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, 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 session_sort(struct session *se, const char *field, int increasing)
 595 {
 596     struct session_sorted_results *sr;
 597     struct client_list *l;
 598
 599     session_enter(se);
 600
 601     /* see if we already have sorted for this critieria */
 602     for (sr = se->sorted_results; sr; sr = sr->next)
 603     {
 604         if (!strcmp(field, sr->field) && increasing == sr->increasing)
 605             break;
 606     }
 607     if (sr)
 608     {
 609         yaz_log(YLOG_LOG, "search_sort: field=%s increasing=%d already fetched",
 610                 field, increasing);
 611         session_leave(se);
 612         return;
 613     }
 614     yaz_log(YLOG_LOG, "search_sort: field=%s increasing=%d must fetch",
 615             field, increasing);
 616     sr = nmem_malloc(se->nmem, sizeof(*sr));
 617     sr->field = nmem_strdup(se->nmem, field);
 618     sr->increasing = increasing;
 619     sr->next = se->sorted_results;
 620     se->sorted_results = sr;
 621
 622     for (l = se->clients; l; l = l->next)
 623     {
 624         struct client *cl = l->client;
 625         struct timeval tval;
 626         if (client_prep_connection(cl, se->service->z3950_operation_timeout,
 627                                    se->service->z3950_session_timeout,
 628                                    se->service->server->iochan_man,
 629                                    &tval))
 630             client_start_search(cl);
 631     }
 632     session_leave(se);
 633 }
 634
 635 enum pazpar2_error_code session_search(struct session *se,
 636                                        const char *query,
 637                                        const char *startrecs,
 638                                        const char *maxrecs,
 639                                        const char *filter,
 640                                        const char *limit,
 641                                        const char **addinfo,
 642                                        const char *sort_field, int increasing)
 643 {
 644     int live_channels = 0;
 645     int no_working = 0;
 646     int no_failed = 0;
 647     struct client_list *l;
 648     struct timeval tval;
 649     facet_limits_t facet_limits;
 650
 651     session_log(se, YLOG_DEBUG, "Search");
 652
 653     *addinfo = 0;
 654
 655     session_remove_clients(se);
 656
 657     session_enter(se);
 658     reclist_destroy(se->reclist);
 659     se->reclist = 0;
 660     relevance_destroy(&se->relevance);
 661     nmem_reset(se->nmem);
 662     se->total_records = se->total_merged = 0;
 663     se->num_termlists = 0;
 664
 665     /* reset list of sorted results and clear to relevance search */
 666     se->sorted_results = nmem_malloc(se->nmem, sizeof(*se->sorted_results));
 667     se->sorted_results->field = nmem_strdup(se->nmem, sort_field);
 668     se->sorted_results->increasing = increasing;
 669     se->sorted_results->next = 0;
 670
 671     live_channels = select_targets(se, filter);
 672     if (!live_channels)
 673     {
 674         session_leave(se);
 675         return PAZPAR2_NO_TARGETS;
 676     }
 677     se->reclist = reclist_create(se->nmem);
 678
 679     yaz_gettimeofday(&tval);
 680
 681     tval.tv_sec += 5;
 682
 683     facet_limits = facet_limits_create(limit);
 684     if (!facet_limits)
 685     {
 686         *addinfo = "limit";
 687         session_leave(se);
 688         return PAZPAR2_MALFORMED_PARAMETER_VALUE;
 689     }
 690     for (l = se->clients; l; l = l->next)
 691     {
 692         struct client *cl = l->client;
 693
 694         if (maxrecs)
 695             client_set_maxrecs(cl, atoi(maxrecs));
 696         if (startrecs)
 697             client_set_startrecs(cl, atoi(startrecs));
 698         if (prepare_session_database(se, client_get_database(cl)) < 0)
 699             ;
 700         else if (client_parse_query(cl, query, facet_limits) < 0)
 701             no_failed++;
 702         else
 703         {
 704             no_working++;
 705             if (client_prep_connection(cl, se->service->z3950_operation_timeout,
 706                                        se->service->z3950_session_timeout,
 707                                        se->service->server->iochan_man,
 708                                        &tval))
 709                 client_start_search(cl);
 710         }
 711     }
 712     facet_limits_destroy(facet_limits);
 713     session_leave(se);
 714     if (no_working == 0)
 715     {
 716         if (no_failed > 0)
 717         {
 718             *addinfo = "query";
 719             return PAZPAR2_MALFORMED_PARAMETER_VALUE;
 720         }
 721         else
 722             return PAZPAR2_NO_TARGETS;
 723     }
 724     return PAZPAR2_NO_ERROR;
 725 }
 726
 727 // Creates a new session_database object for a database
 728 static void session_init_databases_fun(void *context, struct database *db)
 729 {
 730     struct session *se = (struct session *) context;
 731     struct session_database *new = nmem_malloc(se->session_nmem, sizeof(*new));
 732     int i;
 733
 734     new->database = db;
 735
 736     new->map = 0;
 737     assert(db->settings);
 738     new->settings = nmem_malloc(se->session_nmem,
 739                                 sizeof(struct settings *) * db->num_settings);
 740     new->num_settings = db->num_settings;
 741     for (i = 0; i < db->num_settings; i++)
 742     {
 743         struct setting *setting = db->settings[i];
 744         new->settings[i] = setting;
 745     }
 746     new->next = se->databases;
 747     se->databases = new;
 748 }
 749
 750 // Doesn't free memory associated with sdb -- nmem takes care of that
 751 static void session_database_destroy(struct session_database *sdb)
 752 {
 753     sdb->map = 0;
 754 }
 755
 756 // Initialize session_database list -- this represents this session's view
 757 // of the database list -- subject to modification by the settings ws command
 758 void session_init_databases(struct session *se)
 759 {
 760     se->databases = 0;
 761     predef_grep_databases(se, se->service, session_init_databases_fun);
 762 }
 763
 764 // Probably session_init_databases_fun should be refactored instead of
 765 // called here.
 766 static struct session_database *load_session_database(struct session *se,
 767                                                       char *id)
 768 {
 769     struct database *db = new_database(id, se->session_nmem);
 770
 771     session_init_databases_fun((void*) se, db);
 772
 773     // New sdb is head of se->databases list
 774     return se->databases;
 775 }
 776
 777 // Find an existing session database. If not found, load it
 778 static struct session_database *find_session_database(struct session *se,
 779                                                       char *id)
 780 {
 781     struct session_database *sdb;
 782
 783     for (sdb = se->databases; sdb; sdb = sdb->next)
 784         if (!strcmp(sdb->database->id, id))
 785             return sdb;
 786     return load_session_database(se, id);
 787 }
 788
 789 // Apply a session override to a database
 790 void session_apply_setting(struct session *se, char *dbname, char *setting,
 791                            char *value)
 792 {
 793     struct session_database *sdb = find_session_database(se, dbname);
 794     struct conf_service *service = se->service;
 795     struct setting *new = nmem_malloc(se->session_nmem, sizeof(*new));
 796     int offset = settings_create_offset(service, setting);
 797
 798     expand_settings_array(&sdb->settings, &sdb->num_settings, offset,
 799                           se->session_nmem);
 800     new->precedence = 0;
 801     new->target = dbname;
 802     new->name = setting;
 803     new->value = value;
 804     new->next = sdb->settings[offset];
 805     sdb->settings[offset] = new;
 806
 807     // Force later recompute of settings-driven data structures
 808     // (happens when a search starts and client connections are prepared)
 809     switch (offset)
 810     {
 811     case PZ_XSLT:
 812         if (sdb->map)
 813         {
 814             sdb->map = 0;
 815         }
 816         break;
 817     }
 818 }
 819
 820 void session_destroy(struct session *se) {
 821     struct session_database *sdb;
 822     session_log(se, YLOG_DEBUG, "Destroying");
 823     session_use(-1);
 824     session_remove_clients(se);
 825
 826     for (sdb = se->databases; sdb; sdb = sdb->next)
 827         session_database_destroy(sdb);
 828     normalize_cache_destroy(se->normalize_cache);
 829     relevance_destroy(&se->relevance);
 830     reclist_destroy(se->reclist);
 831     nmem_destroy(se->nmem);
 832     service_destroy(se->service);
 833     yaz_mutex_destroy(&se->session_mutex);
 834 }
 835
 836 size_t session_get_memory_status(struct session *session) {
 837     size_t session_nmem;
 838     if (session == 0)
 839         return 0;
 840     session_enter(session);
 841     session_nmem = nmem_total(session->nmem);
 842     session_leave(session);
 843     return session_nmem;
 844 }
 845
 846
 847 struct session *new_session(NMEM nmem, struct conf_service *service,
 848                             unsigned session_id)
 849 {
 850     int i;
 851     struct session *session = nmem_malloc(nmem, sizeof(*session));
 852
 853     char tmp_str[50];
 854
 855     sprintf(tmp_str, "session#%u", session_id);
 856
 857     session->session_id = session_id;
 858     session_log(session, YLOG_DEBUG, "New");
 859     session->service = service;
 860     session->relevance = 0;
 861     session->total_records = 0;
 862     session->number_of_warnings_unknown_elements = 0;
 863     session->number_of_warnings_unknown_metadata = 0;
 864     session->num_termlists = 0;
 865     session->reclist = 0;
 866     session->clients = 0;
 867     session->session_nmem = nmem;
 868     session->nmem = nmem_create();
 869     session->databases = 0;
 870     for (i = 0; i <= SESSION_WATCH_MAX; i++)
 871     {
 872         session->watchlist[i].data = 0;
 873         session->watchlist[i].fun = 0;
 874     }
 875     session->normalize_cache = normalize_cache_create();
 876     session->session_mutex = 0;
 877     pazpar2_mutex_create(&session->session_mutex, tmp_str);
 878     session_use(1);
 879     return session;
 880 }
 881
 882 const char * client_get_suggestions_xml(struct client *cl, WRBUF wrbuf);
 883
 884 static struct hitsbytarget *hitsbytarget_nb(struct session *se,
 885                                             int *count, NMEM nmem)
 886 {
 887     struct hitsbytarget *res = 0;
 888     struct client_list *l;
 889     size_t sz = 0;
 890
 891     for (l = se->clients; l; l = l->next)
 892         sz++;
 893
 894     res = nmem_malloc(nmem, sizeof(*res) * sz);
 895     *count = 0;
 896     for (l = se->clients; l; l = l->next)
 897     {
 898         struct client *cl = l->client;
 899         WRBUF w = wrbuf_alloc();
 900         const char *name = session_setting_oneval(client_get_database(cl),
 901                                                   PZ_NAME);
 902
 903         res[*count].id = client_get_id(cl);
 904         res[*count].name = *name ? name : "Unknown";
 905         res[*count].hits = client_get_hits(cl);
 906         res[*count].records = client_get_num_records(cl);
 907         res[*count].diagnostic = client_get_diagnostic(cl);
 908         res[*count].state = client_get_state_str(cl);
 909         res[*count].connected  = client_get_connection(cl) ? 1 : 0;
 910         session_settings_dump(se, client_get_database(cl), w);
 911         res[*count].settings_xml = nmem_strdup(nmem, wrbuf_cstr(w));
 912         wrbuf_rewind(w);
 913         wrbuf_puts(w, "");
 914         res[*count].suggestions_xml = nmem_strdup(nmem, client_get_suggestions_xml(cl, w));
 915         wrbuf_destroy(w);
 916         (*count)++;
 917     }
 918     return res;
 919 }
 920
 921 struct hitsbytarget *get_hitsbytarget(struct session *se, int *count, NMEM nmem)
 922 {
 923     struct hitsbytarget *p;
 924     session_enter(se);
 925     p = hitsbytarget_nb(se, count, nmem);
 926     session_leave(se);
 927     return p;
 928 }
 929
 930 struct termlist_score **get_termlist_score(struct session *se,
 931                                            const char *name, int *num)
 932 {
 933     int i;
 934     struct termlist_score **tl = 0;
 935
 936     session_enter(se);
 937     for (i = 0; i < se->num_termlists; i++)
 938         if (!strcmp((const char *) se->termlists[i].name, name))
 939         {
 940             tl = termlist_highscore(se->termlists[i].termlist, num);
 941             break;
 942         }
 943     session_leave(se);
 944     return tl;
 945 }
 946
 947 // Compares two hitsbytarget nodes by hitcount
 948 static int cmp_ht(const void *p1, const void *p2)
 949 {
 950     const struct hitsbytarget *h1 = p1;
 951     const struct hitsbytarget *h2 = p2;
 952     return h2->hits - h1->hits;
 953 }
 954
 955 static int targets_termlist_nb(WRBUF wrbuf, struct session *se, int num,
 956                                NMEM nmem)
 957 {
 958     struct hitsbytarget *ht;
 959     int count, i;
 960
 961     ht = hitsbytarget_nb(se, &count, nmem);
 962     qsort(ht, count, sizeof(struct hitsbytarget), cmp_ht);
 963     for (i = 0; i < count && i < num && ht[i].hits > 0; i++)
 964     {
 965
 966         // do only print terms which have display names
 967
 968         wrbuf_puts(wrbuf, "<term>\n");
 969
 970         wrbuf_puts(wrbuf, "<id>");
 971         wrbuf_xmlputs(wrbuf, ht[i].id);
 972         wrbuf_puts(wrbuf, "</id>\n");
 973
 974         wrbuf_puts(wrbuf, "<name>");
 975         if (!ht[i].name || !ht[i].name[0])
 976             wrbuf_xmlputs(wrbuf, "NO TARGET NAME");
 977         else
 978             wrbuf_xmlputs(wrbuf, ht[i].name);
 979         wrbuf_puts(wrbuf, "</name>\n");
 980
 981         wrbuf_printf(wrbuf, "<frequency>" ODR_INT_PRINTF "</frequency>\n",
 982                      ht[i].hits);
 983
 984         wrbuf_puts(wrbuf, "<state>");
 985         wrbuf_xmlputs(wrbuf, ht[i].state);
 986         wrbuf_puts(wrbuf, "</state>\n");
 987
 988         wrbuf_printf(wrbuf, "<diagnostic>%d</diagnostic>\n",
 989                      ht[i].diagnostic);
 990         wrbuf_puts(wrbuf, "</term>\n");
 991     }
 992     return count;
 993 }
 994
 995 void perform_termlist(struct http_channel *c, struct session *se,
 996                       const char *name, int num)
 997 {
 998     int i, j;
 999     NMEM nmem_tmp = nmem_create();
1000     char **names;
1001     int num_names = 0;
1002
1003     if (name)
1004         nmem_strsplit(nmem_tmp, ",", name, &names, &num_names);
1005
1006     session_enter(se);
1007
1008     for (j = 0; j < num_names; j++)
1009     {
1010         const char *tname;
1011
1012         wrbuf_puts(c->wrbuf, "<list name=\"");
1013         wrbuf_xmlputs(c->wrbuf, names[j]);
1014         wrbuf_puts(c->wrbuf, "\">\n");
1015
1016         for (i = 0; i < se->num_termlists; i++)
1017         {
1018             tname = se->termlists[i].name;
1019             if (num_names > 0 && !strcmp(names[j], tname))
1020             {
1021                 struct termlist_score **p = 0;
1022                 int len;
1023                 p = termlist_highscore(se->termlists[i].termlist, &len);
1024                 if (p)
1025                 {
1026                     int i;
1027                     for (i = 0; i < len && i < num; i++)
1028                     {
1029                         // prevent sending empty term elements
1030                         if (!p[i]->display_term || !p[i]->display_term[0])
1031                             continue;
1032
1033                         wrbuf_puts(c->wrbuf, "<term>");
1034                         wrbuf_puts(c->wrbuf, "<name>");
1035                         wrbuf_xmlputs(c->wrbuf, p[i]->display_term);
1036                         wrbuf_puts(c->wrbuf, "</name>");
1037
1038                         wrbuf_printf(c->wrbuf,
1039                                      "<frequency>%d</frequency>",
1040                                      p[i]->frequency);
1041                         wrbuf_puts(c->wrbuf, "</term>\n");
1042                     }
1043                 }
1044             }
1045         }
1046         tname = "xtargets";
1047         if (num_names > 0 && !strcmp(names[j], tname))
1048         {
1049             targets_termlist_nb(c->wrbuf, se, num, c->nmem);
1050         }
1051         wrbuf_puts(c->wrbuf, "</list>\n");
1052     }
1053     session_leave(se);
1054     nmem_destroy(nmem_tmp);
1055 }
1056
1057 #ifdef MISSING_HEADERS
1058 void report_nmem_stats(void)
1059 {
1060     size_t in_use, is_free;
1061
1062     nmem_get_memory_in_use(&in_use);
1063     nmem_get_memory_free(&is_free);
1064
1065     yaz_log(YLOG_LOG, "nmem stat: use=%ld free=%ld",
1066             (long) in_use, (long) is_free);
1067 }
1068 #endif
1069
1070 struct record_cluster *show_single_start(struct session *se, const char *id,
1071                                          struct record_cluster **prev_r,
1072                                          struct record_cluster **next_r)
1073 {
1074     struct record_cluster *r = 0;
1075
1076     session_enter(se);
1077     *prev_r = 0;
1078     *next_r = 0;
1079     if (se->reclist)
1080     {
1081         reclist_enter(se->reclist);
1082         while ((r = reclist_read_record(se->reclist)))
1083         {
1084             if (!strcmp(r->recid, id))
1085             {
1086                 *next_r = reclist_read_record(se->reclist);
1087                 break;
1088             }
1089             *prev_r = r;
1090         }
1091         reclist_leave(se->reclist);
1092     }
1093     if (!r)
1094         session_leave(se);
1095     return r;
1096 }
1097
1098 void show_single_stop(struct session *se, struct record_cluster *rec)
1099 {
1100     session_leave(se);
1101 }
1102
1103 struct record_cluster **show_range_start(struct session *se,
1104                                          struct reclist_sortparms *sp,
1105                                          int start, int *num, int *total, Odr_int *sumhits)
1106 {
1107     struct record_cluster **recs;
1108     struct reclist_sortparms *spp;
1109     int i;
1110 #if USE_TIMING
1111     yaz_timing_t t = yaz_timing_create();
1112 #endif
1113     session_enter(se);
1114     recs = nmem_malloc(se->nmem, *num * sizeof(struct record_cluster *));
1115     if (!se->relevance)
1116     {
1117         *num = 0;
1118         *total = 0;
1119         *sumhits = 0;
1120         recs = 0;
1121     }
1122     else
1123     {
1124         struct client_list *l;
1125
1126         for (spp = sp; spp; spp = spp->next)
1127             if (spp->type == Metadata_sortkey_relevance)
1128             {
1129                 relevance_prepare_read(se->relevance, se->reclist);
1130                 break;
1131             }
1132         reclist_sort(se->reclist, sp);
1133
1134         reclist_enter(se->reclist);
1135         *total = reclist_get_num_records(se->reclist);
1136
1137         *sumhits = 0;
1138         for (l = se->clients; l; l = l->next)
1139             *sumhits += client_get_hits(l->client);
1140
1141         for (i = 0; i < start; i++)
1142             if (!reclist_read_record(se->reclist))
1143             {
1144                 *num = 0;
1145                 recs = 0;
1146                 break;
1147             }
1148
1149         for (i = 0; i < *num; i++)
1150         {
1151             struct record_cluster *r = reclist_read_record(se->reclist);
1152             if (!r)
1153             {
1154                 *num = i;
1155                 break;
1156             }
1157             recs[i] = r;
1158         }
1159         reclist_leave(se->reclist);
1160     }
1161 #if USE_TIMING
1162     yaz_timing_stop(t);
1163     yaz_log(YLOG_LOG, "show %6.5f %3.2f %3.2f",
1164             yaz_timing_get_real(t), yaz_timing_get_user(t),
1165             yaz_timing_get_sys(t));
1166     yaz_timing_destroy(&t);
1167 #endif
1168     return recs;
1169 }
1170
1171 void show_range_stop(struct session *se, struct record_cluster **recs)
1172 {
1173     session_leave(se);
1174 }
1175
1176 void statistics(struct session *se, struct statistics *stat)
1177 {
1178     struct client_list *l;
1179     int count = 0;
1180
1181     memset(stat, 0, sizeof(*stat));
1182     stat->num_hits = 0;
1183     for (l = se->clients; l; l = l->next)
1184     {
1185         struct client *cl = l->client;
1186         if (!client_get_connection(cl))
1187             stat->num_no_connection++;
1188         stat->num_hits += client_get_hits(cl);
1189         switch (client_get_state(cl))
1190         {
1191         case Client_Connecting: stat->num_connecting++; break;
1192         case Client_Working: stat->num_working++; break;
1193         case Client_Idle: stat->num_idle++; break;
1194         case Client_Failed: stat->num_failed++; break;
1195         case Client_Error: stat->num_error++; break;
1196         default: break;
1197         }
1198         count++;
1199     }
1200     stat->num_records = se->total_records;
1201
1202     stat->num_clients = count;
1203 }
1204
1205 static struct record_metadata *record_metadata_init(
1206     NMEM nmem, const char *value, enum conf_metadata_type type,
1207     struct _xmlAttr *attr)
1208 {
1209     struct record_metadata *rec_md = record_metadata_create(nmem);
1210     struct record_metadata_attr **attrp = &rec_md->attributes;
1211
1212     for (; attr; attr = attr->next)
1213     {
1214         if (attr->children && attr->children->content)
1215         {
1216             if (strcmp((const char *) attr->name, "type"))
1217             {  /* skip the "type" attribute.. Its value is already part of
1218                   the element in output (md-%s) and so repeating it here
1219                   is redundant */
1220                 *attrp = nmem_malloc(nmem, sizeof(**attrp));
1221                 (*attrp)->name =
1222                     nmem_strdup(nmem, (const char *) attr->name);
1223                 (*attrp)->value =
1224                     nmem_strdup(nmem, (const char *) attr->children->content);
1225                 attrp = &(*attrp)->next;
1226             }
1227         }
1228     }
1229     *attrp = 0;
1230
1231     if (type == Metadata_type_generic)
1232     {
1233         char *p = nmem_strdup(nmem, value);
1234
1235         p = normalize7bit_generic(p, " ,/.:([");
1236
1237         rec_md->data.text.disp = p;
1238         rec_md->data.text.sort = 0;
1239     }
1240     else if (type == Metadata_type_year || type == Metadata_type_date)
1241     {
1242         int first, last;
1243         int longdate = 0;
1244
1245         if (type == Metadata_type_date)
1246             longdate = 1;
1247         if (extract7bit_dates((char *) value, &first, &last, longdate) < 0)
1248             return 0;
1249
1250         rec_md->data.number.min = first;
1251         rec_md->data.number.max = last;
1252     }
1253     else
1254         return 0;
1255     return rec_md;
1256 }
1257
1258 static int get_mergekey_from_doc(xmlDoc *doc, xmlNode *root, const char *name,
1259                                  struct conf_service *service, WRBUF norm_wr)
1260 {
1261     xmlNode *n;
1262     int no_found = 0;
1263     for (n = root->children; n; n = n->next)
1264     {
1265         if (n->type != XML_ELEMENT_NODE)
1266             continue;
1267         if (!strcmp((const char *) n->name, "metadata"))
1268         {
1269             xmlChar *type = xmlGetProp(n, (xmlChar *) "type");
1270             if (type == NULL) {
1271                 yaz_log(YLOG_FATAL, "Missing type attribute on metadata element. Skipping!");
1272             }
1273             else if (!strcmp(name, (const char *) type))
1274             {
1275                 xmlChar *value = xmlNodeListGetString(doc, n->children, 1);
1276                 if (value)
1277                 {
1278                     const char *norm_str;
1279                     pp2_charset_token_t prt =
1280                         pp2_charset_token_create(service->charsets, "mergekey");
1281
1282                     pp2_charset_token_first(prt, (const char *) value, 0);
1283                     if (wrbuf_len(norm_wr) > 0)
1284                         wrbuf_puts(norm_wr, " ");
1285                     wrbuf_puts(norm_wr, name);
1286                     while ((norm_str =
1287                             pp2_charset_token_next(prt)))
1288                     {
1289                         if (*norm_str)
1290                         {
1291                             wrbuf_puts(norm_wr, " ");
1292                             wrbuf_puts(norm_wr, norm_str);
1293                         }
1294                     }
1295                     xmlFree(value);
1296                     pp2_charset_token_destroy(prt);
1297                     no_found++;
1298                 }
1299             }
1300             xmlFree(type);
1301         }
1302     }
1303     return no_found;
1304 }
1305
1306 static const char *get_mergekey(xmlDoc *doc, struct client *cl, int record_no,
1307                                 struct conf_service *service, NMEM nmem)
1308 {
1309     char *mergekey_norm = 0;
1310     xmlNode *root = xmlDocGetRootElement(doc);
1311     WRBUF norm_wr = wrbuf_alloc();
1312
1313     /* consider mergekey from XSL first */
1314     xmlChar *mergekey = xmlGetProp(root, (xmlChar *) "mergekey");
1315     if (mergekey)
1316     {
1317         const char *norm_str;
1318         pp2_charset_token_t prt =
1319             pp2_charset_token_create(service->charsets, "mergekey");
1320
1321         pp2_charset_token_first(prt, (const char *) mergekey, 0);
1322         while ((norm_str = pp2_charset_token_next(prt)))
1323         {
1324             if (*norm_str)
1325             {
1326                 if (wrbuf_len(norm_wr))
1327                     wrbuf_puts(norm_wr, " ");
1328                 wrbuf_puts(norm_wr, norm_str);
1329             }
1330         }
1331         pp2_charset_token_destroy(prt);
1332         xmlFree(mergekey);
1333     }
1334     else
1335     {
1336         /* no mergekey defined in XSL. Look for mergekey metadata instead */
1337         int field_id;
1338         for (field_id = 0; field_id < service->num_metadata; field_id++)
1339         {
1340             struct conf_metadata *ser_md = &service->metadata[field_id];
1341             if (ser_md->mergekey != Metadata_mergekey_no)
1342             {
1343                 int r = get_mergekey_from_doc(doc, root, ser_md->name,
1344                                               service, norm_wr);
1345                 if (r == 0 && ser_md->mergekey == Metadata_mergekey_required)
1346                 {
1347                     /* no mergekey on this one and it is required..
1348                        Generate unique key instead */
1349                     wrbuf_rewind(norm_wr);
1350                     break;
1351                 }
1352             }
1353         }
1354     }
1355
1356     /* generate unique key if none is not generated already or is empty */
1357     if (wrbuf_len(norm_wr) == 0)
1358     {
1359         wrbuf_printf(norm_wr, "position: %s-%d",
1360                      client_get_id(cl), record_no);
1361     }
1362     else
1363     {
1364         const char *lead = "content: ";
1365         wrbuf_insert(norm_wr, 0, lead, strlen(lead));
1366     }
1367     if (wrbuf_len(norm_wr) > 0)
1368         mergekey_norm = nmem_strdup(nmem, wrbuf_cstr(norm_wr));
1369     wrbuf_destroy(norm_wr);
1370     return mergekey_norm;
1371 }
1372
1373 /** \brief see if metadata for pz:recordfilter exists
1374     \param root xml root element of normalized record
1375     \param sdb session database for client
1376     \retval 0 if there is no metadata for pz:recordfilter
1377     \retval 1 if there is metadata for pz:recordfilter
1378
1379     If there is no pz:recordfilter defined, this function returns 1
1380     as well.
1381 */
1382
1383 static int check_record_filter(xmlNode *root, struct session_database *sdb)
1384 {
1385     int match = 0;
1386     xmlNode *n;
1387     const char *s;
1388     s = session_setting_oneval(sdb, PZ_RECORDFILTER);
1389
1390     if (!s || !*s)
1391         return 1;
1392
1393     for (n = root->children; n; n = n->next)
1394     {
1395         if (n->type != XML_ELEMENT_NODE)
1396             continue;
1397         if (!strcmp((const char *) n->name, "metadata"))
1398         {
1399             xmlChar *type = xmlGetProp(n, (xmlChar *) "type");
1400             if (type)
1401             {
1402                 size_t len;
1403                 int substring;
1404                 const char *eq;
1405
1406                 if ((eq = strchr(s, '=')))
1407                     substring = 0;
1408                 else if ((eq = strchr(s, '~')))
1409                     substring = 1;
1410                 if (eq)
1411                     len = eq - s;
1412                 else
1413                     len = strlen(s);
1414                 if (len == strlen((const char *)type) &&
1415                     !memcmp((const char *) type, s, len))
1416                 {
1417                     xmlChar *value = xmlNodeGetContent(n);
1418                     if (value && *value)
1419                     {
1420                         if (!eq ||
1421                             (substring && strstr((const char *) value, eq+1)) ||
1422                             (!substring && !strcmp((const char *) value, eq + 1)))
1423                             match = 1;
1424                     }
1425                     xmlFree(value);
1426                 }
1427                 xmlFree(type);
1428             }
1429         }
1430     }
1431     return match;
1432 }
1433
1434
1435 static int ingest_to_cluster(struct client *cl,
1436                              xmlDoc *xdoc,
1437                              xmlNode *root,
1438                              int record_no,
1439                              const char *mergekey_norm);
1440
1441 /** \brief ingest XML record
1442     \param cl client holds the result set for record
1443     \param rec record buffer (0 terminated)
1444     \param record_no record position (1, 2, ..)
1445     \param nmem working NMEM
1446     \retval 0 OK
1447     \retval -1 failure
1448     \retval -2 Filtered
1449 */
1450 int ingest_record(struct client *cl, const char *rec,
1451                   int record_no, NMEM nmem)
1452 {
1453     struct session *se = client_get_session(cl);
1454     int ret = 0;
1455     struct session_database *sdb = client_get_database(cl);
1456     struct conf_service *service = se->service;
1457     xmlDoc *xdoc = normalize_record(se, sdb, service, rec, nmem);
1458     xmlNode *root;
1459     const char *mergekey_norm;
1460
1461     if (!xdoc)
1462         return -1;
1463
1464     root = xmlDocGetRootElement(xdoc);
1465
1466     if (!check_record_filter(root, sdb))
1467     {
1468         session_log(se, YLOG_LOG, "Filtered out record no %d from %s",
1469                     record_no, sdb->database->id);
1470         xmlFreeDoc(xdoc);
1471         return -2;
1472     }
1473
1474     mergekey_norm = get_mergekey(xdoc, cl, record_no, service, nmem);
1475     if (!mergekey_norm)
1476     {
1477         session_log(se, YLOG_WARN, "Got no mergekey");
1478         xmlFreeDoc(xdoc);
1479         return -1;
1480     }
1481     session_enter(se);
1482     if (client_get_session(cl) == se)
1483         ingest_to_cluster(cl, xdoc, root, record_no, mergekey_norm);
1484     session_leave(se);
1485
1486     xmlFreeDoc(xdoc);
1487     return ret;
1488 }
1489
1490 static int ingest_to_cluster(struct client *cl,
1491                              xmlDoc *xdoc,
1492                              xmlNode *root,
1493                              int record_no,
1494                              const char *mergekey_norm)
1495 {
1496     xmlNode *n;
1497     xmlChar *type = 0;
1498     xmlChar *value = 0;
1499     struct session *se = client_get_session(cl);
1500     struct conf_service *service = se->service;
1501     int term_factor = 1;
1502     struct record_cluster *cluster;
1503     struct session_database *sdb = client_get_database(cl);
1504     struct record *record = record_create(se->nmem,
1505                                           service->num_metadata,
1506                                           service->num_sortkeys, cl,
1507                                           record_no);
1508
1509     for (n = root->children; n; n = n->next)
1510     {
1511         if (type)
1512             xmlFree(type);
1513         if (value)
1514             xmlFree(value);
1515         type = value = 0;
1516
1517         if (n->type != XML_ELEMENT_NODE)
1518             continue;
1519         if (!strcmp((const char *) n->name, "metadata"))
1520         {
1521             struct conf_metadata *ser_md = 0;
1522             struct record_metadata **wheretoput = 0;
1523             struct record_metadata *rec_md = 0;
1524             int md_field_id = -1;
1525
1526             type = xmlGetProp(n, (xmlChar *) "type");
1527             value = xmlNodeListGetString(xdoc, n->children, 1);
1528
1529             if (!type || !value || !*value)
1530                 continue;
1531
1532             md_field_id
1533                 = conf_service_metadata_field_id(service, (const char *) type);
1534             if (md_field_id < 0)
1535             {
1536                 if (se->number_of_warnings_unknown_metadata == 0)
1537                 {
1538                     session_log(se, YLOG_WARN,
1539                             "Ignoring unknown metadata element: %s", type);
1540                 }
1541                 se->number_of_warnings_unknown_metadata++;
1542                 continue;
1543             }
1544
1545             ser_md = &service->metadata[md_field_id];
1546
1547             // non-merged metadata
1548             rec_md = record_metadata_init(se->nmem, (const char *) value,
1549                                           ser_md->type, n->properties);
1550             if (!rec_md)
1551             {
1552                 session_log(se, YLOG_WARN, "bad metadata data '%s' "
1553                             "for element '%s'", value, type);
1554                 continue;
1555             }
1556             wheretoput = &record->metadata[md_field_id];
1557             while (*wheretoput)
1558                 wheretoput = &(*wheretoput)->next;
1559             *wheretoput = rec_md;
1560         }
1561     }
1562
1563     cluster = reclist_insert(se->reclist, service, record,
1564                              mergekey_norm, &se->total_merged);
1565     if (!cluster)
1566         return -1;
1567
1568     {
1569         const char *use_term_factor_str =
1570             session_setting_oneval(sdb, PZ_TERMLIST_TERM_FACTOR);
1571         if (use_term_factor_str && use_term_factor_str[0] == '1')
1572         {
1573             int maxrecs = client_get_maxrecs(cl);
1574             int hits = (int) client_get_hits(cl);
1575             term_factor = MAX(hits, maxrecs) /  MAX(1, maxrecs);
1576             assert(term_factor >= 1);
1577             yaz_log(YLOG_DEBUG, "Using term factor: %d (%d / %d)", term_factor, MAX(hits, maxrecs), MAX(1, maxrecs));
1578         }
1579     }
1580
1581     if (global_parameters.dump_records)
1582         session_log(se, YLOG_LOG, "Cluster id %s from %s (#%d)", cluster->recid,
1583                     sdb->database->id, record_no);
1584
1585
1586     relevance_newrec(se->relevance, cluster);
1587
1588     // now parsing XML record and adding data to cluster or record metadata
1589     for (n = root->children; n; n = n->next)
1590     {
1591         pp2_charset_token_t prt;
1592         if (type)
1593             xmlFree(type);
1594         if (value)
1595             xmlFree(value);
1596         type = value = 0;
1597
1598         if (n->type != XML_ELEMENT_NODE)
1599             continue;
1600         if (!strcmp((const char *) n->name, "metadata"))
1601         {
1602             struct conf_metadata *ser_md = 0;
1603             struct conf_sortkey *ser_sk = 0;
1604             struct record_metadata **wheretoput = 0;
1605             struct record_metadata *rec_md = 0;
1606             int md_field_id = -1;
1607             int sk_field_id = -1;
1608
1609             type = xmlGetProp(n, (xmlChar *) "type");
1610             value = xmlNodeListGetString(xdoc, n->children, 1);
1611
1612             if (!type || !value || !*value)
1613                 continue;
1614
1615             md_field_id
1616                 = conf_service_metadata_field_id(service, (const char *) type);
1617             if (md_field_id < 0)
1618                 continue;
1619
1620             ser_md = &service->metadata[md_field_id];
1621
1622             if (ser_md->sortkey_offset >= 0)
1623             {
1624                 sk_field_id = ser_md->sortkey_offset;
1625                 ser_sk = &service->sortkeys[sk_field_id];
1626             }
1627
1628             // merged metadata
1629             rec_md = record_metadata_init(se->nmem, (const char *) value,
1630                                           ser_md->type, 0);
1631             if (!rec_md)
1632                 continue;
1633
1634             wheretoput = &cluster->metadata[md_field_id];
1635
1636             // and polulate with data:
1637             // assign cluster or record based on merge action
1638             if (ser_md->merge == Metadata_merge_unique)
1639             {
1640                 while (*wheretoput)
1641                 {
1642                     if (!strcmp((const char *) (*wheretoput)->data.text.disp,
1643                                 rec_md->data.text.disp))
1644                         break;
1645                     wheretoput = &(*wheretoput)->next;
1646                 }
1647                 if (!*wheretoput)
1648                     *wheretoput = rec_md;
1649             }
1650             else if (ser_md->merge == Metadata_merge_longest)
1651             {
1652                 if (!*wheretoput
1653                     || strlen(rec_md->data.text.disp)
1654                     > strlen((*wheretoput)->data.text.disp))
1655                 {
1656                     *wheretoput = rec_md;
1657                     if (ser_sk)
1658                     {
1659                         const char *sort_str = 0;
1660                         int skip_article =
1661                             ser_sk->type == Metadata_sortkey_skiparticle;
1662
1663                         if (!cluster->sortkeys[sk_field_id])
1664                             cluster->sortkeys[sk_field_id] =
1665                                 nmem_malloc(se->nmem,
1666                                             sizeof(union data_types));
1667
1668                         prt =
1669                             pp2_charset_token_create(service->charsets, "sort");
1670
1671                         pp2_charset_token_first(prt, rec_md->data.text.disp,
1672                                                 skip_article);
1673
1674                         pp2_charset_token_next(prt);
1675
1676                         sort_str = pp2_get_sort(prt);
1677
1678                         cluster->sortkeys[sk_field_id]->text.disp =
1679                             rec_md->data.text.disp;
1680                         if (!sort_str)
1681                         {
1682                             sort_str = rec_md->data.text.disp;
1683                             session_log(se, YLOG_WARN,
1684                                     "Could not make sortkey. Bug #1858");
1685                         }
1686                         cluster->sortkeys[sk_field_id]->text.sort =
1687                             nmem_strdup(se->nmem, sort_str);
1688                         pp2_charset_token_destroy(prt);
1689                     }
1690                 }
1691             }
1692             else if (ser_md->merge == Metadata_merge_all)
1693             {
1694                 while (*wheretoput)
1695                     wheretoput = &(*wheretoput)->next;
1696                 *wheretoput = rec_md;
1697             }
1698             else if (ser_md->merge == Metadata_merge_range)
1699             {
1700                 if (!*wheretoput)
1701                 {
1702                     *wheretoput = rec_md;
1703                     if (ser_sk)
1704                         cluster->sortkeys[sk_field_id]
1705                             = &rec_md->data;
1706                 }
1707                 else
1708                 {
1709                     int this_min = rec_md->data.number.min;
1710                     int this_max = rec_md->data.number.max;
1711                     if (this_min < (*wheretoput)->data.number.min)
1712                         (*wheretoput)->data.number.min = this_min;
1713                     if (this_max > (*wheretoput)->data.number.max)
1714                         (*wheretoput)->data.number.max = this_max;
1715                 }
1716             }
1717
1718
1719             // ranking of _all_ fields enabled ...
1720             if (ser_md->rank)
1721                 relevance_countwords(se->relevance, cluster,
1722                                      (char *) value, ser_md->rank,
1723                                      ser_md->name);
1724
1725             // construct facets ... unless the client already has reported them
1726             if (ser_md->termlist && !client_has_facet(cl, (char *) type))
1727             {
1728                 if (ser_md->type == Metadata_type_year)
1729                 {
1730                     char year[64];
1731                     sprintf(year, "%d", rec_md->data.number.max);
1732
1733                     add_facet(se, (char *) type, year, term_factor);
1734                     if (rec_md->data.number.max != rec_md->data.number.min)
1735                     {
1736                         sprintf(year, "%d", rec_md->data.number.min);
1737                         add_facet(se, (char *) type, year, term_factor);
1738                     }
1739                 }
1740                 else
1741                     add_facet(se, (char *) type, (char *) value, term_factor);
1742             }
1743
1744             // cleaning up
1745             xmlFree(type);
1746             xmlFree(value);
1747             type = value = 0;
1748         }
1749         else
1750         {
1751             if (se->number_of_warnings_unknown_elements == 0)
1752                 session_log(se, YLOG_WARN,
1753                         "Unexpected element in internal record: %s", n->name);
1754             se->number_of_warnings_unknown_elements++;
1755         }
1756     }
1757     if (type)
1758         xmlFree(type);
1759     if (value)
1760         xmlFree(value);
1761
1762     relevance_donerecord(se->relevance, cluster);
1763     se->total_records++;
1764
1765     return 0;
1766 }
1767
1768 void session_log(struct session *s, int level, const char *fmt, ...)
1769 {
1770     char buf[1024];
1771     va_list ap;
1772     va_start(ap, fmt);
1773
1774     yaz_vsnprintf(buf, sizeof(buf)-30, fmt, ap);
1775     yaz_log(level, "Session (%u): %s", s->session_id, buf);
1776
1777     va_end(ap);
1778 }
1779
1780 /*
1781  * Local variables:
1782  * c-basic-offset: 4
1783  * c-file-style: "Stroustrup"
1784  * indent-tabs-mode: nil
1785  * End:
1786  * vim: shiftwidth=4 tabstop=8 expandtab
1787  */
1788