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->id);
 250         return 0;
 251     }
 252
 253     if (global_parameters.dump_records)
 254     {
 255         session_log(se, YLOG_LOG, "Un-normalized record from %s", db->id);
 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->id);
 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->id);
 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->id);
 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->id);
 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->id);
 528     struct client_list *l;
 529
 530     client_set_database(cl, db);
 531
 532     client_set_session(cl, se);
 533
 534     l = xmalloc(sizeof(*l));
 535     l->client = cl;
 536     l->next = se->clients;
 537     se->clients = l;
 538 }
 539
 540 static void session_remove_clients(struct session *se)
 541 {
 542     struct client_list *l;
 543
 544     session_enter(se);
 545     l = se->clients;
 546     se->clients = 0;
 547     session_leave(se);
 548
 549     while (l)
 550     {
 551         struct client_list *l_next = l->next;
 552         client_lock(l->client);
 553         client_set_session(l->client, 0);
 554         client_set_database(l->client, 0);
 555         client_unlock(l->client);
 556         client_destroy(l->client);
 557         xfree(l);
 558         l = l_next;
 559     }
 560 }
 561
 562 // Associates a set of clients with a session;
 563 // Note: Session-databases represent databases with per-session
 564 // setting overrides
 565 static int select_targets(struct session *se, const char *filter)
 566 {
 567     return session_grep_databases(se, filter, select_targets_callback);
 568 }
 569
 570 int session_active_clients(struct session *s)
 571 {
 572     struct client_list *l;
 573     int res = 0;
 574
 575     for (l = s->clients; l; l = l->next)
 576         if (client_is_active(l->client))
 577             res++;
 578
 579     return res;
 580 }
 581
 582 int session_is_preferred_clients_ready(struct session *s)
 583 {
 584     struct client_list *l;
 585     int res = 0;
 586
 587     for (l = s->clients; l; l = l->next)
 588         if (client_is_active_preferred(l->client))
 589             res++;
 590     session_log(s, YLOG_DEBUG, "Has %d active preferred clients.", res);
 591     return res == 0;
 592 }
 593
 594 enum pazpar2_error_code search(struct session *se,
 595                                const char *query,
 596                                const char *startrecs, const char *maxrecs,
 597                                const char *filter,
 598                                const char *limit,
 599                                const char **addinfo)
 600 {
 601     int live_channels = 0;
 602     int no_working = 0;
 603     int no_failed = 0;
 604     struct client_list *l;
 605     struct timeval tval;
 606     facet_limits_t facet_limits;
 607
 608     session_log(se, YLOG_DEBUG, "Search");
 609
 610     *addinfo = 0;
 611
 612     session_remove_clients(se);
 613
 614     session_enter(se);
 615     reclist_destroy(se->reclist);
 616     se->reclist = 0;
 617     relevance_destroy(&se->relevance);
 618     nmem_reset(se->nmem);
 619     se->total_records = se->total_hits = se->total_merged = 0;
 620     se->num_termlists = 0;
 621     live_channels = select_targets(se, filter);
 622     if (!live_channels)
 623     {
 624         session_leave(se);
 625         return PAZPAR2_NO_TARGETS;
 626     }
 627     se->reclist = reclist_create(se->nmem);
 628
 629     yaz_gettimeofday(&tval);
 630
 631     tval.tv_sec += 5;
 632
 633     facet_limits = facet_limits_create(limit);
 634     if (!facet_limits)
 635     {
 636         *addinfo = "limit";
 637         session_leave(se);
 638         return PAZPAR2_MALFORMED_PARAMETER_VALUE;
 639     }
 640     for (l = se->clients; l; l = l->next)
 641     {
 642         struct client *cl = l->client;
 643
 644         if (maxrecs)
 645             client_set_maxrecs(cl, atoi(maxrecs));
 646         if (startrecs)
 647             client_set_startrecs(cl, atoi(startrecs));
 648         if (prepare_session_database(se, client_get_database(cl)) < 0)
 649             ;
 650         else if (client_parse_query(cl, query, facet_limits) < 0)
 651             no_failed++;
 652         else
 653         {
 654             no_working++;
 655             if (client_prep_connection(cl, se->service->z3950_operation_timeout,
 656                                        se->service->z3950_session_timeout,
 657                                        se->service->server->iochan_man,
 658                                        &tval))
 659                 client_start_search(cl);
 660         }
 661     }
 662     facet_limits_destroy(facet_limits);
 663     session_leave(se);
 664     if (no_working == 0)
 665     {
 666         if (no_failed > 0)
 667         {
 668             *addinfo = "query";
 669             return PAZPAR2_MALFORMED_PARAMETER_VALUE;
 670         }
 671         else
 672             return PAZPAR2_NO_TARGETS;
 673     }
 674     return PAZPAR2_NO_ERROR;
 675 }
 676
 677 // Creates a new session_database object for a database
 678 static void session_init_databases_fun(void *context, struct database *db)
 679 {
 680     struct session *se = (struct session *) context;
 681     struct session_database *new = nmem_malloc(se->session_nmem, sizeof(*new));
 682     int i;
 683
 684     new->database = db;
 685
 686     new->map = 0;
 687     assert(db->settings);
 688     new->settings = nmem_malloc(se->session_nmem,
 689                                 sizeof(struct settings *) * db->num_settings);
 690     new->num_settings = db->num_settings;
 691     for (i = 0; i < db->num_settings; i++)
 692     {
 693         struct setting *setting = db->settings[i];
 694         new->settings[i] = setting;
 695     }
 696     new->next = se->databases;
 697     se->databases = new;
 698 }
 699
 700 // Doesn't free memory associated with sdb -- nmem takes care of that
 701 static void session_database_destroy(struct session_database *sdb)
 702 {
 703     sdb->map = 0;
 704 }
 705
 706 // Initialize session_database list -- this represents this session's view
 707 // of the database list -- subject to modification by the settings ws command
 708 void session_init_databases(struct session *se)
 709 {
 710     se->databases = 0;
 711     predef_grep_databases(se, se->service, session_init_databases_fun);
 712 }
 713
 714 // Probably session_init_databases_fun should be refactored instead of
 715 // called here.
 716 static struct session_database *load_session_database(struct session *se,
 717                                                       char *id)
 718 {
 719     struct database *db = new_database(id, se->session_nmem);
 720
 721     session_init_databases_fun((void*) se, db);
 722
 723     // New sdb is head of se->databases list
 724     return se->databases;
 725 }
 726
 727 // Find an existing session database. If not found, load it
 728 static struct session_database *find_session_database(struct session *se,
 729                                                       char *id)
 730 {
 731     struct session_database *sdb;
 732
 733     for (sdb = se->databases; sdb; sdb = sdb->next)
 734         if (!strcmp(sdb->database->id, id))
 735             return sdb;
 736     return load_session_database(se, id);
 737 }
 738
 739 // Apply a session override to a database
 740 void session_apply_setting(struct session *se, char *dbname, char *setting,
 741                            char *value)
 742 {
 743     struct session_database *sdb = find_session_database(se, dbname);
 744     struct conf_service *service = se->service;
 745     struct setting *new = nmem_malloc(se->session_nmem, sizeof(*new));
 746     int offset = settings_create_offset(service, setting);
 747
 748     expand_settings_array(&sdb->settings, &sdb->num_settings, offset,
 749                           se->session_nmem);
 750     new->precedence = 0;
 751     new->target = dbname;
 752     new->name = setting;
 753     new->value = value;
 754     new->next = sdb->settings[offset];
 755     sdb->settings[offset] = new;
 756
 757     // Force later recompute of settings-driven data structures
 758     // (happens when a search starts and client connections are prepared)
 759     switch (offset)
 760     {
 761     case PZ_XSLT:
 762         if (sdb->map)
 763         {
 764             sdb->map = 0;
 765         }
 766         break;
 767     }
 768 }
 769
 770 void session_destroy(struct session *se) {
 771     struct session_database *sdb;
 772     session_log(se, YLOG_DEBUG, "Destroying");
 773     session_use(-1);
 774     session_remove_clients(se);
 775
 776     for (sdb = se->databases; sdb; sdb = sdb->next)
 777         session_database_destroy(sdb);
 778     normalize_cache_destroy(se->normalize_cache);
 779     relevance_destroy(&se->relevance);
 780     reclist_destroy(se->reclist);
 781     nmem_destroy(se->nmem);
 782     service_destroy(se->service);
 783     yaz_mutex_destroy(&se->session_mutex);
 784 }
 785
 786 /* Depreciated: use session_destroy */
 787 void destroy_session(struct session *se)
 788 {
 789     session_destroy(se);
 790 }
 791
 792 size_t session_get_memory_status(struct session *session) {
 793     size_t session_nmem;
 794     if (session == 0)
 795         return 0;
 796     session_enter(session);
 797     session_nmem = nmem_total(session->nmem);
 798     session_leave(session);
 799     return session_nmem;
 800 }
 801
 802
 803 struct session *new_session(NMEM nmem, struct conf_service *service,
 804                             unsigned session_id)
 805 {
 806     int i;
 807     struct session *session = nmem_malloc(nmem, sizeof(*session));
 808
 809     char tmp_str[50];
 810
 811     sprintf(tmp_str, "session#%u", session_id);
 812
 813     session->session_id = session_id;
 814     session_log(session, YLOG_DEBUG, "New");
 815     session->service = service;
 816     session->relevance = 0;
 817     session->total_hits = 0;
 818     session->total_records = 0;
 819     session->number_of_warnings_unknown_elements = 0;
 820     session->number_of_warnings_unknown_metadata = 0;
 821     session->num_termlists = 0;
 822     session->reclist = 0;
 823     session->clients = 0;
 824     session->session_nmem = nmem;
 825     session->nmem = nmem_create();
 826     session->databases = 0;
 827     for (i = 0; i <= SESSION_WATCH_MAX; i++)
 828     {
 829         session->watchlist[i].data = 0;
 830         session->watchlist[i].fun = 0;
 831     }
 832     session->normalize_cache = normalize_cache_create();
 833     session->session_mutex = 0;
 834     pazpar2_mutex_create(&session->session_mutex, tmp_str);
 835     session_use(1);
 836     return session;
 837 }
 838
 839 static struct hitsbytarget *hitsbytarget_nb(struct session *se,
 840                                             int *count, NMEM nmem)
 841 {
 842     struct hitsbytarget *res = 0;
 843     struct client_list *l;
 844     size_t sz = 0;
 845
 846     for (l = se->clients; l; l = l->next)
 847         sz++;
 848
 849     res = nmem_malloc(nmem, sizeof(*res) * sz);
 850     *count = 0;
 851     for (l = se->clients; l; l = l->next)
 852     {
 853         struct client *cl = l->client;
 854         WRBUF w = wrbuf_alloc();
 855         const char *name = session_setting_oneval(client_get_database(cl),
 856                                                   PZ_NAME);
 857
 858         res[*count].id = client_get_id(cl);
 859         res[*count].name = *name ? name : "Unknown";
 860         res[*count].hits = client_get_hits(cl);
 861         res[*count].records = client_get_num_records(cl);
 862         res[*count].diagnostic = client_get_diagnostic(cl);
 863         res[*count].state = client_get_state_str(cl);
 864         res[*count].connected  = client_get_connection(cl) ? 1 : 0;
 865         session_settings_dump(se, client_get_database(cl), w);
 866         res[*count].settings_xml = nmem_strdup(nmem, wrbuf_cstr(w));
 867         wrbuf_destroy(w);
 868         (*count)++;
 869     }
 870     return res;
 871 }
 872
 873 struct hitsbytarget *get_hitsbytarget(struct session *se, int *count, NMEM nmem)
 874 {
 875     struct hitsbytarget *p;
 876     session_enter(se);
 877     p = hitsbytarget_nb(se, count, nmem);
 878     session_leave(se);
 879     return p;
 880 }
 881
 882 struct termlist_score **get_termlist_score(struct session *se,
 883                                            const char *name, int *num)
 884 {
 885     int i;
 886     struct termlist_score **tl = 0;
 887
 888     session_enter(se);
 889     for (i = 0; i < se->num_termlists; i++)
 890         if (!strcmp((const char *) se->termlists[i].name, name))
 891         {
 892             tl = termlist_highscore(se->termlists[i].termlist, num);
 893             break;
 894         }
 895     session_leave(se);
 896     return tl;
 897 }
 898
 899 // Compares two hitsbytarget nodes by hitcount
 900 static int cmp_ht(const void *p1, const void *p2)
 901 {
 902     const struct hitsbytarget *h1 = p1;
 903     const struct hitsbytarget *h2 = p2;
 904     return h2->hits - h1->hits;
 905 }
 906
 907 static int targets_termlist_nb(WRBUF wrbuf, struct session *se, int num,
 908                                NMEM nmem)
 909 {
 910     struct hitsbytarget *ht;
 911     int count, i;
 912
 913     ht = hitsbytarget_nb(se, &count, nmem);
 914     qsort(ht, count, sizeof(struct hitsbytarget), cmp_ht);
 915     for (i = 0; i < count && i < num && ht[i].hits > 0; i++)
 916     {
 917
 918         // do only print terms which have display names
 919
 920         wrbuf_puts(wrbuf, "<term>\n");
 921
 922         wrbuf_puts(wrbuf, "<id>");
 923         wrbuf_xmlputs(wrbuf, ht[i].id);
 924         wrbuf_puts(wrbuf, "</id>\n");
 925
 926         wrbuf_puts(wrbuf, "<name>");
 927         if (!ht[i].name || !ht[i].name[0])
 928             wrbuf_xmlputs(wrbuf, "NO TARGET NAME");
 929         else
 930             wrbuf_xmlputs(wrbuf, ht[i].name);
 931         wrbuf_puts(wrbuf, "</name>\n");
 932
 933         wrbuf_printf(wrbuf, "<frequency>" ODR_INT_PRINTF "</frequency>\n",
 934                      ht[i].hits);
 935
 936         wrbuf_puts(wrbuf, "<state>");
 937         wrbuf_xmlputs(wrbuf, ht[i].state);
 938         wrbuf_puts(wrbuf, "</state>\n");
 939
 940         wrbuf_printf(wrbuf, "<diagnostic>%d</diagnostic>\n",
 941                      ht[i].diagnostic);
 942         wrbuf_puts(wrbuf, "</term>\n");
 943     }
 944     return count;
 945 }
 946
 947 void perform_termlist(struct http_channel *c, struct session *se,
 948                       const char *name, int num)
 949 {
 950     int i, j;
 951     NMEM nmem_tmp = nmem_create();
 952     char **names;
 953     int num_names = 0;
 954
 955     if (name)
 956         nmem_strsplit(nmem_tmp, ",", name, &names, &num_names);
 957
 958     session_enter(se);
 959
 960     for (j = 0; j < num_names; j++)
 961     {
 962         const char *tname;
 963         for (i = 0; i < se->num_termlists; i++)
 964         {
 965             tname = se->termlists[i].name;
 966             if (num_names > 0 && !strcmp(names[j], tname))
 967             {
 968                 struct termlist_score **p = 0;
 969                 int len;
 970                 p = termlist_highscore(se->termlists[i].termlist, &len);
 971                 if (p)
 972                 {
 973                     int i;
 974                     wrbuf_puts(c->wrbuf, "<list name=\"");
 975                     wrbuf_xmlputs(c->wrbuf, tname);
 976                     wrbuf_puts(c->wrbuf, "\">\n");
 977                     for (i = 0; i < len && i < num; i++)
 978                     {
 979                         // prevent sending empty term elements
 980                         if (!p[i]->display_term || !p[i]->display_term[0])
 981                             continue;
 982
 983                         wrbuf_puts(c->wrbuf, "<term>");
 984                         wrbuf_puts(c->wrbuf, "<name>");
 985                         wrbuf_xmlputs(c->wrbuf, p[i]->display_term);
 986                         wrbuf_puts(c->wrbuf, "</name>");
 987
 988                         wrbuf_printf(c->wrbuf,
 989                                      "<frequency>%d</frequency>",
 990                                      p[i]->frequency);
 991                         wrbuf_puts(c->wrbuf, "</term>\n");
 992                     }
 993                     wrbuf_puts(c->wrbuf, "</list>\n");
 994                 }
 995             }
 996         }
 997         tname = "xtargets";
 998         if (num_names > 0 && !strcmp(names[j], tname))
 999         {
1000             wrbuf_puts(c->wrbuf, "<list name=\"");
1001             wrbuf_xmlputs(c->wrbuf, tname);
1002             wrbuf_puts(c->wrbuf, "\">\n");
1003             targets_termlist_nb(c->wrbuf, se, num, c->nmem);
1004             wrbuf_puts(c->wrbuf, "</list>\n");
1005         }
1006     }
1007     session_leave(se);
1008     nmem_destroy(nmem_tmp);
1009 }
1010
1011 #ifdef MISSING_HEADERS
1012 void report_nmem_stats(void)
1013 {
1014     size_t in_use, is_free;
1015
1016     nmem_get_memory_in_use(&in_use);
1017     nmem_get_memory_free(&is_free);
1018
1019     yaz_log(YLOG_LOG, "nmem stat: use=%ld free=%ld",
1020             (long) in_use, (long) is_free);
1021 }
1022 #endif
1023
1024 struct record_cluster *show_single_start(struct session *se, const char *id,
1025                                          struct record_cluster **prev_r,
1026                                          struct record_cluster **next_r)
1027 {
1028     struct record_cluster *r = 0;
1029
1030     session_enter(se);
1031     *prev_r = 0;
1032     *next_r = 0;
1033     if (se->reclist)
1034     {
1035         reclist_enter(se->reclist);
1036         while ((r = reclist_read_record(se->reclist)))
1037         {
1038             if (!strcmp(r->recid, id))
1039             {
1040                 *next_r = reclist_read_record(se->reclist);
1041                 break;
1042             }
1043             *prev_r = r;
1044         }
1045         reclist_leave(se->reclist);
1046     }
1047     if (!r)
1048         session_leave(se);
1049     return r;
1050 }
1051
1052 void show_single_stop(struct session *se, struct record_cluster *rec)
1053 {
1054     session_leave(se);
1055 }
1056
1057 struct record_cluster **show_range_start(struct session *se,
1058                                          struct reclist_sortparms *sp,
1059                                          int start, int *num, int *total, Odr_int *sumhits)
1060 {
1061     struct record_cluster **recs;
1062     struct reclist_sortparms *spp;
1063     int i;
1064 #if USE_TIMING
1065     yaz_timing_t t = yaz_timing_create();
1066 #endif
1067     session_enter(se);
1068     recs = nmem_malloc(se->nmem, *num * sizeof(struct record_cluster *));
1069     if (!se->relevance)
1070     {
1071         *num = 0;
1072         *total = 0;
1073         *sumhits = 0;
1074         recs = 0;
1075     }
1076     else
1077     {
1078         for (spp = sp; spp; spp = spp->next)
1079             if (spp->type == Metadata_sortkey_relevance)
1080             {
1081                 relevance_prepare_read(se->relevance, se->reclist);
1082                 break;
1083             }
1084         reclist_sort(se->reclist, sp);
1085
1086         reclist_enter(se->reclist);
1087         *total = reclist_get_num_records(se->reclist);
1088         *sumhits = se->total_hits;
1089
1090         for (i = 0; i < start; i++)
1091             if (!reclist_read_record(se->reclist))
1092             {
1093                 *num = 0;
1094                 recs = 0;
1095                 break;
1096             }
1097
1098         for (i = 0; i < *num; i++)
1099         {
1100             struct record_cluster *r = reclist_read_record(se->reclist);
1101             if (!r)
1102             {
1103                 *num = i;
1104                 break;
1105             }
1106             recs[i] = r;
1107         }
1108         reclist_leave(se->reclist);
1109     }
1110 #if USE_TIMING
1111     yaz_timing_stop(t);
1112     yaz_log(YLOG_LOG, "show %6.5f %3.2f %3.2f",
1113             yaz_timing_get_real(t), yaz_timing_get_user(t),
1114             yaz_timing_get_sys(t));
1115     yaz_timing_destroy(&t);
1116 #endif
1117     return recs;
1118 }
1119
1120 void show_range_stop(struct session *se, struct record_cluster **recs)
1121 {
1122     session_leave(se);
1123 }
1124
1125 void statistics(struct session *se, struct statistics *stat)
1126 {
1127     struct client_list *l;
1128     int count = 0;
1129
1130     memset(stat, 0, sizeof(*stat));
1131     for (l = se->clients; l; l = l->next)
1132     {
1133         struct client *cl = l->client;
1134         if (!client_get_connection(cl))
1135             stat->num_no_connection++;
1136         switch (client_get_state(cl))
1137         {
1138         case Client_Connecting: stat->num_connecting++; break;
1139         case Client_Working: stat->num_working++; break;
1140         case Client_Idle: stat->num_idle++; break;
1141         case Client_Failed: stat->num_failed++; break;
1142         case Client_Error: stat->num_error++; break;
1143         default: break;
1144         }
1145         count++;
1146     }
1147     stat->num_hits = se->total_hits;
1148     stat->num_records = se->total_records;
1149
1150     stat->num_clients = count;
1151 }
1152
1153 static struct record_metadata *record_metadata_init(
1154     NMEM nmem, const char *value, enum conf_metadata_type type,
1155     struct _xmlAttr *attr)
1156 {
1157     struct record_metadata *rec_md = record_metadata_create(nmem);
1158     struct record_metadata_attr **attrp = &rec_md->attributes;
1159
1160     for (; attr; attr = attr->next)
1161     {
1162         if (attr->children && attr->children->content)
1163         {
1164             if (strcmp((const char *) attr->name, "type"))
1165             {  /* skip the "type" attribute.. Its value is already part of
1166                   the element in output (md-%s) and so repeating it here
1167                   is redundant */
1168                 *attrp = nmem_malloc(nmem, sizeof(**attrp));
1169                 (*attrp)->name =
1170                     nmem_strdup(nmem, (const char *) attr->name);
1171                 (*attrp)->value =
1172                     nmem_strdup(nmem, (const char *) attr->children->content);
1173                 attrp = &(*attrp)->next;
1174             }
1175         }
1176     }
1177     *attrp = 0;
1178
1179     if (type == Metadata_type_generic)
1180     {
1181         char *p = nmem_strdup(nmem, value);
1182
1183         p = normalize7bit_generic(p, " ,/.:([");
1184
1185         rec_md->data.text.disp = p;
1186         rec_md->data.text.sort = 0;
1187     }
1188     else if (type == Metadata_type_year || type == Metadata_type_date)
1189     {
1190         int first, last;
1191         int longdate = 0;
1192
1193         if (type == Metadata_type_date)
1194             longdate = 1;
1195         if (extract7bit_dates((char *) value, &first, &last, longdate) < 0)
1196             return 0;
1197
1198         rec_md->data.number.min = first;
1199         rec_md->data.number.max = last;
1200     }
1201     else
1202         return 0;
1203     return rec_md;
1204 }
1205
1206 static int get_mergekey_from_doc(xmlDoc *doc, xmlNode *root, const char *name,
1207                                  struct conf_service *service, WRBUF norm_wr)
1208 {
1209     xmlNode *n;
1210     int no_found = 0;
1211     for (n = root->children; n; n = n->next)
1212     {
1213         if (n->type != XML_ELEMENT_NODE)
1214             continue;
1215         if (!strcmp((const char *) n->name, "metadata"))
1216         {
1217             xmlChar *type = xmlGetProp(n, (xmlChar *) "type");
1218             if (type == NULL) {
1219                 yaz_log(YLOG_FATAL, "Missing type attribute on metadata element. Skipping!");
1220             }
1221             else if (!strcmp(name, (const char *) type))
1222             {
1223                 xmlChar *value = xmlNodeListGetString(doc, n->children, 1);
1224                 if (value)
1225                 {
1226                     const char *norm_str;
1227                     pp2_charset_token_t prt =
1228                         pp2_charset_token_create(service->charsets, "mergekey");
1229
1230                     pp2_charset_token_first(prt, (const char *) value, 0);
1231                     if (wrbuf_len(norm_wr) > 0)
1232                         wrbuf_puts(norm_wr, " ");
1233                     wrbuf_puts(norm_wr, name);
1234                     while ((norm_str =
1235                             pp2_charset_token_next(prt)))
1236                     {
1237                         if (*norm_str)
1238                         {
1239                             wrbuf_puts(norm_wr, " ");
1240                             wrbuf_puts(norm_wr, norm_str);
1241                         }
1242                     }
1243                     xmlFree(value);
1244                     pp2_charset_token_destroy(prt);
1245                     no_found++;
1246                 }
1247             }
1248             xmlFree(type);
1249         }
1250     }
1251     return no_found;
1252 }
1253
1254 static const char *get_mergekey(xmlDoc *doc, struct client *cl, int record_no,
1255                                 struct conf_service *service, NMEM nmem)
1256 {
1257     char *mergekey_norm = 0;
1258     xmlNode *root = xmlDocGetRootElement(doc);
1259     WRBUF norm_wr = wrbuf_alloc();
1260
1261     /* consider mergekey from XSL first */
1262     xmlChar *mergekey = xmlGetProp(root, (xmlChar *) "mergekey");
1263     if (mergekey)
1264     {
1265         const char *norm_str;
1266         pp2_charset_token_t prt =
1267             pp2_charset_token_create(service->charsets, "mergekey");
1268
1269         pp2_charset_token_first(prt, (const char *) mergekey, 0);
1270         while ((norm_str = pp2_charset_token_next(prt)))
1271         {
1272             if (*norm_str)
1273             {
1274                 if (wrbuf_len(norm_wr))
1275                     wrbuf_puts(norm_wr, " ");
1276                 wrbuf_puts(norm_wr, norm_str);
1277             }
1278         }
1279         pp2_charset_token_destroy(prt);
1280         xmlFree(mergekey);
1281     }
1282     else
1283     {
1284         /* no mergekey defined in XSL. Look for mergekey metadata instead */
1285         int field_id;
1286         for (field_id = 0; field_id < service->num_metadata; field_id++)
1287         {
1288             struct conf_metadata *ser_md = &service->metadata[field_id];
1289             if (ser_md->mergekey != Metadata_mergekey_no)
1290             {
1291                 int r = get_mergekey_from_doc(doc, root, ser_md->name,
1292                                               service, norm_wr);
1293                 if (r == 0 && ser_md->mergekey == Metadata_mergekey_required)
1294                 {
1295                     /* no mergekey on this one and it is required..
1296                        Generate unique key instead */
1297                     wrbuf_rewind(norm_wr);
1298                     break;
1299                 }
1300             }
1301         }
1302     }
1303
1304     /* generate unique key if none is not generated already or is empty */
1305     if (wrbuf_len(norm_wr) == 0)
1306     {
1307         wrbuf_printf(norm_wr, "%s-%d",
1308                      client_get_id(cl), record_no);
1309     }
1310     if (wrbuf_len(norm_wr) > 0)
1311         mergekey_norm = nmem_strdup(nmem, wrbuf_cstr(norm_wr));
1312     wrbuf_destroy(norm_wr);
1313     return mergekey_norm;
1314 }
1315
1316 /** \brief see if metadata for pz:recordfilter exists
1317     \param root xml root element of normalized record
1318     \param sdb session database for client
1319     \retval 0 if there is no metadata for pz:recordfilter
1320     \retval 1 if there is metadata for pz:recordfilter
1321
1322     If there is no pz:recordfilter defined, this function returns 1
1323     as well.
1324 */
1325
1326 static int check_record_filter(xmlNode *root, struct session_database *sdb)
1327 {
1328     int match = 0;
1329     xmlNode *n;
1330     const char *s;
1331     s = session_setting_oneval(sdb, PZ_RECORDFILTER);
1332
1333     if (!s || !*s)
1334         return 1;
1335
1336     for (n = root->children; n; n = n->next)
1337     {
1338         if (n->type != XML_ELEMENT_NODE)
1339             continue;
1340         if (!strcmp((const char *) n->name, "metadata"))
1341         {
1342             xmlChar *type = xmlGetProp(n, (xmlChar *) "type");
1343             if (type)
1344             {
1345                 size_t len;
1346                 int substring;
1347                 const char *eq;
1348
1349                 if ((eq = strchr(s, '=')))
1350                     substring = 0;
1351                 else if ((eq = strchr(s, '~')))
1352                     substring = 1;
1353                 if (eq)
1354                     len = eq - s;
1355                 else
1356                     len = strlen(s);
1357                 if (len == strlen((const char *)type) &&
1358                     !memcmp((const char *) type, s, len))
1359                 {
1360                     xmlChar *value = xmlNodeGetContent(n);
1361                     if (value && *value)
1362                     {
1363                         if (!eq ||
1364                             (substring && strstr((const char *) value, eq+1)) ||
1365                             (!substring && !strcmp((const char *) value, eq + 1)))
1366                             match = 1;
1367                     }
1368                     xmlFree(value);
1369                 }
1370                 xmlFree(type);
1371             }
1372         }
1373     }
1374     return match;
1375 }
1376
1377
1378 static int ingest_to_cluster(struct client *cl,
1379                              xmlDoc *xdoc,
1380                              xmlNode *root,
1381                              int record_no,
1382                              const char *mergekey_norm);
1383
1384 /** \brief ingest XML record
1385     \param cl client holds the result set for record
1386     \param rec record buffer (0 terminated)
1387     \param record_no record position (1, 2, ..)
1388     \param nmem working NMEM
1389     \retval 0 OK
1390     \retval -1 failure
1391     \retval -2 Filtered
1392 */
1393 int ingest_record(struct client *cl, const char *rec,
1394                   int record_no, NMEM nmem)
1395 {
1396     struct session *se = client_get_session(cl);
1397     int ret = 0;
1398     struct session_database *sdb = client_get_database(cl);
1399     struct conf_service *service = se->service;
1400     xmlDoc *xdoc = normalize_record(se, sdb, service, rec, nmem);
1401     xmlNode *root;
1402     const char *mergekey_norm;
1403
1404     if (!xdoc)
1405         return -1;
1406
1407     root = xmlDocGetRootElement(xdoc);
1408
1409     if (!check_record_filter(root, sdb))
1410     {
1411         session_log(se, YLOG_LOG, "Filtered out record no %d from %s",
1412                     record_no, sdb->database->id);
1413         xmlFreeDoc(xdoc);
1414         return -2;
1415     }
1416
1417     mergekey_norm = get_mergekey(xdoc, cl, record_no, service, nmem);
1418     if (!mergekey_norm)
1419     {
1420         session_log(se, YLOG_WARN, "Got no mergekey");
1421         xmlFreeDoc(xdoc);
1422         return -1;
1423     }
1424     session_enter(se);
1425     if (client_get_session(cl) == se)
1426         ret = ingest_to_cluster(cl, xdoc, root, record_no, mergekey_norm);
1427     session_leave(se);
1428
1429     xmlFreeDoc(xdoc);
1430     return ret;
1431 }
1432
1433 static int ingest_to_cluster(struct client *cl,
1434                              xmlDoc *xdoc,
1435                              xmlNode *root,
1436                              int record_no,
1437                              const char *mergekey_norm)
1438 {
1439     xmlNode *n;
1440     xmlChar *type = 0;
1441     xmlChar *value = 0;
1442     struct session_database *sdb = client_get_database(cl);
1443     struct session *se = client_get_session(cl);
1444     struct conf_service *service = se->service;
1445     struct record *record = record_create(se->nmem,
1446                                           service->num_metadata,
1447                                           service->num_sortkeys, cl,
1448                                           record_no);
1449     struct record_cluster *cluster = reclist_insert(se->reclist,
1450                                                     service,
1451                                                     record,
1452                                                     mergekey_norm,
1453                                                     &se->total_merged);
1454
1455     const char *use_term_factor_str = session_setting_oneval(sdb, PZ_TERMLIST_TERM_FACTOR);
1456     int use_term_factor = 0;
1457     int term_factor = 1;
1458     if (use_term_factor_str && use_term_factor_str[0] != 0)
1459        use_term_factor =  atoi(use_term_factor_str);
1460     if (use_term_factor) {
1461         int maxrecs = client_get_maxrecs(cl);
1462         int hits = (int) client_get_hits(cl);
1463         term_factor = MAX(hits, maxrecs) /  MAX(1, maxrecs);
1464         assert(term_factor >= 1);
1465         yaz_log(YLOG_DEBUG, "Using term factor: %d (%d / %d)", term_factor, MAX(hits, maxrecs), MAX(1, maxrecs));
1466     }
1467
1468     if (!cluster)
1469         return -1;
1470     if (global_parameters.dump_records)
1471         session_log(se, YLOG_LOG, "Cluster id %s from %s (#%d)", cluster->recid,
1472                     sdb->database->id, record_no);
1473     relevance_newrec(se->relevance, cluster);
1474
1475     // now parsing XML record and adding data to cluster or record metadata
1476     for (n = root->children; n; n = n->next)
1477     {
1478         pp2_charset_token_t prt;
1479         if (type)
1480             xmlFree(type);
1481         if (value)
1482             xmlFree(value);
1483         type = value = 0;
1484
1485         if (n->type != XML_ELEMENT_NODE)
1486             continue;
1487         if (!strcmp((const char *) n->name, "metadata"))
1488         {
1489             struct conf_metadata *ser_md = 0;
1490             struct conf_sortkey *ser_sk = 0;
1491             struct record_metadata **wheretoput = 0;
1492             struct record_metadata *rec_md = 0;
1493             int md_field_id = -1;
1494             int sk_field_id = -1;
1495
1496             type = xmlGetProp(n, (xmlChar *) "type");
1497             value = xmlNodeListGetString(xdoc, n->children, 1);
1498
1499             if (!type || !value || !*value)
1500                 continue;
1501
1502             md_field_id
1503                 = conf_service_metadata_field_id(service, (const char *) type);
1504             if (md_field_id < 0)
1505             {
1506                 if (se->number_of_warnings_unknown_metadata == 0)
1507                 {
1508                     session_log(se, YLOG_WARN,
1509                             "Ignoring unknown metadata element: %s", type);
1510                 }
1511                 se->number_of_warnings_unknown_metadata++;
1512                 continue;
1513             }
1514
1515             ser_md = &service->metadata[md_field_id];
1516
1517             if (ser_md->sortkey_offset >= 0){
1518                 sk_field_id = ser_md->sortkey_offset;
1519                 ser_sk = &service->sortkeys[sk_field_id];
1520             }
1521
1522             // non-merged metadata
1523             rec_md = record_metadata_init(se->nmem, (const char *) value,
1524                                           ser_md->type, n->properties);
1525             if (!rec_md)
1526             {
1527                 session_log(se, YLOG_WARN, "bad metadata data '%s' "
1528                             "for element '%s'", value, type);
1529                 continue;
1530             }
1531             wheretoput = &record->metadata[md_field_id];
1532             while (*wheretoput)
1533                 wheretoput = &(*wheretoput)->next;
1534             *wheretoput = rec_md;
1535
1536             // merged metadata
1537             rec_md = record_metadata_init(se->nmem, (const char *) value,
1538                                           ser_md->type, 0);
1539             wheretoput = &cluster->metadata[md_field_id];
1540
1541             // and polulate with data:
1542             // assign cluster or record based on merge action
1543             if (ser_md->merge == Metadata_merge_unique)
1544             {
1545                 while (*wheretoput)
1546                 {
1547                     if (!strcmp((const char *) (*wheretoput)->data.text.disp,
1548                                 rec_md->data.text.disp))
1549                         break;
1550                     wheretoput = &(*wheretoput)->next;
1551                 }
1552                 if (!*wheretoput)
1553                     *wheretoput = rec_md;
1554             }
1555             else if (ser_md->merge == Metadata_merge_longest)
1556             {
1557                 if (!*wheretoput
1558                     || strlen(rec_md->data.text.disp)
1559                     > strlen((*wheretoput)->data.text.disp))
1560                 {
1561                     *wheretoput = rec_md;
1562                     if (ser_sk)
1563                     {
1564                         const char *sort_str = 0;
1565                         int skip_article =
1566                             ser_sk->type == Metadata_sortkey_skiparticle;
1567
1568                         if (!cluster->sortkeys[sk_field_id])
1569                             cluster->sortkeys[sk_field_id] =
1570                                 nmem_malloc(se->nmem,
1571                                             sizeof(union data_types));
1572
1573                         prt =
1574                             pp2_charset_token_create(service->charsets, "sort");
1575
1576                         pp2_charset_token_first(prt, rec_md->data.text.disp,
1577                                                 skip_article);
1578
1579                         pp2_charset_token_next(prt);
1580
1581                         sort_str = pp2_get_sort(prt);
1582
1583                         cluster->sortkeys[sk_field_id]->text.disp =
1584                             rec_md->data.text.disp;
1585                         if (!sort_str)
1586                         {
1587                             sort_str = rec_md->data.text.disp;
1588                             session_log(se, YLOG_WARN,
1589                                     "Could not make sortkey. Bug #1858");
1590                         }
1591                         cluster->sortkeys[sk_field_id]->text.sort =
1592                             nmem_strdup(se->nmem, sort_str);
1593                         pp2_charset_token_destroy(prt);
1594                     }
1595                 }
1596             }
1597             else if (ser_md->merge == Metadata_merge_all)
1598             {
1599                 while (*wheretoput)
1600                     wheretoput = &(*wheretoput)->next;
1601                 *wheretoput = rec_md;
1602             }
1603             else if (ser_md->merge == Metadata_merge_range)
1604             {
1605                 if (!*wheretoput)
1606                 {
1607                     *wheretoput = rec_md;
1608                     if (ser_sk)
1609                         cluster->sortkeys[sk_field_id]
1610                             = &rec_md->data;
1611                 }
1612                 else
1613                 {
1614                     int this_min = rec_md->data.number.min;
1615                     int this_max = rec_md->data.number.max;
1616                     if (this_min < (*wheretoput)->data.number.min)
1617                         (*wheretoput)->data.number.min = this_min;
1618                     if (this_max > (*wheretoput)->data.number.max)
1619                         (*wheretoput)->data.number.max = this_max;
1620                 }
1621             }
1622
1623
1624             // ranking of _all_ fields enabled ...
1625             if (ser_md->rank)
1626                 relevance_countwords(se->relevance, cluster,
1627                                      (char *) value, ser_md->rank,
1628                                      ser_md->name);
1629
1630             // construct facets ... unless the client already has reported them
1631             if (ser_md->termlist && !client_has_facet(cl, (char *) type))
1632             {
1633                 if (ser_md->type == Metadata_type_year)
1634                 {
1635                     char year[64];
1636                     sprintf(year, "%d", rec_md->data.number.max);
1637
1638                     add_facet(se, (char *) type, year, term_factor);
1639                     if (rec_md->data.number.max != rec_md->data.number.min)
1640                     {
1641                         sprintf(year, "%d", rec_md->data.number.min);
1642                         add_facet(se, (char *) type, year, term_factor);
1643                     }
1644                 }
1645                 else
1646                     add_facet(se, (char *) type, (char *) value, term_factor);
1647             }
1648
1649             // cleaning up
1650             xmlFree(type);
1651             xmlFree(value);
1652             type = value = 0;
1653         }
1654         else
1655         {
1656             if (se->number_of_warnings_unknown_elements == 0)
1657                 session_log(se, YLOG_WARN,
1658                         "Unexpected element in internal record: %s", n->name);
1659             se->number_of_warnings_unknown_elements++;
1660         }
1661     }
1662     if (type)
1663         xmlFree(type);
1664     if (value)
1665         xmlFree(value);
1666
1667     relevance_donerecord(se->relevance, cluster);
1668     se->total_records++;
1669
1670     return 0;
1671 }
1672
1673 void session_log(struct session *s, int level, const char *fmt, ...)
1674 {
1675     char buf[1024];
1676     va_list ap;
1677     va_start(ap, fmt);
1678
1679     yaz_vsnprintf(buf, sizeof(buf)-30, fmt, ap);
1680     yaz_log(level, "Session (%u): %s", s->session_id, buf);
1681
1682     va_end(ap);
1683 }
1684
1685 /*
1686  * Local variables:
1687  * c-basic-offset: 4
1688  * c-file-style: "Stroustrup"
1689  * indent-tabs-mode: nil
1690  * End:
1691  * vim: shiftwidth=4 tabstop=8 expandtab
1692  */
1693