
 693                                 sizeof(struct settings *) * db->num_settings);
 694     new->num_settings = db->num_settings;
 695     for (i = 0; i < db->num_settings; i++)
 696     {
 697         struct setting *setting = db->settings[i];
 698         new->settings[i] = setting;
 699     }
 700     new->next = se->databases;
 701     se->databases = new;
 702 }
 703
 704 // Doesn't free memory associated with sdb -- nmem takes care of that
 705 static void session_database_destroy(struct session_database *sdb)
 706 {
 707     sdb->map = 0;
 708 }
 709
 710 // Initialize session_database list -- this represents this session's view
 711 // of the database list -- subject to modification by the settings ws command
 712 void session_init_databases(struct session *se)
 713 {
 714     se->databases = 0;
 715     predef_grep_databases(se, se->service, session_init_databases_fun);
 716 }
 717
 718 // Probably session_init_databases_fun should be refactored instead of
 719 // called here.
 720 static struct session_database *load_session_database(struct session *se,
 721                                                       char *id)
 722 {
 723     struct database *db = new_database(id, se->session_nmem);
 724
 725     session_init_databases_fun((void*) se, db);
 726
 727     // New sdb is head of se->databases list
 728     return se->databases;
 729 }
 730
 731 // Find an existing session database. If not found, load it
 732 static struct session_database *find_session_database(struct session *se,
 733                                                       char *id)
 734 {
 735     struct session_database *sdb;
 736
 737     for (sdb = se->databases; sdb; sdb = sdb->next)
 738         if (!strcmp(sdb->database->url, id))
 739             return sdb;
 740     return load_session_database(se, id);
 741 }
 742
 743 // Apply a session override to a database
 744 void session_apply_setting(struct session *se, char *dbname, char *setting,
 745                            char *value)
 746 {
 747     struct session_database *sdb = find_session_database(se, dbname);
 748     struct conf_service *service = se->service;
 749     struct setting *new = nmem_malloc(se->session_nmem, sizeof(*new));
 750     int offset = settings_create_offset(service, setting);
 751
 752     expand_settings_array(&sdb->settings, &sdb->num_settings, offset,
 753                           se->session_nmem);
 754     new->precedence = 0;
 755     new->target = dbname;
 756     new->name = setting;
 757     new->value = value;
 758     new->next = sdb->settings[offset];
 759     sdb->settings[offset] = new;
 760
 761     // Force later recompute of settings-driven data structures
 762     // (happens when a search starts and client connections are prepared)
 763     switch (offset)
 764     {
 765     case PZ_XSLT:
 766         if (sdb->map)
 767         {
 768             sdb->map = 0;
 769         }
 770         break;
 771     }
 772 }
 773
 774 void session_destroy(struct session *se) {
 775     struct session_database *sdb;
 776     session_log(se, YLOG_DEBUG, "Destroying");
 777     session_use(-1);
 778     session_remove_clients(se);
 779
 780     for (sdb = se->databases; sdb; sdb = sdb->next)
 781         session_database_destroy(sdb);
 782     normalize_cache_destroy(se->normalize_cache);
 783     relevance_destroy(&se->relevance);
 784     reclist_destroy(se->reclist);
 785     nmem_destroy(se->nmem);
 786     service_destroy(se->service);
 787     yaz_mutex_destroy(&se->session_mutex);
 788 }
 789
 790 /* Depreciated: use session_destroy */
 791 void destroy_session(struct session *se)
 792 {
 793     session_destroy(se);
 794 }
 795
 796 size_t session_get_memory_status(struct session *session) {
 797     size_t session_nmem;
 798     if (session == 0)
 799         return 0;
 800     session_enter(session);
 801     session_nmem = nmem_total(session->nmem);
 802     session_leave(session);
 803     return session_nmem;
 804 }
 805
 806
 807 struct session *new_session(NMEM nmem, struct conf_service *service,
 808                             unsigned session_id)
 809 {
 810     int i;
 811     struct session *session = nmem_malloc(nmem, sizeof(*session));
 812
 813     char tmp_str[50];
 814
 815     sprintf(tmp_str, "session#%u", session_id);
 816
 817     session->session_id = session_id;
 818     session_log(session, YLOG_DEBUG, "New");
 819     session->service = service;
 820     session->relevance = 0;
 821     session->total_hits = 0;
 822     session->total_records = 0;
 823     session->number_of_warnings_unknown_elements = 0;
 824     session->number_of_warnings_unknown_metadata = 0;
 825     session->num_termlists = 0;
 826     session->reclist = 0;
 827     session->clients = 0;
 828     session->session_nmem = nmem;
 829     session->nmem = nmem_create();
 830     session->databases = 0;
 831     for (i = 0; i <= SESSION_WATCH_MAX; i++)
 832     {
 833         session->watchlist[i].data = 0;
 834         session->watchlist[i].fun = 0;
 835     }
 836     session->normalize_cache = normalize_cache_create();
 837     session->session_mutex = 0;
 838     pazpar2_mutex_create(&session->session_mutex, tmp_str);
 839     session_use(1);
 840     return session;
 841 }
 842
 843 static struct hitsbytarget *hitsbytarget_nb(struct session *se,
 844                                             int *count, NMEM nmem)
 845 {
 846     struct hitsbytarget *res = 0;
 847     struct client_list *l;
 848     size_t sz = 0;
 849
 850     for (l = se->clients; l; l = l->next)
 851         sz++;
 852
 853     res = nmem_malloc(nmem, sizeof(*res) * sz);
 854     *count = 0;
 855     for (l = se->clients; l; l = l->next)
 856     {
 857         struct client *cl = l->client;
 858         WRBUF w = wrbuf_alloc();
 859         const char *name = session_setting_oneval(client_get_database(cl),
 860                                                   PZ_NAME);
 861
 862         res[*count].id = client_get_database(cl)->database->url;
 863         res[*count].name = *name ? name : "Unknown";
 864         res[*count].hits = client_get_hits(cl);
 865         res[*count].records = client_get_num_records(cl);
 866         res[*count].diagnostic = client_get_diagnostic(cl);
 867         res[*count].state = client_get_state_str(cl);
 868         res[*count].connected  = client_get_connection(cl) ? 1 : 0;
 869         session_settings_dump(se, client_get_database(cl), w);
 870         res[*count].settings_xml = nmem_strdup(nmem, wrbuf_cstr(w));
 871         wrbuf_destroy(w);
 872         (*count)++;
 873     }
 874     return res;
 875 }
 876
 877 struct hitsbytarget *get_hitsbytarget(struct session *se, int *count, NMEM nmem)
 878 {
 879     struct hitsbytarget *p;
 880     session_enter(se);
 881     p = hitsbytarget_nb(se, count, nmem);
 882     session_leave(se);
 883     return p;
 884 }
 885
 886 struct termlist_score **get_termlist_score(struct session *se,
 887                                            const char *name, int *num)
 888 {
 889     int i;
 890     struct termlist_score **tl = 0;
 891
 892     session_enter(se);
 893     for (i = 0; i < se->num_termlists; i++)
 894         if (!strcmp((const char *) se->termlists[i].name, name))
 895         {
 896             tl = termlist_highscore(se->termlists[i].termlist, num);
 897             break;
 898         }
 899     session_leave(se);
 900     return tl;
 901 }
 902
 903 // Compares two hitsbytarget nodes by hitcount
 904 static int cmp_ht(const void *p1, const void *p2)
 905 {
 906     const struct hitsbytarget *h1 = p1;
 907     const struct hitsbytarget *h2 = p2;
 908     return h2->hits - h1->hits;
 909 }
 910
 911 static int targets_termlist_nb(WRBUF wrbuf, struct session *se, int num,
 912                                NMEM nmem)
 913 {
 914     struct hitsbytarget *ht;
 915     int count, i;
 916
 917     ht = hitsbytarget_nb(se, &count, nmem);
 918     qsort(ht, count, sizeof(struct hitsbytarget), cmp_ht);
 919     for (i = 0; i < count && i < num && ht[i].hits > 0; i++)
 920     {
 921
 922         // do only print terms which have display names
 923
 924         wrbuf_puts(wrbuf, "<term>\n");
 925
 926         wrbuf_puts(wrbuf, "<id>");
 927         wrbuf_xmlputs(wrbuf, ht[i].id);
 928         wrbuf_puts(wrbuf, "</id>\n");
 929
 930         wrbuf_puts(wrbuf, "<name>");
 931         if (!ht[i].name || !ht[i].name[0])
 932             wrbuf_xmlputs(wrbuf, "NO TARGET NAME");
 933         else
 934             wrbuf_xmlputs(wrbuf, ht[i].name);
 935         wrbuf_puts(wrbuf, "</name>\n");
 936
 937         wrbuf_printf(wrbuf, "<frequency>" ODR_INT_PRINTF "</frequency>\n",
 938                      ht[i].hits);
 939
 940         wrbuf_puts(wrbuf, "<state>");
 941         wrbuf_xmlputs(wrbuf, ht[i].state);
 942         wrbuf_puts(wrbuf, "</state>\n");
 943
 944         wrbuf_printf(wrbuf, "<diagnostic>%d</diagnostic>\n",
 945                      ht[i].diagnostic);
 946         wrbuf_puts(wrbuf, "</term>\n");
 947     }
 948     return count;
 949 }
 950
 951 void perform_termlist(struct http_channel *c, struct session *se,
 952                       const char *name, int num)
 953 {
 954     int i, j;
 955     NMEM nmem_tmp = nmem_create();
 956     char **names;
 957     int num_names = 0;
 958
 959     if (name)
 960         nmem_strsplit(nmem_tmp, ",", name, &names, &num_names);
 961
 962     session_enter(se);
 963
 964     for (j = 0; j < num_names; j++)
 965     {
 966         const char *tname;
 967         for (i = 0; i < se->num_termlists; i++)
 968         {
 969             tname = se->termlists[i].name;
 970             if (num_names > 0 && !strcmp(names[j], tname))
 971             {
 972                 struct termlist_score **p = 0;
 973                 int len;
 974                 p = termlist_highscore(se->termlists[i].termlist, &len);
 975                 if (p)
 976                 {
 977                     int i;
 978                     wrbuf_puts(c->wrbuf, "<list name=\"");
 979                     wrbuf_xmlputs(c->wrbuf, tname);
 980                     wrbuf_puts(c->wrbuf, "\">\n");
 981                     for (i = 0; i < len && i < num; i++)
 982                     {
 983                         // prevent sending empty term elements
 984                         if (!p[i]->display_term || !p[i]->display_term[0])
 985                             continue;
 986
 987                         wrbuf_puts(c->wrbuf, "<term>");
 988                         wrbuf_puts(c->wrbuf, "<name>");
 989                         wrbuf_xmlputs(c->wrbuf, p[i]->display_term);
 990                         wrbuf_puts(c->wrbuf, "</name>");
 991
 992                         wrbuf_printf(c->wrbuf,
 993                                      "<frequency>%d</frequency>",
 994                                      p[i]->frequency);
 995                         wrbuf_puts(c->wrbuf, "</term>\n");
 996                     }
 997                     wrbuf_puts(c->wrbuf, "</list>\n");
 998                 }
 999             }
1000         }
1001         tname = "xtargets";
1002         if (num_names > 0 && !strcmp(names[j], tname))
1003         {
1004             wrbuf_puts(c->wrbuf, "<list name=\"");
1005             wrbuf_xmlputs(c->wrbuf, tname);
1006             wrbuf_puts(c->wrbuf, "\">\n");
1007             targets_termlist_nb(c->wrbuf, se, num, c->nmem);
1008             wrbuf_puts(c->wrbuf, "</list>\n");
1009         }
1010     }
1011     session_leave(se);
1012     nmem_destroy(nmem_tmp);
1013 }
1014
1015 #ifdef MISSING_HEADERS
1016 void report_nmem_stats(void)
1017 {
1018     size_t in_use, is_free;
1019
1020     nmem_get_memory_in_use(&in_use);
1021     nmem_get_memory_free(&is_free);
1022
1023     yaz_log(YLOG_LOG, "nmem stat: use=%ld free=%ld",
1024             (long) in_use, (long) is_free);
1025 }
1026 #endif
1027
1028 struct record_cluster *show_single_start(struct session *se, const char *id,
1029                                          struct record_cluster **prev_r,
1030                                          struct record_cluster **next_r)
1031 {
1032     struct record_cluster *r = 0;
1033
1034     session_enter(se);
1035     *prev_r = 0;
1036     *next_r = 0;
1037     if (se->reclist)
1038     {
1039         reclist_enter(se->reclist);
1040         while ((r = reclist_read_record(se->reclist)))
1041         {
1042             if (!strcmp(r->recid, id))
1043             {
1044                 *next_r = reclist_read_record(se->reclist);
1045                 break;
1046             }
1047             *prev_r = r;
1048         }
1049         reclist_leave(se->reclist);
1050     }
1051     if (!r)
1052         session_leave(se);
1053     return r;
1054 }
1055
1056 void show_single_stop(struct session *se, struct record_cluster *rec)
1057 {
1058     session_leave(se);
1059 }
1060
1061 struct record_cluster **show_range_start(struct session *se,
1062                                          struct reclist_sortparms *sp,
1063                                          int start, int *num, int *total, Odr_int *sumhits)
1064 {
1065     struct record_cluster **recs;
1066     struct reclist_sortparms *spp;
1067     int i;
1068 #if USE_TIMING
1069     yaz_timing_t t = yaz_timing_create();
1070 #endif
1071     session_enter(se);
1072     recs = nmem_malloc(se->nmem, *num * sizeof(struct record_cluster *));
1073     if (!se->relevance)
1074     {
1075         *num = 0;
1076         *total = 0;
1077         *sumhits = 0;
1078         recs = 0;
1079     }
1080     else
1081     {
1082         for (spp = sp; spp; spp = spp->next)
1083             if (spp->type == Metadata_sortkey_relevance)
1084             {
1085                 relevance_prepare_read(se->relevance, se->reclist);
1086                 break;
1087             }
1088         reclist_sort(se->reclist, sp);
1089
1090         reclist_enter(se->reclist);
1091         *total = reclist_get_num_records(se->reclist);
1092         *sumhits = se->total_hits;
1093
1094         for (i = 0; i < start; i++)
1095             if (!reclist_read_record(se->reclist))
1096             {
1097                 *num = 0;
1098                 recs = 0;
1099                 break;
1100             }
1101
1102         for (i = 0; i < *num; i++)
1103         {
1104             struct record_cluster *r = reclist_read_record(se->reclist);
1105             if (!r)
1106             {
1107                 *num = i;
1108                 break;
1109             }
1110             recs[i] = r;
1111         }
1112         reclist_leave(se->reclist);
1113     }
1114 #if USE_TIMING
1115     yaz_timing_stop(t);
1116     yaz_log(YLOG_LOG, "show %6.5f %3.2f %3.2f",
1117             yaz_timing_get_real(t), yaz_timing_get_user(t),
1118             yaz_timing_get_sys(t));
1119     yaz_timing_destroy(&t);
1120 #endif
1121     return recs;
1122 }
1123
1124 void show_range_stop(struct session *se, struct record_cluster **recs)
1125 {
1126     session_leave(se);
1127 }
1128
1129 void statistics(struct session *se, struct statistics *stat)
1130 {
1131     struct client_list *l;
1132     int count = 0;
1133
1134     memset(stat, 0, sizeof(*stat));
1135     for (l = se->clients; l; l = l->next)
1136     {
1137         struct client *cl = l->client;
1138         if (!client_get_connection(cl))
1139             stat->num_no_connection++;
1140         switch (client_get_state(cl))
1141         {
1142         case Client_Connecting: stat->num_connecting++; break;
1143         case Client_Working: stat->num_working++; break;
1144         case Client_Idle: stat->num_idle++; break;
1145         case Client_Failed: stat->num_failed++; break;
1146         case Client_Error: stat->num_error++; break;
1147         default: break;
1148         }
1149         count++;
1150     }
1151     stat->num_hits = se->total_hits;
1152     stat->num_records = se->total_records;
1153
1154     stat->num_clients = count;
1155 }
1156
1157 static struct record_metadata *record_metadata_init(
1158     NMEM nmem, const char *value, enum conf_metadata_type type,
1159     struct _xmlAttr *attr)
1160 {
1161     struct record_metadata *rec_md = record_metadata_create(nmem);
1162     struct record_metadata_attr **attrp = &rec_md->attributes;
1163
1164     for (; attr; attr = attr->next)
1165     {
1166         if (attr->children && attr->children->content)
1167         {
1168             if (strcmp((const char *) attr->name, "type"))
1169             {  /* skip the "type" attribute.. Its value is already part of
1170                   the element in output (md-%s) and so repeating it here
1171                   is redundant */
1172                 *attrp = nmem_malloc(nmem, sizeof(**attrp));
1173                 (*attrp)->name =
1174                     nmem_strdup(nmem, (const char *) attr->name);
1175                 (*attrp)->value =
1176                     nmem_strdup(nmem, (const char *) attr->children->content);
1177                 attrp = &(*attrp)->next;
1178             }
1179         }
1180     }
1181     *attrp = 0;
1182
1183     if (type == Metadata_type_generic)
1184     {
1185         char *p = nmem_strdup(nmem, value);
1186
1187         p = normalize7bit_generic(p, " ,/.:([");
1188
1189         rec_md->data.text.disp = p;
1190         rec_md->data.text.sort = 0;
1191     }
1192     else if (type == Metadata_type_year || type == Metadata_type_date)
1193     {
1194         int first, last;
1195         int longdate = 0;
1196
1197         if (type == Metadata_type_date)
1198             longdate = 1;
1199         if (extract7bit_dates((char *) value, &first, &last, longdate) < 0)
1200             return 0;
1201
1202         rec_md->data.number.min = first;
1203         rec_md->data.number.max = last;
1204     }
1205     else
1206         return 0;
1207     return rec_md;
1208 }
1209
1210 static int get_mergekey_from_doc(xmlDoc *doc, xmlNode *root, const char *name,
1211                                  struct conf_service *service, WRBUF norm_wr)
1212 {
1213     xmlNode *n;
1214     int no_found = 0;
1215     for (n = root->children; n; n = n->next)
1216     {
1217         if (n->type != XML_ELEMENT_NODE)
1218             continue;
1219         if (!strcmp((const char *) n->name, "metadata"))
1220         {
1221             xmlChar *type = xmlGetProp(n, (xmlChar *) "type");
1222             if (type == NULL) {
1223                 yaz_log(YLOG_FATAL, "Missing type attribute on metadata element. Skipping!");
1224             }
1225             else if (!strcmp(name, (const char *) type))
1226             {
1227                 xmlChar *value = xmlNodeListGetString(doc, n->children, 1);
1228                 if (value)
1229                 {
1230                     const char *norm_str;
1231                     pp2_charset_token_t prt =
1232                         pp2_charset_token_create(service->charsets, "mergekey");
1233
1234                     pp2_charset_token_first(prt, (const char *) value, 0);
1235                     if (wrbuf_len(norm_wr) > 0)
1236                         wrbuf_puts(norm_wr, " ");
1237                     wrbuf_puts(norm_wr, name);
1238                     while ((norm_str =
1239                             pp2_charset_token_next(prt)))
1240                     {
1241                         if (*norm_str)
1242                         {
1243                             wrbuf_puts(norm_wr, " ");
1244                             wrbuf_puts(norm_wr, norm_str);
1245                         }
1246                     }
1247                     xmlFree(value);
1248                     pp2_charset_token_destroy(prt);
1249                     no_found++;
1250                 }
1251             }
1252             xmlFree(type);
1253         }
1254     }
1255     return no_found;
1256 }
1257
1258 static const char *get_mergekey(xmlDoc *doc, struct client *cl, int record_no,
1259                                 struct conf_service *service, NMEM nmem)
1260 {
1261     char *mergekey_norm = 0;
1262     xmlNode *root = xmlDocGetRootElement(doc);
1263     WRBUF norm_wr = wrbuf_alloc();
1264
1265     /* consider mergekey from XSL first */
1266     xmlChar *mergekey = xmlGetProp(root, (xmlChar *) "mergekey");
1267     if (mergekey)
1268     {
1269         const char *norm_str;
1270         pp2_charset_token_t prt =
1271             pp2_charset_token_create(service->charsets, "mergekey");
1272
1273         pp2_charset_token_first(prt, (const char *) mergekey, 0);
1274         while ((norm_str = pp2_charset_token_next(prt)))
1275         {
1276             if (*norm_str)
1277             {
1278                 if (wrbuf_len(norm_wr))
1279                     wrbuf_puts(norm_wr, " ");
1280                 wrbuf_puts(norm_wr, norm_str);
1281             }
1282         }
1283         pp2_charset_token_destroy(prt);
1284         xmlFree(mergekey);
1285     }
1286     else
1287     {
1288         /* no mergekey defined in XSL. Look for mergekey metadata instead */
1289         int field_id;
1290         for (field_id = 0; field_id < service->num_metadata; field_id++)
1291         {
1292             struct conf_metadata *ser_md = &service->metadata[field_id];
1293             if (ser_md->mergekey != Metadata_mergekey_no)
1294             {
1295                 int r = get_mergekey_from_doc(doc, root, ser_md->name,
1296                                               service, norm_wr);
1297                 if (r == 0 && ser_md->mergekey == Metadata_mergekey_required)
1298                 {
1299                     /* no mergekey on this one and it is required..
1300                        Generate unique key instead */
1301                     wrbuf_rewind(norm_wr);
1302                     break;
1303                 }
1304             }
1305         }
1306     }
1307
1308     /* generate unique key if none is not generated already or is empty */
1309     if (wrbuf_len(norm_wr) == 0)
1310     {
1311         wrbuf_printf(norm_wr, "%s-%d",
1312                      client_get_database(cl)->database->url, record_no);
1313     }
1314     if (wrbuf_len(norm_wr) > 0)
1315         mergekey_norm = nmem_strdup(nmem, wrbuf_cstr(norm_wr));
1316     wrbuf_destroy(norm_wr);
1317     return mergekey_norm;
1318 }
1319
1320 /** \brief see if metadata for pz:recordfilter exists
1321     \param root xml root element of normalized record
1322     \param sdb session database for client
1323     \retval 0 if there is no metadata for pz:recordfilter
1324     \retval 1 if there is metadata for pz:recordfilter
1325
1326     If there is no pz:recordfilter defined, this function returns 1
1327     as well.
1328 */
1329
1330 static int check_record_filter(xmlNode *root, struct session_database *sdb)
1331 {
1332     int match = 0;
1333     xmlNode *n;
1334     const char *s;
1335     s = session_setting_oneval(sdb, PZ_RECORDFILTER);
1336
1337     if (!s || !*s)
1338         return 1;
1339
1340     for (n = root->children; n; n = n->next)
1341     {
1342         if (n->type != XML_ELEMENT_NODE)
1343             continue;
1344         if (!strcmp((const char *) n->name, "metadata"))
1345         {
1346             xmlChar *type = xmlGetProp(n, (xmlChar *) "type");
1347             if (type)
1348             {
1349                 size_t len;
1350                 int substring;
1351                 const char *eq;
1352
1353                 if ((eq = strchr(s, '=')))
1354                     substring = 0;
1355                 else if ((eq = strchr(s, '~')))
1356                     substring = 1;
1357                 if (eq)
1358                     len = eq - s;
1359                 else
1360                     len = strlen(s);
1361                 if (len == strlen((const char *)type) &&
1362                     !memcmp((const char *) type, s, len))
1363                 {
1364                     xmlChar *value = xmlNodeGetContent(n);
1365                     if (value && *value)
1366                     {
1367                         if (!eq ||
1368                             (substring && strstr((const char *) value, eq+1)) ||
1369                             (!substring && !strcmp((const char *) value, eq + 1)))
1370                             match = 1;
1371                     }
1372                     xmlFree(value);
1373                 }
1374                 xmlFree(type);
1375             }
1376         }
1377     }
1378     return match;
1379 }
1380
1381
1382 static int ingest_to_cluster(struct client *cl,
1383                              xmlDoc *xdoc,
1384                              xmlNode *root,
1385                              int record_no,
1386                              const char *mergekey_norm);
1387
1388 /** \brief ingest XML record
1389     \param cl client holds the result set for record
1390     \param rec record buffer (0 terminated)
1391     \param record_no record position (1, 2, ..)
1392     \param nmem working NMEM
1393     \retval 0 OK
1394     \retval -1 failure
1395     \retval -2 Filtered
1396 */
1397 int ingest_record(struct client *cl, const char *rec,
1398                   int record_no, NMEM nmem)
1399 {
1400     struct session *se = client_get_session(cl);
1401     int ret = 0;
1402     struct session_database *sdb = client_get_database(cl);
1403     struct conf_service *service = se->service;
1404     xmlDoc *xdoc = normalize_record(se, sdb, service, rec, nmem);
1405     xmlNode *root;
1406     const char *mergekey_norm;
1407
1408     if (!xdoc)
1409         return -1;
1410
1411     root = xmlDocGetRootElement(xdoc);
1412
1413     if (!check_record_filter(root, sdb))
1414     {
1415         session_log(se, YLOG_LOG, "Filtered out record no %d from %s",
1416                     record_no, sdb->database->url);
1417         xmlFreeDoc(xdoc);
1418         return -2;
1419     }
1420
1421     mergekey_norm = get_mergekey(xdoc, cl, record_no, service, nmem);
1422     if (!mergekey_norm)
1423     {
1424         session_log(se, YLOG_WARN, "Got no mergekey");
1425         xmlFreeDoc(xdoc);
1426         return -1;
1427     }
1428     session_enter(se);
1429     if (client_get_session(cl) == se)
1430         ret = ingest_to_cluster(cl, xdoc, root, record_no, mergekey_norm);
1431     session_leave(se);
1432
1433     xmlFreeDoc(xdoc);
1434     return ret;
1435 }
1436
1437 static int ingest_to_cluster(struct client *cl,
1438                              xmlDoc *xdoc,
1439                              xmlNode *root,
1440                              int record_no,
1441                              const char *mergekey_norm)
1442 {
1443     xmlNode *n;
1444     xmlChar *type = 0;
1445     xmlChar *value = 0;
1446     struct session_database *sdb = client_get_database(cl);
1447     struct session *se = client_get_session(cl);
1448     struct conf_service *service = se->service;
1449     struct record *record = record_create(se->nmem,
1450                                           service->num_metadata,
1451                                           service->num_sortkeys, cl,
1452                                           record_no);
1453     struct record_cluster *cluster = reclist_insert(se->reclist,
1454                                                     service,
1455                                                     record,
1456                                                     mergekey_norm,
1457                                                     &se->total_merged);
1458
1459     const char *use_term_factor_str = session_setting_oneval(sdb, PZ_TERMLIST_TERM_FACTOR);
1460     int use_term_factor = 0;
1461     int term_factor = 1;
1462     if (use_term_factor_str && use_term_factor_str[0] != 0)
1463        use_term_factor =  atoi(use_term_factor_str);
1464     if (use_term_factor) {
1465         int maxrecs = client_get_maxrecs(cl);
1466         int hits = (int) client_get_hits(cl);
1467         term_factor = MAX(hits, maxrecs) /  MAX(1, maxrecs);
1468         assert(term_factor >= 1);
1469         yaz_log(YLOG_DEBUG, "Using term factor: %d (%d / %d)", term_factor, MAX(hits, maxrecs), MAX(1, maxrecs));
1470     }
1471
1472     if (!cluster)
1473         return -1;
1474     if (global_parameters.dump_records)
1475         session_log(se, YLOG_LOG, "Cluster id %s from %s (#%d)", cluster->recid,
1476                     sdb->database->url, record_no);
1477     relevance_newrec(se->relevance, cluster);
1478
1479     // now parsing XML record and adding data to cluster or record metadata
1480     for (n = root->children; n; n = n->next)
1481     {
1482         pp2_charset_token_t prt;
1483         if (type)
1484             xmlFree(type);
1485         if (value)
1486             xmlFree(value);
1487         type = value = 0;
1488
1489         if (n->type != XML_ELEMENT_NODE)
1490             continue;
1491         if (!strcmp((const char *) n->name, "metadata"))
1492         {
1493             struct conf_metadata *ser_md = 0;
1494             struct conf_sortkey *ser_sk = 0;
1495             struct record_metadata **wheretoput = 0;
1496             struct record_metadata *rec_md = 0;
1497             int md_field_id = -1;
1498             int sk_field_id = -1;
1499
1500             type = xmlGetProp(n, (xmlChar *) "type");
1501             value = xmlNodeListGetString(xdoc, n->children, 1);
1502
1503             if (!type || !value || !*value)
1504                 continue;
1505
1506             md_field_id
1507                 = conf_service_metadata_field_id(service, (const char *) type);
1508             if (md_field_id < 0)
1509             {
1510                 if (se->number_of_warnings_unknown_metadata == 0)
1511                 {
1512                     session_log(se, YLOG_WARN,
1513                             "Ignoring unknown metadata element: %s", type);
1514                 }
1515                 se->number_of_warnings_unknown_metadata++;
1516                 continue;
1517             }
1518
1519             ser_md = &service->metadata[md_field_id];
1520
1521             if (ser_md->sortkey_offset >= 0){
1522                 sk_field_id = ser_md->sortkey_offset;
1523                 ser_sk = &service->sortkeys[sk_field_id];
1524             }
1525
1526             // non-merged metadata
1527             rec_md = record_metadata_init(se->nmem, (const char *) value,
1528                                           ser_md->type, n->properties);
1529             if (!rec_md)
1530             {
1531                 session_log(se, YLOG_WARN, "bad metadata data '%s' "
1532                             "for element '%s'", value, type);
1533                 continue;
1534             }
1535             wheretoput = &record->metadata[md_field_id];
1536             while (*wheretoput)
1537                 wheretoput = &(*wheretoput)->next;
1538             *wheretoput = rec_md;
1539
1540             // merged metadata
1541             rec_md = record_metadata_init(se->nmem, (const char *) value,
1542                                           ser_md->type, 0);
1543             wheretoput = &cluster->metadata[md_field_id];
1544
1545             // and polulate with data:
1546             // assign cluster or record based on merge action
1547             if (ser_md->merge == Metadata_merge_unique)
1548             {
1549                 while (*wheretoput)
1550                 {
1551                     if (!strcmp((const char *) (*wheretoput)->data.text.disp,
1552                                 rec_md->data.text.disp))
1553                         break;
1554                     wheretoput = &(*wheretoput)->next;
1555                 }
1556                 if (!*wheretoput)
1557                     *wheretoput = rec_md;
1558             }
1559             else if (ser_md->merge == Metadata_merge_longest)
1560             {
1561                 if (!*wheretoput
1562                     || strlen(rec_md->data.text.disp)
1563                     > strlen((*wheretoput)->data.text.disp))
1564                 {
1565                     *wheretoput = rec_md;
1566                     if (ser_sk)
1567                     {
1568                         const char *sort_str = 0;
1569                         int skip_article =
1570                             ser_sk->type == Metadata_sortkey_skiparticle;
1571
1572                         if (!cluster->sortkeys[sk_field_id])
1573                             cluster->sortkeys[sk_field_id] =
1574                                 nmem_malloc(se->nmem,
1575                                             sizeof(union data_types));
1576
1577                         prt =
1578                             pp2_charset_token_create(service->charsets, "sort");
1579
1580                         pp2_charset_token_first(prt, rec_md->data.text.disp,
1581                                                 skip_article);
1582
1583                         pp2_charset_token_next(prt);
1584
1585                         sort_str = pp2_get_sort(prt);
1586
1587                         cluster->sortkeys[sk_field_id]->text.disp =
1588                             rec_md->data.text.disp;
1589                         if (!sort_str)
1590                         {
1591                             sort_str = rec_md->data.text.disp;
1592                             session_log(se, YLOG_WARN,
1593                                     "Could not make sortkey. Bug #1858");
1594                         }
1595                         cluster->sortkeys[sk_field_id]->text.sort =
1596                             nmem_strdup(se->nmem, sort_str);
1597                         pp2_charset_token_destroy(prt);
1598                     }
1599                 }
1600             }
1601             else if (ser_md->merge == Metadata_merge_all)
1602             {
1603                 while (*wheretoput)
1604                     wheretoput = &(*wheretoput)->next;
1605                 *wheretoput = rec_md;
1606             }
1607             else if (ser_md->merge == Metadata_merge_range)
1608             {
1609                 if (!*wheretoput)
1610                 {
1611                     *wheretoput = rec_md;
1612                     if (ser_sk)
1613                         cluster->sortkeys[sk_field_id]
1614                             = &rec_md->data;
1615                 }
1616                 else
1617                 {
1618                     int this_min = rec_md->data.number.min;
1619                     int this_max = rec_md->data.number.max;
1620                     if (this_min < (*wheretoput)->data.number.min)
1621                         (*wheretoput)->data.number.min = this_min;
1622                     if (this_max > (*wheretoput)->data.number.max)
1623                         (*wheretoput)->data.number.max = this_max;
1624                 }
1625             }
1626
1627
1628             // ranking of _all_ fields enabled ...
1629             if (ser_md->rank)
1630                 relevance_countwords(se->relevance, cluster,
1631                                      (char *) value, ser_md->rank,
1632                                      ser_md->name);
1633
1634             // construct facets ... unless the client already has reported them
1635             if (ser_md->termlist && !client_has_facet(cl, (char *) type))
1636             {
1637                 if (ser_md->type == Metadata_type_year)
1638                 {
1639                     char year[64];
1640                     sprintf(year, "%d", rec_md->data.number.max);
1641
1642                     add_facet(se, (char *) type, year, term_factor);
1643                     if (rec_md->data.number.max != rec_md->data.number.min)
1644                     {
1645                         sprintf(year, "%d", rec_md->data.number.min);
1646                         add_facet(se, (char *) type, year, term_factor);
1647                     }
1648                 }
1649                 else
1650                     add_facet(se, (char *) type, (char *) value, term_factor);
1651             }
1652
1653             // cleaning up
1654             xmlFree(type);
1655             xmlFree(value);
1656             type = value = 0;
1657         }
1658         else
1659         {
1660             if (se->number_of_warnings_unknown_elements == 0)
1661                 session_log(se, YLOG_WARN,
1662                         "Unexpected element in internal record: %s", n->name);
1663             se->number_of_warnings_unknown_elements++;
1664         }
1665     }
1666     if (type)
1667         xmlFree(type);
1668     if (value)
1669         xmlFree(value);
1670
1671     relevance_donerecord(se->relevance, cluster);
1672     se->total_records++;
1673
1674     return 0;
1675 }
1676
1677 void session_log(struct session *s, int level, const char *fmt, ...)
1678 {
1679     char buf[1024];
1680     va_list ap;
1681     va_start(ap, fmt);
1682
1683     yaz_vsnprintf(buf, sizeof(buf)-30, fmt, ap);
1684     yaz_log(level, "Session (%u): %s", s->session_id, buf);
1685
1686     va_end(ap);
1687 }
1688
1689 /*
1690  * Local variables:
1691  * c-basic-offset: 4
1692  * c-file-style: "Stroustrup"
1693  * indent-tabs-mode: nil
1694  * End:
1695  * vim: shiftwidth=4 tabstop=8 expandtab
1696  */
1697