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();
 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     resolve_database(se->service, db->database, url);
 535
 536     client_set_database(cl, db);
 537
 538     client_set_session(cl, se);
 539
 540     l = xmalloc(sizeof(*l));
 541     l->client = cl;
 542     l->next = se->clients;
 543     se->clients = l;
 544 }
 545
 546 static void session_remove_clients(struct session *se)
 547 {
 548     struct client_list *l;
 549
 550     session_enter(se);
 551     l = se->clients;
 552     se->clients = 0;
 553     session_leave(se);
 554
 555     while (l)
 556     {
 557         struct client_list *l_next = l->next;
 558         client_lock(l->client);
 559         client_set_session(l->client, 0);
 560         client_set_database(l->client, 0);
 561         client_unlock(l->client);
 562         client_destroy(l->client);
 563         xfree(l);
 564         l = l_next;
 565     }
 566 }
 567
 568 // Associates a set of clients with a session;
 569 // Note: Session-databases represent databases with per-session
 570 // setting overrides
 571 static int select_targets(struct session *se, const char *filter)
 572 {
 573     return session_grep_databases(se, filter, select_targets_callback);
 574 }
 575
 576 int session_active_clients(struct session *s)
 577 {
 578     struct client_list *l;
 579     int res = 0;
 580
 581     for (l = s->clients; l; l = l->next)
 582         if (client_is_active(l->client))
 583             res++;
 584
 585     return res;
 586 }
 587
 588 int session_is_preferred_clients_ready(struct session *s)
 589 {
 590     struct client_list *l;
 591     int res = 0;
 592
 593     for (l = s->clients; l; l = l->next)
 594         if (client_is_active_preferred(l->client))
 595             res++;
 596     session_log(s, YLOG_DEBUG, "Has %d active preferred clients.", res);
 597     return res == 0;
 598 }
 599
 600 enum pazpar2_error_code search(struct session *se,
 601                                const char *query,
 602                                const char *startrecs, const char *maxrecs,
 603                                const char *filter,
 604                                const char *limit,
 605                                const char **addinfo)
 606 {
 607     int live_channels = 0;
 608     int no_working = 0;
 609     int no_failed = 0;
 610     struct client_list *l;
 611     struct timeval tval;
 612     facet_limits_t facet_limits;
 613
 614     session_log(se, YLOG_DEBUG, "Search");
 615
 616     *addinfo = 0;
 617
 618     session_remove_clients(se);
 619
 620     session_enter(se);
 621     reclist_destroy(se->reclist);
 622     se->reclist = 0;
 623     relevance_destroy(&se->relevance);
 624     nmem_reset(se->nmem);
 625     se->total_records = se->total_hits = se->total_merged = 0;
 626     se->num_termlists = 0;
 627     live_channels = select_targets(se, filter);
 628     if (!live_channels)
 629     {
 630         session_leave(se);
 631         return PAZPAR2_NO_TARGETS;
 632     }
 633     se->reclist = reclist_create(se->nmem);
 634
 635     yaz_gettimeofday(&tval);
 636
 637     tval.tv_sec += 5;
 638
 639     facet_limits = facet_limits_create(limit);
 640     if (!facet_limits)
 641     {
 642         *addinfo = "limit";
 643         session_leave(se);
 644         return PAZPAR2_MALFORMED_PARAMETER_VALUE;
 645     }
 646     for (l = se->clients; l; l = l->next)
 647     {
 648         struct client *cl = l->client;
 649
 650         if (maxrecs)
 651             client_set_maxrecs(cl, atoi(maxrecs));
 652         if (startrecs)
 653             client_set_startrecs(cl, atoi(startrecs));
 654         if (prepare_session_database(se, client_get_database(cl)) < 0)
 655             ;
 656         else if (client_parse_query(cl, query, facet_limits) < 0)
 657             no_failed++;
 658         else
 659         {
 660             no_working++;
 661             if (client_prep_connection(cl, se->service->z3950_operation_timeout,
 662                                        se->service->z3950_session_timeout,
 663                                        se->service->server->iochan_man,
 664                                        &tval))
 665                 client_start_search(cl);
 666         }
 667     }
 668     facet_limits_destroy(facet_limits);
 669     session_leave(se);
 670     if (no_working == 0)
 671     {
 672         if (no_failed > 0)
 673         {
 674             *addinfo = "query";
 675             return PAZPAR2_MALFORMED_PARAMETER_VALUE;
 676         }
 677         else
 678             return PAZPAR2_NO_TARGETS;
 679     }
 680     return PAZPAR2_NO_ERROR;
 681 }
 682
 683 // Creates a new session_database object for a database
 684 static void session_init_databases_fun(void *context, struct database *db)
 685 {
 686     struct session *se = (struct session *) context;
 687     struct session_database *new = nmem_malloc(se->session_nmem, sizeof(*new));
 688     int i;
 689
 690     new->database = db;
 691
 692     new->map = 0;
 693     assert(db->settings);
 694     new->settings = nmem_malloc(se->session_nmem,
 695                                 sizeof(struct settings *) * db->num_settings);
 696     new->num_settings = db->num_settings;
 697     for (i = 0; i < db->num_settings; i++)
 698     {
 699         struct setting *setting = db->settings[i];
 700         new->settings[i] = setting;
 701     }
 702     new->next = se->databases;
 703     se->databases = new;
 704 }
 705
 706 // Doesn't free memory associated with sdb -- nmem takes care of that
 707 static void session_database_destroy(struct session_database *sdb)
 708 {
 709     sdb->map = 0;
 710 }
 711
 712 // Initialize session_database list -- this represents this session's view
 713 // of the database list -- subject to modification by the settings ws command
 714 void session_init_databases(struct session *se)
 715 {
 716     se->databases = 0;
 717     predef_grep_databases(se, se->service, session_init_databases_fun);
 718 }
 719
 720 // Probably session_init_databases_fun should be refactored instead of
 721 // called here.
 722 static struct session_database *load_session_database(struct session *se,
 723                                                       char *id)
 724 {
 725     struct database *db = new_database(id, se->session_nmem);
 726
 727     session_init_databases_fun((void*) se, db);
 728
 729     // New sdb is head of se->databases list
 730     return se->databases;
 731 }
 732
 733 // Find an existing session database. If not found, load it
 734 static struct session_database *find_session_database(struct session *se,
 735                                                       char *id)
 736 {
 737     struct session_database *sdb;
 738
 739     for (sdb = se->databases; sdb; sdb = sdb->next)
 740         if (!strcmp(sdb->database->url, id))
 741             return sdb;
 742     return load_session_database(se, id);
 743 }
 744
 745 // Apply a session override to a database
 746 void session_apply_setting(struct session *se, char *dbname, char *setting,
 747                            char *value)
 748 {
 749     struct session_database *sdb = find_session_database(se, dbname);
 750     struct conf_service *service = se->service;
 751     struct setting *new = nmem_malloc(se->session_nmem, sizeof(*new));
 752     int offset = settings_create_offset(service, setting);
 753
 754     expand_settings_array(&sdb->settings, &sdb->num_settings, offset,
 755                           se->session_nmem);
 756     new->precedence = 0;
 757     new->target = dbname;
 758     new->name = setting;
 759     new->value = value;
 760     new->next = sdb->settings[offset];
 761     sdb->settings[offset] = new;
 762
 763     // Force later recompute of settings-driven data structures
 764     // (happens when a search starts and client connections are prepared)
 765     switch (offset)
 766     {
 767     case PZ_XSLT:
 768         if (sdb->map)
 769         {
 770             sdb->map = 0;
 771         }
 772         break;
 773     }
 774 }
 775
 776 void session_destroy(struct session *se) {
 777     struct session_database *sdb;
 778     session_log(se, YLOG_DEBUG, "Destroying");
 779     session_use(-1);
 780     session_remove_clients(se);
 781
 782     for (sdb = se->databases; sdb; sdb = sdb->next)
 783         session_database_destroy(sdb);
 784     normalize_cache_destroy(se->normalize_cache);
 785     relevance_destroy(&se->relevance);
 786     reclist_destroy(se->reclist);
 787     nmem_destroy(se->nmem);
 788     service_destroy(se->service);
 789     yaz_mutex_destroy(&se->session_mutex);
 790 }
 791
 792 /* Depreciated: use session_destroy */
 793 void destroy_session(struct session *se)
 794 {
 795     session_destroy(se);
 796 }
 797
 798 size_t session_get_memory_status(struct session *session) {
 799     size_t session_nmem;
 800     if (session == 0)
 801         return 0;
 802     session_enter(session);
 803     session_nmem = nmem_total(session->nmem);
 804     session_leave(session);
 805     return session_nmem;
 806 }
 807
 808
 809 struct session *new_session(NMEM nmem, struct conf_service *service,
 810                             unsigned session_id)
 811 {
 812     int i;
 813     struct session *session = nmem_malloc(nmem, sizeof(*session));
 814
 815     char tmp_str[50];
 816
 817     sprintf(tmp_str, "session#%u", session_id);
 818
 819     session->session_id = session_id;
 820     session_log(session, YLOG_DEBUG, "New");
 821     session->service = service;
 822     session->relevance = 0;
 823     session->total_hits = 0;
 824     session->total_records = 0;
 825     session->number_of_warnings_unknown_elements = 0;
 826     session->number_of_warnings_unknown_metadata = 0;
 827     session->num_termlists = 0;
 828     session->reclist = 0;
 829     session->clients = 0;
 830     session->session_nmem = nmem;
 831     session->nmem = nmem_create();
 832     session->databases = 0;
 833     for (i = 0; i <= SESSION_WATCH_MAX; i++)
 834     {
 835         session->watchlist[i].data = 0;
 836         session->watchlist[i].fun = 0;
 837     }
 838     session->normalize_cache = normalize_cache_create();
 839     session->session_mutex = 0;
 840     pazpar2_mutex_create(&session->session_mutex, tmp_str);
 841     session_use(1);
 842     return session;
 843 }
 844
 845 static struct hitsbytarget *hitsbytarget_nb(struct session *se,
 846                                             int *count, NMEM nmem)
 847 {
 848     struct hitsbytarget *res = 0;
 849     struct client_list *l;
 850     size_t sz = 0;
 851
 852     for (l = se->clients; l; l = l->next)
 853         sz++;
 854
 855     res = nmem_malloc(nmem, sizeof(*res) * sz);
 856     *count = 0;
 857     for (l = se->clients; l; l = l->next)
 858     {
 859         struct client *cl = l->client;
 860         WRBUF w = wrbuf_alloc();
 861         const char *name = session_setting_oneval(client_get_database(cl),
 862                                                   PZ_NAME);
 863
 864         res[*count].id = client_get_database(cl)->database->url;
 865         res[*count].name = *name ? name : "Unknown";
 866         res[*count].hits = client_get_hits(cl);
 867         res[*count].records = client_get_num_records(cl);
 868         res[*count].diagnostic = client_get_diagnostic(cl);
 869         res[*count].state = client_get_state_str(cl);
 870         res[*count].connected  = client_get_connection(cl) ? 1 : 0;
 871         session_settings_dump(se, client_get_database(cl), w);
 872         res[*count].settings_xml = nmem_strdup(nmem, wrbuf_cstr(w));
 873         wrbuf_destroy(w);
 874         (*count)++;
 875     }
 876     return res;
 877 }
 878
 879 struct hitsbytarget *get_hitsbytarget(struct session *se, int *count, NMEM nmem)
 880 {
 881     struct hitsbytarget *p;
 882     session_enter(se);
 883     p = hitsbytarget_nb(se, count, nmem);
 884     session_leave(se);
 885     return p;
 886 }
 887
 888 struct termlist_score **get_termlist_score(struct session *se,
 889                                            const char *name, int *num)
 890 {
 891     int i;
 892     struct termlist_score **tl = 0;
 893
 894     session_enter(se);
 895     for (i = 0; i < se->num_termlists; i++)
 896         if (!strcmp((const char *) se->termlists[i].name, name))
 897         {
 898             tl = termlist_highscore(se->termlists[i].termlist, num);
 899             break;
 900         }
 901     session_leave(se);
 902     return tl;
 903 }
 904
 905 // Compares two hitsbytarget nodes by hitcount
 906 static int cmp_ht(const void *p1, const void *p2)
 907 {
 908     const struct hitsbytarget *h1 = p1;
 909     const struct hitsbytarget *h2 = p2;
 910     return h2->hits - h1->hits;
 911 }
 912
 913 static int targets_termlist_nb(WRBUF wrbuf, struct session *se, int num,
 914                                NMEM nmem)
 915 {
 916     struct hitsbytarget *ht;
 917     int count, i;
 918
 919     ht = hitsbytarget_nb(se, &count, nmem);
 920     qsort(ht, count, sizeof(struct hitsbytarget), cmp_ht);
 921     for (i = 0; i < count && i < num && ht[i].hits > 0; i++)
 922     {
 923
 924         // do only print terms which have display names
 925
 926         wrbuf_puts(wrbuf, "<term>\n");
 927
 928         wrbuf_puts(wrbuf, "<id>");
 929         wrbuf_xmlputs(wrbuf, ht[i].id);
 930         wrbuf_puts(wrbuf, "</id>\n");
 931
 932         wrbuf_puts(wrbuf, "<name>");
 933         if (!ht[i].name || !ht[i].name[0])
 934             wrbuf_xmlputs(wrbuf, "NO TARGET NAME");
 935         else
 936             wrbuf_xmlputs(wrbuf, ht[i].name);
 937         wrbuf_puts(wrbuf, "</name>\n");
 938
 939         wrbuf_printf(wrbuf, "<frequency>" ODR_INT_PRINTF "</frequency>\n",
 940                      ht[i].hits);
 941
 942         wrbuf_puts(wrbuf, "<state>");
 943         wrbuf_xmlputs(wrbuf, ht[i].state);
 944         wrbuf_puts(wrbuf, "</state>\n");
 945
 946         wrbuf_printf(wrbuf, "<diagnostic>%d</diagnostic>\n",
 947                      ht[i].diagnostic);
 948         wrbuf_puts(wrbuf, "</term>\n");
 949     }
 950     return count;
 951 }
 952
 953 void perform_termlist(struct http_channel *c, struct session *se,
 954                       const char *name, int num)
 955 {
 956     int i, j;
 957     NMEM nmem_tmp = nmem_create();
 958     char **names;
 959     int num_names = 0;
 960
 961     if (name)
 962         nmem_strsplit(nmem_tmp, ",", name, &names, &num_names);
 963
 964     session_enter(se);
 965
 966     for (j = 0; j < num_names; j++)
 967     {
 968         const char *tname;
 969         for (i = 0; i < se->num_termlists; i++)
 970         {
 971             tname = se->termlists[i].name;
 972             if (num_names > 0 && !strcmp(names[j], tname))
 973             {
 974                 struct termlist_score **p = 0;
 975                 int len;
 976                 p = termlist_highscore(se->termlists[i].termlist, &len);
 977                 if (p)
 978                 {
 979                     int i;
 980                     wrbuf_puts(c->wrbuf, "<list name=\"");
 981                     wrbuf_xmlputs(c->wrbuf, tname);
 982                     wrbuf_puts(c->wrbuf, "\">\n");
 983                     for (i = 0; i < len && i < num; i++)
 984                     {
 985                         // prevent sending empty term elements
 986                         if (!p[i]->display_term || !p[i]->display_term[0])
 987                             continue;
 988
 989                         wrbuf_puts(c->wrbuf, "<term>");
 990                         wrbuf_puts(c->wrbuf, "<name>");
 991                         wrbuf_xmlputs(c->wrbuf, p[i]->display_term);
 992                         wrbuf_puts(c->wrbuf, "</name>");
 993
 994                         wrbuf_printf(c->wrbuf,
 995                                      "<frequency>%d</frequency>",
 996                                      p[i]->frequency);
 997                         wrbuf_puts(c->wrbuf, "</term>\n");
 998                     }
 999                     wrbuf_puts(c->wrbuf, "</list>\n");
1000                 }
1001             }
1002         }
1003         tname = "xtargets";
1004         if (num_names > 0 && !strcmp(names[j], tname))
1005         {
1006             wrbuf_puts(c->wrbuf, "<list name=\"");
1007             wrbuf_xmlputs(c->wrbuf, tname);
1008             wrbuf_puts(c->wrbuf, "\">\n");
1009             targets_termlist_nb(c->wrbuf, se, num, c->nmem);
1010             wrbuf_puts(c->wrbuf, "</list>\n");
1011         }
1012     }
1013     session_leave(se);
1014     nmem_destroy(nmem_tmp);
1015 }
1016
1017 #ifdef MISSING_HEADERS
1018 void report_nmem_stats(void)
1019 {
1020     size_t in_use, is_free;
1021
1022     nmem_get_memory_in_use(&in_use);
1023     nmem_get_memory_free(&is_free);
1024
1025     yaz_log(YLOG_LOG, "nmem stat: use=%ld free=%ld",
1026             (long) in_use, (long) is_free);
1027 }
1028 #endif
1029
1030 struct record_cluster *show_single_start(struct session *se, const char *id,
1031                                          struct record_cluster **prev_r,
1032                                          struct record_cluster **next_r)
1033 {
1034     struct record_cluster *r = 0;
1035
1036     session_enter(se);
1037     *prev_r = 0;
1038     *next_r = 0;
1039     if (se->reclist)
1040     {
1041         reclist_enter(se->reclist);
1042         while ((r = reclist_read_record(se->reclist)))
1043         {
1044             if (!strcmp(r->recid, id))
1045             {
1046                 *next_r = reclist_read_record(se->reclist);
1047                 break;
1048             }
1049             *prev_r = r;
1050         }
1051         reclist_leave(se->reclist);
1052     }
1053     if (!r)
1054         session_leave(se);
1055     return r;
1056 }
1057
1058 void show_single_stop(struct session *se, struct record_cluster *rec)
1059 {
1060     session_leave(se);
1061 }
1062
1063 struct record_cluster **show_range_start(struct session *se,
1064                                          struct reclist_sortparms *sp,
1065                                          int start, int *num, int *total, Odr_int *sumhits)
1066 {
1067     struct record_cluster **recs;
1068     struct reclist_sortparms *spp;
1069     int i;
1070 #if USE_TIMING
1071     yaz_timing_t t = yaz_timing_create();
1072 #endif
1073     session_enter(se);
1074     recs = nmem_malloc(se->nmem, *num * sizeof(struct record_cluster *));
1075     if (!se->relevance)
1076     {
1077         *num = 0;
1078         *total = 0;
1079         *sumhits = 0;
1080         recs = 0;
1081     }
1082     else
1083     {
1084         for (spp = sp; spp; spp = spp->next)
1085             if (spp->type == Metadata_sortkey_relevance)
1086             {
1087                 relevance_prepare_read(se->relevance, se->reclist);
1088                 break;
1089             }
1090         reclist_sort(se->reclist, sp);
1091
1092         reclist_enter(se->reclist);
1093         *total = reclist_get_num_records(se->reclist);
1094         *sumhits = se->total_hits;
1095
1096         for (i = 0; i < start; i++)
1097             if (!reclist_read_record(se->reclist))
1098             {
1099                 *num = 0;
1100                 recs = 0;
1101                 break;
1102             }
1103
1104         for (i = 0; i < *num; i++)
1105         {
1106             struct record_cluster *r = reclist_read_record(se->reclist);
1107             if (!r)
1108             {
1109                 *num = i;
1110                 break;
1111             }
1112             recs[i] = r;
1113         }
1114         reclist_leave(se->reclist);
1115     }
1116 #if USE_TIMING
1117     yaz_timing_stop(t);
1118     yaz_log(YLOG_LOG, "show %6.5f %3.2f %3.2f",
1119             yaz_timing_get_real(t), yaz_timing_get_user(t),
1120             yaz_timing_get_sys(t));
1121     yaz_timing_destroy(&t);
1122 #endif
1123     return recs;
1124 }
1125
1126 void show_range_stop(struct session *se, struct record_cluster **recs)
1127 {
1128     session_leave(se);
1129 }
1130
1131 void statistics(struct session *se, struct statistics *stat)
1132 {
1133     struct client_list *l;
1134     int count = 0;
1135
1136     memset(stat, 0, sizeof(*stat));
1137     for (l = se->clients; l; l = l->next)
1138     {
1139         struct client *cl = l->client;
1140         if (!client_get_connection(cl))
1141             stat->num_no_connection++;
1142         switch (client_get_state(cl))
1143         {
1144         case Client_Connecting: stat->num_connecting++; break;
1145         case Client_Working: stat->num_working++; break;
1146         case Client_Idle: stat->num_idle++; break;
1147         case Client_Failed: stat->num_failed++; break;
1148         case Client_Error: stat->num_error++; break;
1149         default: break;
1150         }
1151         count++;
1152     }
1153     stat->num_hits = se->total_hits;
1154     stat->num_records = se->total_records;
1155
1156     stat->num_clients = count;
1157 }
1158
1159 static struct record_metadata *record_metadata_init(
1160     NMEM nmem, const char *value, enum conf_metadata_type type,
1161     struct _xmlAttr *attr)
1162 {
1163     struct record_metadata *rec_md = record_metadata_create(nmem);
1164     struct record_metadata_attr **attrp = &rec_md->attributes;
1165
1166     for (; attr; attr = attr->next)
1167     {
1168         if (attr->children && attr->children->content)
1169         {
1170             if (strcmp((const char *) attr->name, "type"))
1171             {  /* skip the "type" attribute.. Its value is already part of
1172                   the element in output (md-%s) and so repeating it here
1173                   is redundant */
1174                 *attrp = nmem_malloc(nmem, sizeof(**attrp));
1175                 (*attrp)->name =
1176                     nmem_strdup(nmem, (const char *) attr->name);
1177                 (*attrp)->value =
1178                     nmem_strdup(nmem, (const char *) attr->children->content);
1179                 attrp = &(*attrp)->next;
1180             }
1181         }
1182     }
1183     *attrp = 0;
1184
1185     if (type == Metadata_type_generic)
1186     {
1187         char *p = nmem_strdup(nmem, value);
1188
1189         p = normalize7bit_generic(p, " ,/.:([");
1190
1191         rec_md->data.text.disp = p;
1192         rec_md->data.text.sort = 0;
1193     }
1194     else if (type == Metadata_type_year || type == Metadata_type_date)
1195     {
1196         int first, last;
1197         int longdate = 0;
1198
1199         if (type == Metadata_type_date)
1200             longdate = 1;
1201         if (extract7bit_dates((char *) value, &first, &last, longdate) < 0)
1202             return 0;
1203
1204         rec_md->data.number.min = first;
1205         rec_md->data.number.max = last;
1206     }
1207     else
1208         return 0;
1209     return rec_md;
1210 }
1211
1212 static int get_mergekey_from_doc(xmlDoc *doc, xmlNode *root, const char *name,
1213                                  struct conf_service *service, WRBUF norm_wr)
1214 {
1215     xmlNode *n;
1216     int no_found = 0;
1217     for (n = root->children; n; n = n->next)
1218     {
1219         if (n->type != XML_ELEMENT_NODE)
1220             continue;
1221         if (!strcmp((const char *) n->name, "metadata"))
1222         {
1223             xmlChar *type = xmlGetProp(n, (xmlChar *) "type");
1224             if (type == NULL) {
1225                 yaz_log(YLOG_FATAL, "Missing type attribute on metadata element. Skipping!");
1226             }
1227             else if (!strcmp(name, (const char *) type))
1228             {
1229                 xmlChar *value = xmlNodeListGetString(doc, n->children, 1);
1230                 if (value)
1231                 {
1232                     const char *norm_str;
1233                     pp2_charset_token_t prt =
1234                         pp2_charset_token_create(service->charsets, "mergekey");
1235
1236                     pp2_charset_token_first(prt, (const char *) value, 0);
1237                     if (wrbuf_len(norm_wr) > 0)
1238                         wrbuf_puts(norm_wr, " ");
1239                     wrbuf_puts(norm_wr, name);
1240                     while ((norm_str =
1241                             pp2_charset_token_next(prt)))
1242                     {
1243                         if (*norm_str)
1244                         {
1245                             wrbuf_puts(norm_wr, " ");
1246                             wrbuf_puts(norm_wr, norm_str);
1247                         }
1248                     }
1249                     xmlFree(value);
1250                     pp2_charset_token_destroy(prt);
1251                     no_found++;
1252                 }
1253             }
1254             xmlFree(type);
1255         }
1256     }
1257     return no_found;
1258 }
1259
1260 static const char *get_mergekey(xmlDoc *doc, struct client *cl, int record_no,
1261                                 struct conf_service *service, NMEM nmem)
1262 {
1263     char *mergekey_norm = 0;
1264     xmlNode *root = xmlDocGetRootElement(doc);
1265     WRBUF norm_wr = wrbuf_alloc();
1266
1267     /* consider mergekey from XSL first */
1268     xmlChar *mergekey = xmlGetProp(root, (xmlChar *) "mergekey");
1269     if (mergekey)
1270     {
1271         const char *norm_str;
1272         pp2_charset_token_t prt =
1273             pp2_charset_token_create(service->charsets, "mergekey");
1274
1275         pp2_charset_token_first(prt, (const char *) mergekey, 0);
1276         while ((norm_str = pp2_charset_token_next(prt)))
1277         {
1278             if (*norm_str)
1279             {
1280                 if (wrbuf_len(norm_wr))
1281                     wrbuf_puts(norm_wr, " ");
1282                 wrbuf_puts(norm_wr, norm_str);
1283             }
1284         }
1285         pp2_charset_token_destroy(prt);
1286         xmlFree(mergekey);
1287     }
1288     else
1289     {
1290         /* no mergekey defined in XSL. Look for mergekey metadata instead */
1291         int field_id;
1292         for (field_id = 0; field_id < service->num_metadata; field_id++)
1293         {
1294             struct conf_metadata *ser_md = &service->metadata[field_id];
1295             if (ser_md->mergekey != Metadata_mergekey_no)
1296             {
1297                 int r = get_mergekey_from_doc(doc, root, ser_md->name,
1298                                               service, norm_wr);
1299                 if (r == 0 && ser_md->mergekey == Metadata_mergekey_required)
1300                 {
1301                     /* no mergekey on this one and it is required..
1302                        Generate unique key instead */
1303                     wrbuf_rewind(norm_wr);
1304                     break;
1305                 }
1306             }
1307         }
1308     }
1309
1310     /* generate unique key if none is not generated already or is empty */
1311     if (wrbuf_len(norm_wr) == 0)
1312     {
1313         wrbuf_printf(norm_wr, "%s-%d",
1314                      client_get_database(cl)->database->url, record_no);
1315     }
1316     if (wrbuf_len(norm_wr) > 0)
1317         mergekey_norm = nmem_strdup(nmem, wrbuf_cstr(norm_wr));
1318     wrbuf_destroy(norm_wr);
1319     return mergekey_norm;
1320 }
1321
1322 /** \brief see if metadata for pz:recordfilter exists
1323     \param root xml root element of normalized record
1324     \param sdb session database for client
1325     \retval 0 if there is no metadata for pz:recordfilter
1326     \retval 1 if there is metadata for pz:recordfilter
1327
1328     If there is no pz:recordfilter defined, this function returns 1
1329     as well.
1330 */
1331
1332 static int check_record_filter(xmlNode *root, struct session_database *sdb)
1333 {
1334     int match = 0;
1335     xmlNode *n;
1336     const char *s;
1337     s = session_setting_oneval(sdb, PZ_RECORDFILTER);
1338
1339     if (!s || !*s)
1340         return 1;
1341
1342     for (n = root->children; n; n = n->next)
1343     {
1344         if (n->type != XML_ELEMENT_NODE)
1345             continue;
1346         if (!strcmp((const char *) n->name, "metadata"))
1347         {
1348             xmlChar *type = xmlGetProp(n, (xmlChar *) "type");
1349             if (type)
1350             {
1351                 size_t len;
1352                 int substring;
1353                 const char *eq;
1354
1355                 if ((eq = strchr(s, '=')))
1356                     substring = 0;
1357                 else if ((eq = strchr(s, '~')))
1358                     substring = 1;
1359                 if (eq)
1360                     len = eq - s;
1361                 else
1362                     len = strlen(s);
1363                 if (len == strlen((const char *)type) &&
1364                     !memcmp((const char *) type, s, len))
1365                 {
1366                     xmlChar *value = xmlNodeGetContent(n);
1367                     if (value && *value)
1368                     {
1369                         if (!eq ||
1370                             (substring && strstr((const char *) value, eq+1)) ||
1371                             (!substring && !strcmp((const char *) value, eq + 1)))
1372                             match = 1;
1373                     }
1374                     xmlFree(value);
1375                 }
1376                 xmlFree(type);
1377             }
1378         }
1379     }
1380     return match;
1381 }
1382
1383
1384 static int ingest_to_cluster(struct client *cl,
1385                              xmlDoc *xdoc,
1386                              xmlNode *root,
1387                              int record_no,
1388                              const char *mergekey_norm);
1389
1390 /** \brief ingest XML record
1391     \param cl client holds the result set for record
1392     \param rec record buffer (0 terminated)
1393     \param record_no record position (1, 2, ..)
1394     \param nmem working NMEM
1395     \retval 0 OK
1396     \retval -1 failure
1397     \retval -2 Filtered
1398 */
1399 int ingest_record(struct client *cl, const char *rec,
1400                   int record_no, NMEM nmem)
1401 {
1402     struct session *se = client_get_session(cl);
1403     int ret = 0;
1404     struct session_database *sdb = client_get_database(cl);
1405     struct conf_service *service = se->service;
1406     xmlDoc *xdoc = normalize_record(se, sdb, service, rec, nmem);
1407     xmlNode *root;
1408     const char *mergekey_norm;
1409
1410     if (!xdoc)
1411         return -1;
1412
1413     root = xmlDocGetRootElement(xdoc);
1414
1415     if (!check_record_filter(root, sdb))
1416     {
1417         session_log(se, YLOG_LOG, "Filtered out record no %d from %s",
1418                     record_no, sdb->database->url);
1419         xmlFreeDoc(xdoc);
1420         return -2;
1421     }
1422
1423     mergekey_norm = get_mergekey(xdoc, cl, record_no, service, nmem);
1424     if (!mergekey_norm)
1425     {
1426         session_log(se, YLOG_WARN, "Got no mergekey");
1427         xmlFreeDoc(xdoc);
1428         return -1;
1429     }
1430     session_enter(se);
1431     if (client_get_session(cl) == se)
1432         ret = ingest_to_cluster(cl, xdoc, root, record_no, mergekey_norm);
1433     session_leave(se);
1434
1435     xmlFreeDoc(xdoc);
1436     return ret;
1437 }
1438
1439 static int ingest_to_cluster(struct client *cl,
1440                              xmlDoc *xdoc,
1441                              xmlNode *root,
1442                              int record_no,
1443                              const char *mergekey_norm)
1444 {
1445     xmlNode *n;
1446     xmlChar *type = 0;
1447     xmlChar *value = 0;
1448     struct session_database *sdb = client_get_database(cl);
1449     struct session *se = client_get_session(cl);
1450     struct conf_service *service = se->service;
1451     struct record *record = record_create(se->nmem,
1452                                           service->num_metadata,
1453                                           service->num_sortkeys, cl,
1454                                           record_no);
1455     struct record_cluster *cluster = reclist_insert(se->reclist,
1456                                                     service,
1457                                                     record,
1458                                                     mergekey_norm,
1459                                                     &se->total_merged);
1460
1461     const char *use_term_factor_str = session_setting_oneval(sdb, PZ_TERMLIST_TERM_FACTOR);
1462     int use_term_factor = 0;
1463     int term_factor = 1;
1464     if (use_term_factor_str && use_term_factor_str[0] != 0)
1465        use_term_factor =  atoi(use_term_factor_str);
1466     if (use_term_factor) {
1467         int maxrecs = client_get_maxrecs(cl);
1468         int hits = (int) client_get_hits(cl);
1469         term_factor = MAX(hits, maxrecs) /  MAX(1, maxrecs);
1470         assert(term_factor >= 1);
1471         yaz_log(YLOG_DEBUG, "Using term factor: %d (%d / %d)", term_factor, MAX(hits, maxrecs), MAX(1, maxrecs));
1472     }
1473
1474     if (!cluster)
1475         return -1;
1476     if (global_parameters.dump_records)
1477         session_log(se, YLOG_LOG, "Cluster id %s from %s (#%d)", cluster->recid,
1478                     sdb->database->url, record_no);
1479     relevance_newrec(se->relevance, cluster);
1480
1481     // now parsing XML record and adding data to cluster or record metadata
1482     for (n = root->children; n; n = n->next)
1483     {
1484         pp2_charset_token_t prt;
1485         if (type)
1486             xmlFree(type);
1487         if (value)
1488             xmlFree(value);
1489         type = value = 0;
1490
1491         if (n->type != XML_ELEMENT_NODE)
1492             continue;
1493         if (!strcmp((const char *) n->name, "metadata"))
1494         {
1495             struct conf_metadata *ser_md = 0;
1496             struct conf_sortkey *ser_sk = 0;
1497             struct record_metadata **wheretoput = 0;
1498             struct record_metadata *rec_md = 0;
1499             int md_field_id = -1;
1500             int sk_field_id = -1;
1501
1502             type = xmlGetProp(n, (xmlChar *) "type");
1503             value = xmlNodeListGetString(xdoc, n->children, 1);
1504
1505             if (!type || !value || !*value)
1506                 continue;
1507
1508             md_field_id
1509                 = conf_service_metadata_field_id(service, (const char *) type);
1510             if (md_field_id < 0)
1511             {
1512                 if (se->number_of_warnings_unknown_metadata == 0)
1513                 {
1514                     session_log(se, YLOG_WARN,
1515                             "Ignoring unknown metadata element: %s", type);
1516                 }
1517                 se->number_of_warnings_unknown_metadata++;
1518                 continue;
1519             }
1520
1521             ser_md = &service->metadata[md_field_id];
1522
1523             if (ser_md->sortkey_offset >= 0){
1524                 sk_field_id = ser_md->sortkey_offset;
1525                 ser_sk = &service->sortkeys[sk_field_id];
1526             }
1527
1528             // non-merged metadata
1529             rec_md = record_metadata_init(se->nmem, (const char *) value,
1530                                           ser_md->type, n->properties);
1531             if (!rec_md)
1532             {
1533                 session_log(se, YLOG_WARN, "bad metadata data '%s' "
1534                             "for element '%s'", value, type);
1535                 continue;
1536             }
1537             wheretoput = &record->metadata[md_field_id];
1538             while (*wheretoput)
1539                 wheretoput = &(*wheretoput)->next;
1540             *wheretoput = rec_md;
1541
1542             // merged metadata
1543             rec_md = record_metadata_init(se->nmem, (const char *) value,
1544                                           ser_md->type, 0);
1545             wheretoput = &cluster->metadata[md_field_id];
1546
1547             // and polulate with data:
1548             // assign cluster or record based on merge action
1549             if (ser_md->merge == Metadata_merge_unique)
1550             {
1551                 while (*wheretoput)
1552                 {
1553                     if (!strcmp((const char *) (*wheretoput)->data.text.disp,
1554                                 rec_md->data.text.disp))
1555                         break;
1556                     wheretoput = &(*wheretoput)->next;
1557                 }
1558                 if (!*wheretoput)
1559                     *wheretoput = rec_md;
1560             }
1561             else if (ser_md->merge == Metadata_merge_longest)
1562             {
1563                 if (!*wheretoput
1564                     || strlen(rec_md->data.text.disp)
1565                     > strlen((*wheretoput)->data.text.disp))
1566                 {
1567                     *wheretoput = rec_md;
1568                     if (ser_sk)
1569                     {
1570                         const char *sort_str = 0;
1571                         int skip_article =
1572                             ser_sk->type == Metadata_sortkey_skiparticle;
1573
1574                         if (!cluster->sortkeys[sk_field_id])
1575                             cluster->sortkeys[sk_field_id] =
1576                                 nmem_malloc(se->nmem,
1577                                             sizeof(union data_types));
1578
1579                         prt =
1580                             pp2_charset_token_create(service->charsets, "sort");
1581
1582                         pp2_charset_token_first(prt, rec_md->data.text.disp,
1583                                                 skip_article);
1584
1585                         pp2_charset_token_next(prt);
1586
1587                         sort_str = pp2_get_sort(prt);
1588
1589                         cluster->sortkeys[sk_field_id]->text.disp =
1590                             rec_md->data.text.disp;
1591                         if (!sort_str)
1592                         {
1593                             sort_str = rec_md->data.text.disp;
1594                             session_log(se, YLOG_WARN,
1595                                     "Could not make sortkey. Bug #1858");
1596                         }
1597                         cluster->sortkeys[sk_field_id]->text.sort =
1598                             nmem_strdup(se->nmem, sort_str);
1599                         pp2_charset_token_destroy(prt);
1600                     }
1601                 }
1602             }
1603             else if (ser_md->merge == Metadata_merge_all)
1604             {
1605                 while (*wheretoput)
1606                     wheretoput = &(*wheretoput)->next;
1607                 *wheretoput = rec_md;
1608             }
1609             else if (ser_md->merge == Metadata_merge_range)
1610             {
1611                 if (!*wheretoput)
1612                 {
1613                     *wheretoput = rec_md;
1614                     if (ser_sk)
1615                         cluster->sortkeys[sk_field_id]
1616                             = &rec_md->data;
1617                 }
1618                 else
1619                 {
1620                     int this_min = rec_md->data.number.min;
1621                     int this_max = rec_md->data.number.max;
1622                     if (this_min < (*wheretoput)->data.number.min)
1623                         (*wheretoput)->data.number.min = this_min;
1624                     if (this_max > (*wheretoput)->data.number.max)
1625                         (*wheretoput)->data.number.max = this_max;
1626                 }
1627             }
1628
1629
1630             // ranking of _all_ fields enabled ...
1631             if (ser_md->rank)
1632                 relevance_countwords(se->relevance, cluster,
1633                                      (char *) value, ser_md->rank,
1634                                      ser_md->name);
1635
1636             // construct facets ... unless the client already has reported them
1637             if (ser_md->termlist && !client_has_facet(cl, (char *) type))
1638             {
1639                 if (ser_md->type == Metadata_type_year)
1640                 {
1641                     char year[64];
1642                     sprintf(year, "%d", rec_md->data.number.max);
1643
1644                     add_facet(se, (char *) type, year, term_factor);
1645                     if (rec_md->data.number.max != rec_md->data.number.min)
1646                     {
1647                         sprintf(year, "%d", rec_md->data.number.min);
1648                         add_facet(se, (char *) type, year, term_factor);
1649                     }
1650                 }
1651                 else
1652                     add_facet(se, (char *) type, (char *) value, term_factor);
1653             }
1654
1655             // cleaning up
1656             xmlFree(type);
1657             xmlFree(value);
1658             type = value = 0;
1659         }
1660         else
1661         {
1662             if (se->number_of_warnings_unknown_elements == 0)
1663                 session_log(se, YLOG_WARN,
1664                         "Unexpected element in internal record: %s", n->name);
1665             se->number_of_warnings_unknown_elements++;
1666         }
1667     }
1668     if (type)
1669         xmlFree(type);
1670     if (value)
1671         xmlFree(value);
1672
1673     relevance_donerecord(se->relevance, cluster);
1674     se->total_records++;
1675
1676     return 0;
1677 }
1678
1679 void session_log(struct session *s, int level, const char *fmt, ...)
1680 {
1681     char buf[1024];
1682     va_list ap;
1683     va_start(ap, fmt);
1684
1685     yaz_vsnprintf(buf, sizeof(buf)-30, fmt, ap);
1686     yaz_log(level, "Session (%u): %s", s->session_id, buf);
1687
1688     va_end(ap);
1689 }
1690
1691 /*
1692  * Local variables:
1693  * c-basic-offset: 4
1694  * c-file-style: "Stroustrup"
1695  * indent-tabs-mode: nil
1696  * End:
1697  * vim: shiftwidth=4 tabstop=8 expandtab
1698  */
1699