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 void search_sort(struct session *se, const char *field, int increasing)
 595 {
 596     struct client_list *l;
 597     struct timeval tval;
 598
 599     session_enter(se);
 600     for (l = se->clients; l; l = l->next)
 601     {
 602         struct client *cl = l->client;
 603         struct session_database *sdb = client_get_database(cl);
 604         struct setting *s;
 605         const char *strategy_plus_sort = 0;
 606
 607         for (s = sdb->settings[PZ_SORTMAP]; s; s = s->next)
 608         {
 609             char *p = strchr(s->name + 3, ':');
 610             if (!p)
 611             {
 612                 yaz_log(YLOG_WARN, "Malformed sortmap name: %s", s->name);
 613                 continue;
 614             }
 615             p++;
 616             if (!strcmp(p, field))
 617             {
 618                 strategy_plus_sort = s->value;
 619                 break;
 620             }
 621         }
 622
 623         if (strategy_plus_sort)
 624         {
 625             if (client_prep_connection(cl, se->service->z3950_operation_timeout,
 626                                        se->service->z3950_session_timeout,
 627                                        se->service->server->iochan_man,
 628                                        &tval))
 629             {
 630                 char **array;
 631                 int num;
 632                 nmem_strsplit(se->nmem, ":", strategy_plus_sort, &array, &num);
 633
 634                 if (num == 2)
 635                 {
 636                     const char *sort_spec = array[1];
 637                     while (*sort_spec == ' ')
 638                         sort_spec++;
 639                     client_start_search(cl, array[0], sort_spec);
 640                 }
 641             }
 642         }
 643     }
 644     session_leave(se);
 645 }
 646
 647 enum pazpar2_error_code search(struct session *se,
 648                                const char *query,
 649                                const char *startrecs, const char *maxrecs,
 650                                const char *filter,
 651                                const char *limit,
 652                                const char **addinfo)
 653 {
 654     int live_channels = 0;
 655     int no_working = 0;
 656     int no_failed = 0;
 657     struct client_list *l;
 658     struct timeval tval;
 659     facet_limits_t facet_limits;
 660
 661     session_log(se, YLOG_DEBUG, "Search");
 662
 663     *addinfo = 0;
 664
 665     session_remove_clients(se);
 666
 667     session_enter(se);
 668     reclist_destroy(se->reclist);
 669     se->reclist = 0;
 670     relevance_destroy(&se->relevance);
 671     nmem_reset(se->nmem);
 672     se->total_records = se->total_merged = 0;
 673     se->num_termlists = 0;
 674     live_channels = select_targets(se, filter);
 675     if (!live_channels)
 676     {
 677         session_leave(se);
 678         return PAZPAR2_NO_TARGETS;
 679     }
 680     se->reclist = reclist_create(se->nmem);
 681
 682     yaz_gettimeofday(&tval);
 683
 684     tval.tv_sec += 5;
 685
 686     facet_limits = facet_limits_create(limit);
 687     if (!facet_limits)
 688     {
 689         *addinfo = "limit";
 690         session_leave(se);
 691         return PAZPAR2_MALFORMED_PARAMETER_VALUE;
 692     }
 693     for (l = se->clients; l; l = l->next)
 694     {
 695         struct client *cl = l->client;
 696
 697         if (maxrecs)
 698             client_set_maxrecs(cl, atoi(maxrecs));
 699         if (startrecs)
 700             client_set_startrecs(cl, atoi(startrecs));
 701         if (prepare_session_database(se, client_get_database(cl)) < 0)
 702             ;
 703         else if (client_parse_query(cl, query, facet_limits) < 0)
 704             no_failed++;
 705         else
 706         {
 707             no_working++;
 708             if (client_prep_connection(cl, se->service->z3950_operation_timeout,
 709                                        se->service->z3950_session_timeout,
 710                                        se->service->server->iochan_man,
 711                                        &tval))
 712                 client_start_search(cl, 0, 0);
 713         }
 714     }
 715     facet_limits_destroy(facet_limits);
 716     session_leave(se);
 717     if (no_working == 0)
 718     {
 719         if (no_failed > 0)
 720         {
 721             *addinfo = "query";
 722             return PAZPAR2_MALFORMED_PARAMETER_VALUE;
 723         }
 724         else
 725             return PAZPAR2_NO_TARGETS;
 726     }
 727     return PAZPAR2_NO_ERROR;
 728 }
 729
 730 // Creates a new session_database object for a database
 731 static void session_init_databases_fun(void *context, struct database *db)
 732 {
 733     struct session *se = (struct session *) context;
 734     struct session_database *new = nmem_malloc(se->session_nmem, sizeof(*new));
 735     int i;
 736
 737     new->database = db;
 738
 739     new->map = 0;
 740     assert(db->settings);
 741     new->settings = nmem_malloc(se->session_nmem,
 742                                 sizeof(struct settings *) * db->num_settings);
 743     new->num_settings = db->num_settings;
 744     for (i = 0; i < db->num_settings; i++)
 745     {
 746         struct setting *setting = db->settings[i];
 747         new->settings[i] = setting;
 748     }
 749     new->next = se->databases;
 750     se->databases = new;
 751 }
 752
 753 // Doesn't free memory associated with sdb -- nmem takes care of that
 754 static void session_database_destroy(struct session_database *sdb)
 755 {
 756     sdb->map = 0;
 757 }
 758
 759 // Initialize session_database list -- this represents this session's view
 760 // of the database list -- subject to modification by the settings ws command
 761 void session_init_databases(struct session *se)
 762 {
 763     se->databases = 0;
 764     predef_grep_databases(se, se->service, session_init_databases_fun);
 765 }
 766
 767 // Probably session_init_databases_fun should be refactored instead of
 768 // called here.
 769 static struct session_database *load_session_database(struct session *se,
 770                                                       char *id)
 771 {
 772     struct database *db = new_database(id, se->session_nmem);
 773
 774     session_init_databases_fun((void*) se, db);
 775
 776     // New sdb is head of se->databases list
 777     return se->databases;
 778 }
 779
 780 // Find an existing session database. If not found, load it
 781 static struct session_database *find_session_database(struct session *se,
 782                                                       char *id)
 783 {
 784     struct session_database *sdb;
 785
 786     for (sdb = se->databases; sdb; sdb = sdb->next)
 787         if (!strcmp(sdb->database->id, id))
 788             return sdb;
 789     return load_session_database(se, id);
 790 }
 791
 792 // Apply a session override to a database
 793 void session_apply_setting(struct session *se, char *dbname, char *setting,
 794                            char *value)
 795 {
 796     struct session_database *sdb = find_session_database(se, dbname);
 797     struct conf_service *service = se->service;
 798     struct setting *new = nmem_malloc(se->session_nmem, sizeof(*new));
 799     int offset = settings_create_offset(service, setting);
 800
 801     expand_settings_array(&sdb->settings, &sdb->num_settings, offset,
 802                           se->session_nmem);
 803     new->precedence = 0;
 804     new->target = dbname;
 805     new->name = setting;
 806     new->value = value;
 807     new->next = sdb->settings[offset];
 808     sdb->settings[offset] = new;
 809
 810     // Force later recompute of settings-driven data structures
 811     // (happens when a search starts and client connections are prepared)
 812     switch (offset)
 813     {
 814     case PZ_XSLT:
 815         if (sdb->map)
 816         {
 817             sdb->map = 0;
 818         }
 819         break;
 820     }
 821 }
 822
 823 void session_destroy(struct session *se) {
 824     struct session_database *sdb;
 825     session_log(se, YLOG_DEBUG, "Destroying");
 826     session_use(-1);
 827     session_remove_clients(se);
 828
 829     for (sdb = se->databases; sdb; sdb = sdb->next)
 830         session_database_destroy(sdb);
 831     normalize_cache_destroy(se->normalize_cache);
 832     relevance_destroy(&se->relevance);
 833     reclist_destroy(se->reclist);
 834     nmem_destroy(se->nmem);
 835     service_destroy(se->service);
 836     yaz_mutex_destroy(&se->session_mutex);
 837 }
 838
 839 /* Depreciated: use session_destroy */
 840 void destroy_session(struct session *se)
 841 {
 842     session_destroy(se);
 843 }
 844
 845 size_t session_get_memory_status(struct session *session) {
 846     size_t session_nmem;
 847     if (session == 0)
 848         return 0;
 849     session_enter(session);
 850     session_nmem = nmem_total(session->nmem);
 851     session_leave(session);
 852     return session_nmem;
 853 }
 854
 855
 856 struct session *new_session(NMEM nmem, struct conf_service *service,
 857                             unsigned session_id)
 858 {
 859     int i;
 860     struct session *session = nmem_malloc(nmem, sizeof(*session));
 861
 862     char tmp_str[50];
 863
 864     sprintf(tmp_str, "session#%u", session_id);
 865
 866     session->session_id = session_id;
 867     session_log(session, YLOG_DEBUG, "New");
 868     session->service = service;
 869     session->relevance = 0;
 870     session->total_records = 0;
 871     session->number_of_warnings_unknown_elements = 0;
 872     session->number_of_warnings_unknown_metadata = 0;
 873     session->num_termlists = 0;
 874     session->reclist = 0;
 875     session->clients = 0;
 876     session->session_nmem = nmem;
 877     session->nmem = nmem_create();
 878     session->databases = 0;
 879     for (i = 0; i <= SESSION_WATCH_MAX; i++)
 880     {
 881         session->watchlist[i].data = 0;
 882         session->watchlist[i].fun = 0;
 883     }
 884     session->normalize_cache = normalize_cache_create();
 885     session->session_mutex = 0;
 886     pazpar2_mutex_create(&session->session_mutex, tmp_str);
 887     session_use(1);
 888     return session;
 889 }
 890
 891 static struct hitsbytarget *hitsbytarget_nb(struct session *se,
 892                                             int *count, NMEM nmem)
 893 {
 894     struct hitsbytarget *res = 0;
 895     struct client_list *l;
 896     size_t sz = 0;
 897
 898     for (l = se->clients; l; l = l->next)
 899         sz++;
 900
 901     res = nmem_malloc(nmem, sizeof(*res) * sz);
 902     *count = 0;
 903     for (l = se->clients; l; l = l->next)
 904     {
 905         struct client *cl = l->client;
 906         WRBUF w = wrbuf_alloc();
 907         const char *name = session_setting_oneval(client_get_database(cl),
 908                                                   PZ_NAME);
 909
 910         res[*count].id = client_get_id(cl);
 911         res[*count].name = *name ? name : "Unknown";
 912         res[*count].hits = client_get_hits(cl);
 913         res[*count].records = client_get_num_records(cl);
 914         res[*count].diagnostic = client_get_diagnostic(cl);
 915         res[*count].state = client_get_state_str(cl);
 916         res[*count].connected  = client_get_connection(cl) ? 1 : 0;
 917         session_settings_dump(se, client_get_database(cl), w);
 918         res[*count].settings_xml = nmem_strdup(nmem, wrbuf_cstr(w));
 919         wrbuf_destroy(w);
 920         (*count)++;
 921     }
 922     return res;
 923 }
 924
 925 struct hitsbytarget *get_hitsbytarget(struct session *se, int *count, NMEM nmem)
 926 {
 927     struct hitsbytarget *p;
 928     session_enter(se);
 929     p = hitsbytarget_nb(se, count, nmem);
 930     session_leave(se);
 931     return p;
 932 }
 933
 934 struct termlist_score **get_termlist_score(struct session *se,
 935                                            const char *name, int *num)
 936 {
 937     int i;
 938     struct termlist_score **tl = 0;
 939
 940     session_enter(se);
 941     for (i = 0; i < se->num_termlists; i++)
 942         if (!strcmp((const char *) se->termlists[i].name, name))
 943         {
 944             tl = termlist_highscore(se->termlists[i].termlist, num);
 945             break;
 946         }
 947     session_leave(se);
 948     return tl;
 949 }
 950
 951 // Compares two hitsbytarget nodes by hitcount
 952 static int cmp_ht(const void *p1, const void *p2)
 953 {
 954     const struct hitsbytarget *h1 = p1;
 955     const struct hitsbytarget *h2 = p2;
 956     return h2->hits - h1->hits;
 957 }
 958
 959 static int targets_termlist_nb(WRBUF wrbuf, struct session *se, int num,
 960                                NMEM nmem)
 961 {
 962     struct hitsbytarget *ht;
 963     int count, i;
 964
 965     ht = hitsbytarget_nb(se, &count, nmem);
 966     qsort(ht, count, sizeof(struct hitsbytarget), cmp_ht);
 967     for (i = 0; i < count && i < num && ht[i].hits > 0; i++)
 968     {
 969
 970         // do only print terms which have display names
 971
 972         wrbuf_puts(wrbuf, "<term>\n");
 973
 974         wrbuf_puts(wrbuf, "<id>");
 975         wrbuf_xmlputs(wrbuf, ht[i].id);
 976         wrbuf_puts(wrbuf, "</id>\n");
 977
 978         wrbuf_puts(wrbuf, "<name>");
 979         if (!ht[i].name || !ht[i].name[0])
 980             wrbuf_xmlputs(wrbuf, "NO TARGET NAME");
 981         else
 982             wrbuf_xmlputs(wrbuf, ht[i].name);
 983         wrbuf_puts(wrbuf, "</name>\n");
 984
 985         wrbuf_printf(wrbuf, "<frequency>" ODR_INT_PRINTF "</frequency>\n",
 986                      ht[i].hits);
 987
 988         wrbuf_puts(wrbuf, "<state>");
 989         wrbuf_xmlputs(wrbuf, ht[i].state);
 990         wrbuf_puts(wrbuf, "</state>\n");
 991
 992         wrbuf_printf(wrbuf, "<diagnostic>%d</diagnostic>\n",
 993                      ht[i].diagnostic);
 994         wrbuf_puts(wrbuf, "</term>\n");
 995     }
 996     return count;
 997 }
 998
 999 void perform_termlist(struct http_channel *c, struct session *se,
1000                       const char *name, int num)
1001 {
1002     int i, j;
1003     NMEM nmem_tmp = nmem_create();
1004     char **names;
1005     int num_names = 0;
1006
1007     if (name)
1008         nmem_strsplit(nmem_tmp, ",", name, &names, &num_names);
1009
1010     session_enter(se);
1011
1012     for (j = 0; j < num_names; j++)
1013     {
1014         const char *tname;
1015         for (i = 0; i < se->num_termlists; i++)
1016         {
1017             tname = se->termlists[i].name;
1018             if (num_names > 0 && !strcmp(names[j], tname))
1019             {
1020                 struct termlist_score **p = 0;
1021                 int len;
1022                 p = termlist_highscore(se->termlists[i].termlist, &len);
1023                 if (p)
1024                 {
1025                     int i;
1026                     wrbuf_puts(c->wrbuf, "<list name=\"");
1027                     wrbuf_xmlputs(c->wrbuf, tname);
1028                     wrbuf_puts(c->wrbuf, "\">\n");
1029                     for (i = 0; i < len && i < num; i++)
1030                     {
1031                         // prevent sending empty term elements
1032                         if (!p[i]->display_term || !p[i]->display_term[0])
1033                             continue;
1034
1035                         wrbuf_puts(c->wrbuf, "<term>");
1036                         wrbuf_puts(c->wrbuf, "<name>");
1037                         wrbuf_xmlputs(c->wrbuf, p[i]->display_term);
1038                         wrbuf_puts(c->wrbuf, "</name>");
1039
1040                         wrbuf_printf(c->wrbuf,
1041                                      "<frequency>%d</frequency>",
1042                                      p[i]->frequency);
1043                         wrbuf_puts(c->wrbuf, "</term>\n");
1044                     }
1045                     wrbuf_puts(c->wrbuf, "</list>\n");
1046                 }
1047             }
1048         }
1049         tname = "xtargets";
1050         if (num_names > 0 && !strcmp(names[j], tname))
1051         {
1052             wrbuf_puts(c->wrbuf, "<list name=\"");
1053             wrbuf_xmlputs(c->wrbuf, tname);
1054             wrbuf_puts(c->wrbuf, "\">\n");
1055             targets_termlist_nb(c->wrbuf, se, num, c->nmem);
1056             wrbuf_puts(c->wrbuf, "</list>\n");
1057         }
1058     }
1059     session_leave(se);
1060     nmem_destroy(nmem_tmp);
1061 }
1062
1063 #ifdef MISSING_HEADERS
1064 void report_nmem_stats(void)
1065 {
1066     size_t in_use, is_free;
1067
1068     nmem_get_memory_in_use(&in_use);
1069     nmem_get_memory_free(&is_free);
1070
1071     yaz_log(YLOG_LOG, "nmem stat: use=%ld free=%ld",
1072             (long) in_use, (long) is_free);
1073 }
1074 #endif
1075
1076 struct record_cluster *show_single_start(struct session *se, const char *id,
1077                                          struct record_cluster **prev_r,
1078                                          struct record_cluster **next_r)
1079 {
1080     struct record_cluster *r = 0;
1081
1082     session_enter(se);
1083     *prev_r = 0;
1084     *next_r = 0;
1085     if (se->reclist)
1086     {
1087         reclist_enter(se->reclist);
1088         while ((r = reclist_read_record(se->reclist)))
1089         {
1090             if (!strcmp(r->recid, id))
1091             {
1092                 *next_r = reclist_read_record(se->reclist);
1093                 break;
1094             }
1095             *prev_r = r;
1096         }
1097         reclist_leave(se->reclist);
1098     }
1099     if (!r)
1100         session_leave(se);
1101     return r;
1102 }
1103
1104 void show_single_stop(struct session *se, struct record_cluster *rec)
1105 {
1106     session_leave(se);
1107 }
1108
1109 struct record_cluster **show_range_start(struct session *se,
1110                                          struct reclist_sortparms *sp,
1111                                          int start, int *num, int *total, Odr_int *sumhits)
1112 {
1113     struct record_cluster **recs;
1114     struct reclist_sortparms *spp;
1115     int i;
1116 #if USE_TIMING
1117     yaz_timing_t t = yaz_timing_create();
1118 #endif
1119     session_enter(se);
1120     recs = nmem_malloc(se->nmem, *num * sizeof(struct record_cluster *));
1121     if (!se->relevance)
1122     {
1123         *num = 0;
1124         *total = 0;
1125         *sumhits = 0;
1126         recs = 0;
1127     }
1128     else
1129     {
1130         struct client_list *l;
1131
1132         for (spp = sp; spp; spp = spp->next)
1133             if (spp->type == Metadata_sortkey_relevance)
1134             {
1135                 relevance_prepare_read(se->relevance, se->reclist);
1136                 break;
1137             }
1138         reclist_sort(se->reclist, sp);
1139
1140         reclist_enter(se->reclist);
1141         *total = reclist_get_num_records(se->reclist);
1142
1143         *sumhits = 0;
1144         for (l = se->clients; l; l = l->next)
1145             *sumhits += client_get_hits(l->client);
1146
1147         for (i = 0; i < start; i++)
1148             if (!reclist_read_record(se->reclist))
1149             {
1150                 *num = 0;
1151                 recs = 0;
1152                 break;
1153             }
1154
1155         for (i = 0; i < *num; i++)
1156         {
1157             struct record_cluster *r = reclist_read_record(se->reclist);
1158             if (!r)
1159             {
1160                 *num = i;
1161                 break;
1162             }
1163             recs[i] = r;
1164         }
1165         reclist_leave(se->reclist);
1166     }
1167 #if USE_TIMING
1168     yaz_timing_stop(t);
1169     yaz_log(YLOG_LOG, "show %6.5f %3.2f %3.2f",
1170             yaz_timing_get_real(t), yaz_timing_get_user(t),
1171             yaz_timing_get_sys(t));
1172     yaz_timing_destroy(&t);
1173 #endif
1174     return recs;
1175 }
1176
1177 void show_range_stop(struct session *se, struct record_cluster **recs)
1178 {
1179     session_leave(se);
1180 }
1181
1182 void statistics(struct session *se, struct statistics *stat)
1183 {
1184     struct client_list *l;
1185     int count = 0;
1186
1187     memset(stat, 0, sizeof(*stat));
1188     stat->num_hits = 0;
1189     for (l = se->clients; l; l = l->next)
1190     {
1191         struct client *cl = l->client;
1192         if (!client_get_connection(cl))
1193             stat->num_no_connection++;
1194         stat->num_hits += client_get_hits(cl);
1195         switch (client_get_state(cl))
1196         {
1197         case Client_Connecting: stat->num_connecting++; break;
1198         case Client_Working: stat->num_working++; break;
1199         case Client_Idle: stat->num_idle++; break;
1200         case Client_Failed: stat->num_failed++; break;
1201         case Client_Error: stat->num_error++; break;
1202         default: break;
1203         }
1204         count++;
1205     }
1206     stat->num_records = se->total_records;
1207
1208     stat->num_clients = count;
1209 }
1210
1211 static struct record_metadata *record_metadata_init(
1212     NMEM nmem, const char *value, enum conf_metadata_type type,
1213     struct _xmlAttr *attr)
1214 {
1215     struct record_metadata *rec_md = record_metadata_create(nmem);
1216     struct record_metadata_attr **attrp = &rec_md->attributes;
1217
1218     for (; attr; attr = attr->next)
1219     {
1220         if (attr->children && attr->children->content)
1221         {
1222             if (strcmp((const char *) attr->name, "type"))
1223             {  /* skip the "type" attribute.. Its value is already part of
1224                   the element in output (md-%s) and so repeating it here
1225                   is redundant */
1226                 *attrp = nmem_malloc(nmem, sizeof(**attrp));
1227                 (*attrp)->name =
1228                     nmem_strdup(nmem, (const char *) attr->name);
1229                 (*attrp)->value =
1230                     nmem_strdup(nmem, (const char *) attr->children->content);
1231                 attrp = &(*attrp)->next;
1232             }
1233         }
1234     }
1235     *attrp = 0;
1236
1237     if (type == Metadata_type_generic)
1238     {
1239         char *p = nmem_strdup(nmem, value);
1240
1241         p = normalize7bit_generic(p, " ,/.:([");
1242
1243         rec_md->data.text.disp = p;
1244         rec_md->data.text.sort = 0;
1245     }
1246     else if (type == Metadata_type_year || type == Metadata_type_date)
1247     {
1248         int first, last;
1249         int longdate = 0;
1250
1251         if (type == Metadata_type_date)
1252             longdate = 1;
1253         if (extract7bit_dates((char *) value, &first, &last, longdate) < 0)
1254             return 0;
1255
1256         rec_md->data.number.min = first;
1257         rec_md->data.number.max = last;
1258     }
1259     else
1260         return 0;
1261     return rec_md;
1262 }
1263
1264 static int get_mergekey_from_doc(xmlDoc *doc, xmlNode *root, const char *name,
1265                                  struct conf_service *service, WRBUF norm_wr)
1266 {
1267     xmlNode *n;
1268     int no_found = 0;
1269     for (n = root->children; n; n = n->next)
1270     {
1271         if (n->type != XML_ELEMENT_NODE)
1272             continue;
1273         if (!strcmp((const char *) n->name, "metadata"))
1274         {
1275             xmlChar *type = xmlGetProp(n, (xmlChar *) "type");
1276             if (type == NULL) {
1277                 yaz_log(YLOG_FATAL, "Missing type attribute on metadata element. Skipping!");
1278             }
1279             else if (!strcmp(name, (const char *) type))
1280             {
1281                 xmlChar *value = xmlNodeListGetString(doc, n->children, 1);
1282                 if (value)
1283                 {
1284                     const char *norm_str;
1285                     pp2_charset_token_t prt =
1286                         pp2_charset_token_create(service->charsets, "mergekey");
1287
1288                     pp2_charset_token_first(prt, (const char *) value, 0);
1289                     if (wrbuf_len(norm_wr) > 0)
1290                         wrbuf_puts(norm_wr, " ");
1291                     wrbuf_puts(norm_wr, name);
1292                     while ((norm_str =
1293                             pp2_charset_token_next(prt)))
1294                     {
1295                         if (*norm_str)
1296                         {
1297                             wrbuf_puts(norm_wr, " ");
1298                             wrbuf_puts(norm_wr, norm_str);
1299                         }
1300                     }
1301                     xmlFree(value);
1302                     pp2_charset_token_destroy(prt);
1303                     no_found++;
1304                 }
1305             }
1306             xmlFree(type);
1307         }
1308     }
1309     return no_found;
1310 }
1311
1312 static const char *get_mergekey(xmlDoc *doc, struct client *cl, int record_no,
1313                                 struct conf_service *service, NMEM nmem)
1314 {
1315     char *mergekey_norm = 0;
1316     xmlNode *root = xmlDocGetRootElement(doc);
1317     WRBUF norm_wr = wrbuf_alloc();
1318
1319     /* consider mergekey from XSL first */
1320     xmlChar *mergekey = xmlGetProp(root, (xmlChar *) "mergekey");
1321     if (mergekey)
1322     {
1323         const char *norm_str;
1324         pp2_charset_token_t prt =
1325             pp2_charset_token_create(service->charsets, "mergekey");
1326
1327         pp2_charset_token_first(prt, (const char *) mergekey, 0);
1328         while ((norm_str = pp2_charset_token_next(prt)))
1329         {
1330             if (*norm_str)
1331             {
1332                 if (wrbuf_len(norm_wr))
1333                     wrbuf_puts(norm_wr, " ");
1334                 wrbuf_puts(norm_wr, norm_str);
1335             }
1336         }
1337         pp2_charset_token_destroy(prt);
1338         xmlFree(mergekey);
1339     }
1340     else
1341     {
1342         /* no mergekey defined in XSL. Look for mergekey metadata instead */
1343         int field_id;
1344         for (field_id = 0; field_id < service->num_metadata; field_id++)
1345         {
1346             struct conf_metadata *ser_md = &service->metadata[field_id];
1347             if (ser_md->mergekey != Metadata_mergekey_no)
1348             {
1349                 int r = get_mergekey_from_doc(doc, root, ser_md->name,
1350                                               service, norm_wr);
1351                 if (r == 0 && ser_md->mergekey == Metadata_mergekey_required)
1352                 {
1353                     /* no mergekey on this one and it is required..
1354                        Generate unique key instead */
1355                     wrbuf_rewind(norm_wr);
1356                     break;
1357                 }
1358             }
1359         }
1360     }
1361
1362     /* generate unique key if none is not generated already or is empty */
1363     if (wrbuf_len(norm_wr) == 0)
1364     {
1365         wrbuf_printf(norm_wr, "%s-%d",
1366                      client_get_id(cl), record_no);
1367     }
1368     if (wrbuf_len(norm_wr) > 0)
1369         mergekey_norm = nmem_strdup(nmem, wrbuf_cstr(norm_wr));
1370     wrbuf_destroy(norm_wr);
1371     return mergekey_norm;
1372 }
1373
1374 /** \brief see if metadata for pz:recordfilter exists
1375     \param root xml root element of normalized record
1376     \param sdb session database for client
1377     \retval 0 if there is no metadata for pz:recordfilter
1378     \retval 1 if there is metadata for pz:recordfilter
1379
1380     If there is no pz:recordfilter defined, this function returns 1
1381     as well.
1382 */
1383
1384 static int check_record_filter(xmlNode *root, struct session_database *sdb)
1385 {
1386     int match = 0;
1387     xmlNode *n;
1388     const char *s;
1389     s = session_setting_oneval(sdb, PZ_RECORDFILTER);
1390
1391     if (!s || !*s)
1392         return 1;
1393
1394     for (n = root->children; n; n = n->next)
1395     {
1396         if (n->type != XML_ELEMENT_NODE)
1397             continue;
1398         if (!strcmp((const char *) n->name, "metadata"))
1399         {
1400             xmlChar *type = xmlGetProp(n, (xmlChar *) "type");
1401             if (type)
1402             {
1403                 size_t len;
1404                 int substring;
1405                 const char *eq;
1406
1407                 if ((eq = strchr(s, '=')))
1408                     substring = 0;
1409                 else if ((eq = strchr(s, '~')))
1410                     substring = 1;
1411                 if (eq)
1412                     len = eq - s;
1413                 else
1414                     len = strlen(s);
1415                 if (len == strlen((const char *)type) &&
1416                     !memcmp((const char *) type, s, len))
1417                 {
1418                     xmlChar *value = xmlNodeGetContent(n);
1419                     if (value && *value)
1420                     {
1421                         if (!eq ||
1422                             (substring && strstr((const char *) value, eq+1)) ||
1423                             (!substring && !strcmp((const char *) value, eq + 1)))
1424                             match = 1;
1425                     }
1426                     xmlFree(value);
1427                 }
1428                 xmlFree(type);
1429             }
1430         }
1431     }
1432     return match;
1433 }
1434
1435
1436 static int ingest_to_cluster(struct client *cl,
1437                              xmlDoc *xdoc,
1438                              xmlNode *root,
1439                              int record_no,
1440                              const char *mergekey_norm);
1441
1442 /** \brief ingest XML record
1443     \param cl client holds the result set for record
1444     \param rec record buffer (0 terminated)
1445     \param record_no record position (1, 2, ..)
1446     \param nmem working NMEM
1447     \retval 0 OK
1448     \retval -1 failure
1449     \retval -2 Filtered
1450 */
1451 int ingest_record(struct client *cl, const char *rec,
1452                   int record_no, NMEM nmem)
1453 {
1454     struct session *se = client_get_session(cl);
1455     int ret = 0;
1456     struct session_database *sdb = client_get_database(cl);
1457     struct conf_service *service = se->service;
1458     xmlDoc *xdoc = normalize_record(se, sdb, service, rec, nmem);
1459     xmlNode *root;
1460     const char *mergekey_norm;
1461
1462     if (!xdoc)
1463         return -1;
1464
1465     root = xmlDocGetRootElement(xdoc);
1466
1467     if (!check_record_filter(root, sdb))
1468     {
1469         session_log(se, YLOG_LOG, "Filtered out record no %d from %s",
1470                     record_no, sdb->database->id);
1471         xmlFreeDoc(xdoc);
1472         return -2;
1473     }
1474
1475     mergekey_norm = get_mergekey(xdoc, cl, record_no, service, nmem);
1476     if (!mergekey_norm)
1477     {
1478         session_log(se, YLOG_WARN, "Got no mergekey");
1479         xmlFreeDoc(xdoc);
1480         return -1;
1481     }
1482     session_enter(se);
1483     if (client_get_session(cl) == se)
1484         ret = ingest_to_cluster(cl, xdoc, root, record_no, mergekey_norm);
1485     session_leave(se);
1486
1487     xmlFreeDoc(xdoc);
1488     return ret;
1489 }
1490
1491 static int ingest_to_cluster(struct client *cl,
1492                              xmlDoc *xdoc,
1493                              xmlNode *root,
1494                              int record_no,
1495                              const char *mergekey_norm)
1496 {
1497     xmlNode *n;
1498     xmlChar *type = 0;
1499     xmlChar *value = 0;
1500     struct session_database *sdb = client_get_database(cl);
1501     struct session *se = client_get_session(cl);
1502     struct conf_service *service = se->service;
1503     struct record *record = record_create(se->nmem,
1504                                           service->num_metadata,
1505                                           service->num_sortkeys, cl,
1506                                           record_no);
1507     struct record_cluster *cluster = reclist_insert(se->reclist,
1508                                                     service,
1509                                                     record,
1510                                                     mergekey_norm,
1511                                                     &se->total_merged);
1512
1513     const char *use_term_factor_str = session_setting_oneval(sdb, PZ_TERMLIST_TERM_FACTOR);
1514     int use_term_factor = 0;
1515     int term_factor = 1;
1516     if (use_term_factor_str && use_term_factor_str[0] != 0)
1517        use_term_factor =  atoi(use_term_factor_str);
1518     if (use_term_factor) {
1519         int maxrecs = client_get_maxrecs(cl);
1520         int hits = (int) client_get_hits(cl);
1521         term_factor = MAX(hits, maxrecs) /  MAX(1, maxrecs);
1522         assert(term_factor >= 1);
1523         yaz_log(YLOG_DEBUG, "Using term factor: %d (%d / %d)", term_factor, MAX(hits, maxrecs), MAX(1, maxrecs));
1524     }
1525
1526     if (!cluster)
1527         return -1;
1528     if (global_parameters.dump_records)
1529         session_log(se, YLOG_LOG, "Cluster id %s from %s (#%d)", cluster->recid,
1530                     sdb->database->id, record_no);
1531     relevance_newrec(se->relevance, cluster);
1532
1533     // now parsing XML record and adding data to cluster or record metadata
1534     for (n = root->children; n; n = n->next)
1535     {
1536         pp2_charset_token_t prt;
1537         if (type)
1538             xmlFree(type);
1539         if (value)
1540             xmlFree(value);
1541         type = value = 0;
1542
1543         if (n->type != XML_ELEMENT_NODE)
1544             continue;
1545         if (!strcmp((const char *) n->name, "metadata"))
1546         {
1547             struct conf_metadata *ser_md = 0;
1548             struct conf_sortkey *ser_sk = 0;
1549             struct record_metadata **wheretoput = 0;
1550             struct record_metadata *rec_md = 0;
1551             int md_field_id = -1;
1552             int sk_field_id = -1;
1553
1554             type = xmlGetProp(n, (xmlChar *) "type");
1555             value = xmlNodeListGetString(xdoc, n->children, 1);
1556
1557             if (!type || !value || !*value)
1558                 continue;
1559
1560             md_field_id
1561                 = conf_service_metadata_field_id(service, (const char *) type);
1562             if (md_field_id < 0)
1563             {
1564                 if (se->number_of_warnings_unknown_metadata == 0)
1565                 {
1566                     session_log(se, YLOG_WARN,
1567                             "Ignoring unknown metadata element: %s", type);
1568                 }
1569                 se->number_of_warnings_unknown_metadata++;
1570                 continue;
1571             }
1572
1573             ser_md = &service->metadata[md_field_id];
1574
1575             if (ser_md->sortkey_offset >= 0){
1576                 sk_field_id = ser_md->sortkey_offset;
1577                 ser_sk = &service->sortkeys[sk_field_id];
1578             }
1579
1580             // non-merged metadata
1581             rec_md = record_metadata_init(se->nmem, (const char *) value,
1582                                           ser_md->type, n->properties);
1583             if (!rec_md)
1584             {
1585                 session_log(se, YLOG_WARN, "bad metadata data '%s' "
1586                             "for element '%s'", value, type);
1587                 continue;
1588             }
1589             wheretoput = &record->metadata[md_field_id];
1590             while (*wheretoput)
1591                 wheretoput = &(*wheretoput)->next;
1592             *wheretoput = rec_md;
1593
1594             // merged metadata
1595             rec_md = record_metadata_init(se->nmem, (const char *) value,
1596                                           ser_md->type, 0);
1597             wheretoput = &cluster->metadata[md_field_id];
1598
1599             // and polulate with data:
1600             // assign cluster or record based on merge action
1601             if (ser_md->merge == Metadata_merge_unique)
1602             {
1603                 while (*wheretoput)
1604                 {
1605                     if (!strcmp((const char *) (*wheretoput)->data.text.disp,
1606                                 rec_md->data.text.disp))
1607                         break;
1608                     wheretoput = &(*wheretoput)->next;
1609                 }
1610                 if (!*wheretoput)
1611                     *wheretoput = rec_md;
1612             }
1613             else if (ser_md->merge == Metadata_merge_longest)
1614             {
1615                 if (!*wheretoput
1616                     || strlen(rec_md->data.text.disp)
1617                     > strlen((*wheretoput)->data.text.disp))
1618                 {
1619                     *wheretoput = rec_md;
1620                     if (ser_sk)
1621                     {
1622                         const char *sort_str = 0;
1623                         int skip_article =
1624                             ser_sk->type == Metadata_sortkey_skiparticle;
1625
1626                         if (!cluster->sortkeys[sk_field_id])
1627                             cluster->sortkeys[sk_field_id] =
1628                                 nmem_malloc(se->nmem,
1629                                             sizeof(union data_types));
1630
1631                         prt =
1632                             pp2_charset_token_create(service->charsets, "sort");
1633
1634                         pp2_charset_token_first(prt, rec_md->data.text.disp,
1635                                                 skip_article);
1636
1637                         pp2_charset_token_next(prt);
1638
1639                         sort_str = pp2_get_sort(prt);
1640
1641                         cluster->sortkeys[sk_field_id]->text.disp =
1642                             rec_md->data.text.disp;
1643                         if (!sort_str)
1644                         {
1645                             sort_str = rec_md->data.text.disp;
1646                             session_log(se, YLOG_WARN,
1647                                     "Could not make sortkey. Bug #1858");
1648                         }
1649                         cluster->sortkeys[sk_field_id]->text.sort =
1650                             nmem_strdup(se->nmem, sort_str);
1651                         pp2_charset_token_destroy(prt);
1652                     }
1653                 }
1654             }
1655             else if (ser_md->merge == Metadata_merge_all)
1656             {
1657                 while (*wheretoput)
1658                     wheretoput = &(*wheretoput)->next;
1659                 *wheretoput = rec_md;
1660             }
1661             else if (ser_md->merge == Metadata_merge_range)
1662             {
1663                 if (!*wheretoput)
1664                 {
1665                     *wheretoput = rec_md;
1666                     if (ser_sk)
1667                         cluster->sortkeys[sk_field_id]
1668                             = &rec_md->data;
1669                 }
1670                 else
1671                 {
1672                     int this_min = rec_md->data.number.min;
1673                     int this_max = rec_md->data.number.max;
1674                     if (this_min < (*wheretoput)->data.number.min)
1675                         (*wheretoput)->data.number.min = this_min;
1676                     if (this_max > (*wheretoput)->data.number.max)
1677                         (*wheretoput)->data.number.max = this_max;
1678                 }
1679             }
1680
1681
1682             // ranking of _all_ fields enabled ...
1683             if (ser_md->rank)
1684                 relevance_countwords(se->relevance, cluster,
1685                                      (char *) value, ser_md->rank,
1686                                      ser_md->name);
1687
1688             // construct facets ... unless the client already has reported them
1689             if (ser_md->termlist && !client_has_facet(cl, (char *) type))
1690             {
1691                 if (ser_md->type == Metadata_type_year)
1692                 {
1693                     char year[64];
1694                     sprintf(year, "%d", rec_md->data.number.max);
1695
1696                     add_facet(se, (char *) type, year, term_factor);
1697                     if (rec_md->data.number.max != rec_md->data.number.min)
1698                     {
1699                         sprintf(year, "%d", rec_md->data.number.min);
1700                         add_facet(se, (char *) type, year, term_factor);
1701                     }
1702                 }
1703                 else
1704                     add_facet(se, (char *) type, (char *) value, term_factor);
1705             }
1706
1707             // cleaning up
1708             xmlFree(type);
1709             xmlFree(value);
1710             type = value = 0;
1711         }
1712         else
1713         {
1714             if (se->number_of_warnings_unknown_elements == 0)
1715                 session_log(se, YLOG_WARN,
1716                         "Unexpected element in internal record: %s", n->name);
1717             se->number_of_warnings_unknown_elements++;
1718         }
1719     }
1720     if (type)
1721         xmlFree(type);
1722     if (value)
1723         xmlFree(value);
1724
1725     relevance_donerecord(se->relevance, cluster);
1726     se->total_records++;
1727
1728     return 0;
1729 }
1730
1731 void session_log(struct session *s, int level, const char *fmt, ...)
1732 {
1733     char buf[1024];
1734     va_list ap;
1735     va_start(ap, fmt);
1736
1737     yaz_vsnprintf(buf, sizeof(buf)-30, fmt, ap);
1738     yaz_log(level, "Session (%u): %s", s->session_id, buf);
1739
1740     va_end(ap);
1741 }
1742
1743 /*
1744  * Local variables:
1745  * c-basic-offset: 4
1746  * c-file-style: "Stroustrup"
1747  * indent-tabs-mode: nil
1748  * End:
1749  * vim: shiftwidth=4 tabstop=8 expandtab
1750  */
1751