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_hits = 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_hits = 0;
 871     session->total_records = 0;
 872     session->number_of_warnings_unknown_elements = 0;
 873     session->number_of_warnings_unknown_metadata = 0;
 874     session->num_termlists = 0;
 875     session->reclist = 0;
 876     session->clients = 0;
 877     session->session_nmem = nmem;
 878     session->nmem = nmem_create();
 879     session->databases = 0;
 880     for (i = 0; i <= SESSION_WATCH_MAX; i++)
 881     {
 882         session->watchlist[i].data = 0;
 883         session->watchlist[i].fun = 0;
 884     }
 885     session->normalize_cache = normalize_cache_create();
 886     session->session_mutex = 0;
 887     pazpar2_mutex_create(&session->session_mutex, tmp_str);
 888     session_use(1);
 889     return session;
 890 }
 891
 892 static struct hitsbytarget *hitsbytarget_nb(struct session *se,
 893                                             int *count, NMEM nmem)
 894 {
 895     struct hitsbytarget *res = 0;
 896     struct client_list *l;
 897     size_t sz = 0;
 898
 899     for (l = se->clients; l; l = l->next)
 900         sz++;
 901
 902     res = nmem_malloc(nmem, sizeof(*res) * sz);
 903     *count = 0;
 904     for (l = se->clients; l; l = l->next)
 905     {
 906         struct client *cl = l->client;
 907         WRBUF w = wrbuf_alloc();
 908         const char *name = session_setting_oneval(client_get_database(cl),
 909                                                   PZ_NAME);
 910
 911         res[*count].id = client_get_id(cl);
 912         res[*count].name = *name ? name : "Unknown";
 913         res[*count].hits = client_get_hits(cl);
 914         res[*count].records = client_get_num_records(cl);
 915         res[*count].diagnostic = client_get_diagnostic(cl);
 916         res[*count].state = client_get_state_str(cl);
 917         res[*count].connected  = client_get_connection(cl) ? 1 : 0;
 918         session_settings_dump(se, client_get_database(cl), w);
 919         res[*count].settings_xml = nmem_strdup(nmem, wrbuf_cstr(w));
 920         wrbuf_destroy(w);
 921         (*count)++;
 922     }
 923     return res;
 924 }
 925
 926 struct hitsbytarget *get_hitsbytarget(struct session *se, int *count, NMEM nmem)
 927 {
 928     struct hitsbytarget *p;
 929     session_enter(se);
 930     p = hitsbytarget_nb(se, count, nmem);
 931     session_leave(se);
 932     return p;
 933 }
 934
 935 struct termlist_score **get_termlist_score(struct session *se,
 936                                            const char *name, int *num)
 937 {
 938     int i;
 939     struct termlist_score **tl = 0;
 940
 941     session_enter(se);
 942     for (i = 0; i < se->num_termlists; i++)
 943         if (!strcmp((const char *) se->termlists[i].name, name))
 944         {
 945             tl = termlist_highscore(se->termlists[i].termlist, num);
 946             break;
 947         }
 948     session_leave(se);
 949     return tl;
 950 }
 951
 952 // Compares two hitsbytarget nodes by hitcount
 953 static int cmp_ht(const void *p1, const void *p2)
 954 {
 955     const struct hitsbytarget *h1 = p1;
 956     const struct hitsbytarget *h2 = p2;
 957     return h2->hits - h1->hits;
 958 }
 959
 960 static int targets_termlist_nb(WRBUF wrbuf, struct session *se, int num,
 961                                NMEM nmem)
 962 {
 963     struct hitsbytarget *ht;
 964     int count, i;
 965
 966     ht = hitsbytarget_nb(se, &count, nmem);
 967     qsort(ht, count, sizeof(struct hitsbytarget), cmp_ht);
 968     for (i = 0; i < count && i < num && ht[i].hits > 0; i++)
 969     {
 970
 971         // do only print terms which have display names
 972
 973         wrbuf_puts(wrbuf, "<term>\n");
 974
 975         wrbuf_puts(wrbuf, "<id>");
 976         wrbuf_xmlputs(wrbuf, ht[i].id);
 977         wrbuf_puts(wrbuf, "</id>\n");
 978
 979         wrbuf_puts(wrbuf, "<name>");
 980         if (!ht[i].name || !ht[i].name[0])
 981             wrbuf_xmlputs(wrbuf, "NO TARGET NAME");
 982         else
 983             wrbuf_xmlputs(wrbuf, ht[i].name);
 984         wrbuf_puts(wrbuf, "</name>\n");
 985
 986         wrbuf_printf(wrbuf, "<frequency>" ODR_INT_PRINTF "</frequency>\n",
 987                      ht[i].hits);
 988
 989         wrbuf_puts(wrbuf, "<state>");
 990         wrbuf_xmlputs(wrbuf, ht[i].state);
 991         wrbuf_puts(wrbuf, "</state>\n");
 992
 993         wrbuf_printf(wrbuf, "<diagnostic>%d</diagnostic>\n",
 994                      ht[i].diagnostic);
 995         wrbuf_puts(wrbuf, "</term>\n");
 996     }
 997     return count;
 998 }
 999
1000 void perform_termlist(struct http_channel *c, struct session *se,
1001                       const char *name, int num)
1002 {
1003     int i, j;
1004     NMEM nmem_tmp = nmem_create();
1005     char **names;
1006     int num_names = 0;
1007
1008     if (name)
1009         nmem_strsplit(nmem_tmp, ",", name, &names, &num_names);
1010
1011     session_enter(se);
1012
1013     for (j = 0; j < num_names; j++)
1014     {
1015         const char *tname;
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                     wrbuf_puts(c->wrbuf, "<list name=\"");
1028                     wrbuf_xmlputs(c->wrbuf, tname);
1029                     wrbuf_puts(c->wrbuf, "\">\n");
1030                     for (i = 0; i < len && i < num; i++)
1031                     {
1032                         // prevent sending empty term elements
1033                         if (!p[i]->display_term || !p[i]->display_term[0])
1034                             continue;
1035
1036                         wrbuf_puts(c->wrbuf, "<term>");
1037                         wrbuf_puts(c->wrbuf, "<name>");
1038                         wrbuf_xmlputs(c->wrbuf, p[i]->display_term);
1039                         wrbuf_puts(c->wrbuf, "</name>");
1040
1041                         wrbuf_printf(c->wrbuf,
1042                                      "<frequency>%d</frequency>",
1043                                      p[i]->frequency);
1044                         wrbuf_puts(c->wrbuf, "</term>\n");
1045                     }
1046                     wrbuf_puts(c->wrbuf, "</list>\n");
1047                 }
1048             }
1049         }
1050         tname = "xtargets";
1051         if (num_names > 0 && !strcmp(names[j], tname))
1052         {
1053             wrbuf_puts(c->wrbuf, "<list name=\"");
1054             wrbuf_xmlputs(c->wrbuf, tname);
1055             wrbuf_puts(c->wrbuf, "\">\n");
1056             targets_termlist_nb(c->wrbuf, se, num, c->nmem);
1057             wrbuf_puts(c->wrbuf, "</list>\n");
1058         }
1059     }
1060     session_leave(se);
1061     nmem_destroy(nmem_tmp);
1062 }
1063
1064 #ifdef MISSING_HEADERS
1065 void report_nmem_stats(void)
1066 {
1067     size_t in_use, is_free;
1068
1069     nmem_get_memory_in_use(&in_use);
1070     nmem_get_memory_free(&is_free);
1071
1072     yaz_log(YLOG_LOG, "nmem stat: use=%ld free=%ld",
1073             (long) in_use, (long) is_free);
1074 }
1075 #endif
1076
1077 struct record_cluster *show_single_start(struct session *se, const char *id,
1078                                          struct record_cluster **prev_r,
1079                                          struct record_cluster **next_r)
1080 {
1081     struct record_cluster *r = 0;
1082
1083     session_enter(se);
1084     *prev_r = 0;
1085     *next_r = 0;
1086     if (se->reclist)
1087     {
1088         reclist_enter(se->reclist);
1089         while ((r = reclist_read_record(se->reclist)))
1090         {
1091             if (!strcmp(r->recid, id))
1092             {
1093                 *next_r = reclist_read_record(se->reclist);
1094                 break;
1095             }
1096             *prev_r = r;
1097         }
1098         reclist_leave(se->reclist);
1099     }
1100     if (!r)
1101         session_leave(se);
1102     return r;
1103 }
1104
1105 void show_single_stop(struct session *se, struct record_cluster *rec)
1106 {
1107     session_leave(se);
1108 }
1109
1110 struct record_cluster **show_range_start(struct session *se,
1111                                          struct reclist_sortparms *sp,
1112                                          int start, int *num, int *total, Odr_int *sumhits)
1113 {
1114     struct record_cluster **recs;
1115     struct reclist_sortparms *spp;
1116     int i;
1117 #if USE_TIMING
1118     yaz_timing_t t = yaz_timing_create();
1119 #endif
1120     session_enter(se);
1121     recs = nmem_malloc(se->nmem, *num * sizeof(struct record_cluster *));
1122     if (!se->relevance)
1123     {
1124         *num = 0;
1125         *total = 0;
1126         *sumhits = 0;
1127         recs = 0;
1128     }
1129     else
1130     {
1131         for (spp = sp; spp; spp = spp->next)
1132             if (spp->type == Metadata_sortkey_relevance)
1133             {
1134                 relevance_prepare_read(se->relevance, se->reclist);
1135                 break;
1136             }
1137         reclist_sort(se->reclist, sp);
1138
1139         reclist_enter(se->reclist);
1140         *total = reclist_get_num_records(se->reclist);
1141         *sumhits = se->total_hits;
1142
1143         for (i = 0; i < start; i++)
1144             if (!reclist_read_record(se->reclist))
1145             {
1146                 *num = 0;
1147                 recs = 0;
1148                 break;
1149             }
1150
1151         for (i = 0; i < *num; i++)
1152         {
1153             struct record_cluster *r = reclist_read_record(se->reclist);
1154             if (!r)
1155             {
1156                 *num = i;
1157                 break;
1158             }
1159             recs[i] = r;
1160         }
1161         reclist_leave(se->reclist);
1162     }
1163 #if USE_TIMING
1164     yaz_timing_stop(t);
1165     yaz_log(YLOG_LOG, "show %6.5f %3.2f %3.2f",
1166             yaz_timing_get_real(t), yaz_timing_get_user(t),
1167             yaz_timing_get_sys(t));
1168     yaz_timing_destroy(&t);
1169 #endif
1170     return recs;
1171 }
1172
1173 void show_range_stop(struct session *se, struct record_cluster **recs)
1174 {
1175     session_leave(se);
1176 }
1177
1178 void statistics(struct session *se, struct statistics *stat)
1179 {
1180     struct client_list *l;
1181     int count = 0;
1182
1183     memset(stat, 0, sizeof(*stat));
1184     for (l = se->clients; l; l = l->next)
1185     {
1186         struct client *cl = l->client;
1187         if (!client_get_connection(cl))
1188             stat->num_no_connection++;
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_hits = se->total_hits;
1201     stat->num_records = se->total_records;
1202
1203     stat->num_clients = count;
1204 }
1205
1206 static struct record_metadata *record_metadata_init(
1207     NMEM nmem, const char *value, enum conf_metadata_type type,
1208     struct _xmlAttr *attr)
1209 {
1210     struct record_metadata *rec_md = record_metadata_create(nmem);
1211     struct record_metadata_attr **attrp = &rec_md->attributes;
1212
1213     for (; attr; attr = attr->next)
1214     {
1215         if (attr->children && attr->children->content)
1216         {
1217             if (strcmp((const char *) attr->name, "type"))
1218             {  /* skip the "type" attribute.. Its value is already part of
1219                   the element in output (md-%s) and so repeating it here
1220                   is redundant */
1221                 *attrp = nmem_malloc(nmem, sizeof(**attrp));
1222                 (*attrp)->name =
1223                     nmem_strdup(nmem, (const char *) attr->name);
1224                 (*attrp)->value =
1225                     nmem_strdup(nmem, (const char *) attr->children->content);
1226                 attrp = &(*attrp)->next;
1227             }
1228         }
1229     }
1230     *attrp = 0;
1231
1232     if (type == Metadata_type_generic)
1233     {
1234         char *p = nmem_strdup(nmem, value);
1235
1236         p = normalize7bit_generic(p, " ,/.:([");
1237
1238         rec_md->data.text.disp = p;
1239         rec_md->data.text.sort = 0;
1240     }
1241     else if (type == Metadata_type_year || type == Metadata_type_date)
1242     {
1243         int first, last;
1244         int longdate = 0;
1245
1246         if (type == Metadata_type_date)
1247             longdate = 1;
1248         if (extract7bit_dates((char *) value, &first, &last, longdate) < 0)
1249             return 0;
1250
1251         rec_md->data.number.min = first;
1252         rec_md->data.number.max = last;
1253     }
1254     else
1255         return 0;
1256     return rec_md;
1257 }
1258
1259 static int get_mergekey_from_doc(xmlDoc *doc, xmlNode *root, const char *name,
1260                                  struct conf_service *service, WRBUF norm_wr)
1261 {
1262     xmlNode *n;
1263     int no_found = 0;
1264     for (n = root->children; n; n = n->next)
1265     {
1266         if (n->type != XML_ELEMENT_NODE)
1267             continue;
1268         if (!strcmp((const char *) n->name, "metadata"))
1269         {
1270             xmlChar *type = xmlGetProp(n, (xmlChar *) "type");
1271             if (type == NULL) {
1272                 yaz_log(YLOG_FATAL, "Missing type attribute on metadata element. Skipping!");
1273             }
1274             else if (!strcmp(name, (const char *) type))
1275             {
1276                 xmlChar *value = xmlNodeListGetString(doc, n->children, 1);
1277                 if (value)
1278                 {
1279                     const char *norm_str;
1280                     pp2_charset_token_t prt =
1281                         pp2_charset_token_create(service->charsets, "mergekey");
1282
1283                     pp2_charset_token_first(prt, (const char *) value, 0);
1284                     if (wrbuf_len(norm_wr) > 0)
1285                         wrbuf_puts(norm_wr, " ");
1286                     wrbuf_puts(norm_wr, name);
1287                     while ((norm_str =
1288                             pp2_charset_token_next(prt)))
1289                     {
1290                         if (*norm_str)
1291                         {
1292                             wrbuf_puts(norm_wr, " ");
1293                             wrbuf_puts(norm_wr, norm_str);
1294                         }
1295                     }
1296                     xmlFree(value);
1297                     pp2_charset_token_destroy(prt);
1298                     no_found++;
1299                 }
1300             }
1301             xmlFree(type);
1302         }
1303     }
1304     return no_found;
1305 }
1306
1307 static const char *get_mergekey(xmlDoc *doc, struct client *cl, int record_no,
1308                                 struct conf_service *service, NMEM nmem)
1309 {
1310     char *mergekey_norm = 0;
1311     xmlNode *root = xmlDocGetRootElement(doc);
1312     WRBUF norm_wr = wrbuf_alloc();
1313
1314     /* consider mergekey from XSL first */
1315     xmlChar *mergekey = xmlGetProp(root, (xmlChar *) "mergekey");
1316     if (mergekey)
1317     {
1318         const char *norm_str;
1319         pp2_charset_token_t prt =
1320             pp2_charset_token_create(service->charsets, "mergekey");
1321
1322         pp2_charset_token_first(prt, (const char *) mergekey, 0);
1323         while ((norm_str = pp2_charset_token_next(prt)))
1324         {
1325             if (*norm_str)
1326             {
1327                 if (wrbuf_len(norm_wr))
1328                     wrbuf_puts(norm_wr, " ");
1329                 wrbuf_puts(norm_wr, norm_str);
1330             }
1331         }
1332         pp2_charset_token_destroy(prt);
1333         xmlFree(mergekey);
1334     }
1335     else
1336     {
1337         /* no mergekey defined in XSL. Look for mergekey metadata instead */
1338         int field_id;
1339         for (field_id = 0; field_id < service->num_metadata; field_id++)
1340         {
1341             struct conf_metadata *ser_md = &service->metadata[field_id];
1342             if (ser_md->mergekey != Metadata_mergekey_no)
1343             {
1344                 int r = get_mergekey_from_doc(doc, root, ser_md->name,
1345                                               service, norm_wr);
1346                 if (r == 0 && ser_md->mergekey == Metadata_mergekey_required)
1347                 {
1348                     /* no mergekey on this one and it is required..
1349                        Generate unique key instead */
1350                     wrbuf_rewind(norm_wr);
1351                     break;
1352                 }
1353             }
1354         }
1355     }
1356
1357     /* generate unique key if none is not generated already or is empty */
1358     if (wrbuf_len(norm_wr) == 0)
1359     {
1360         wrbuf_printf(norm_wr, "%s-%d",
1361                      client_get_id(cl), record_no);
1362     }
1363     if (wrbuf_len(norm_wr) > 0)
1364         mergekey_norm = nmem_strdup(nmem, wrbuf_cstr(norm_wr));
1365     wrbuf_destroy(norm_wr);
1366     return mergekey_norm;
1367 }
1368
1369 /** \brief see if metadata for pz:recordfilter exists
1370     \param root xml root element of normalized record
1371     \param sdb session database for client
1372     \retval 0 if there is no metadata for pz:recordfilter
1373     \retval 1 if there is metadata for pz:recordfilter
1374
1375     If there is no pz:recordfilter defined, this function returns 1
1376     as well.
1377 */
1378
1379 static int check_record_filter(xmlNode *root, struct session_database *sdb)
1380 {
1381     int match = 0;
1382     xmlNode *n;
1383     const char *s;
1384     s = session_setting_oneval(sdb, PZ_RECORDFILTER);
1385
1386     if (!s || !*s)
1387         return 1;
1388
1389     for (n = root->children; n; n = n->next)
1390     {
1391         if (n->type != XML_ELEMENT_NODE)
1392             continue;
1393         if (!strcmp((const char *) n->name, "metadata"))
1394         {
1395             xmlChar *type = xmlGetProp(n, (xmlChar *) "type");
1396             if (type)
1397             {
1398                 size_t len;
1399                 int substring;
1400                 const char *eq;
1401
1402                 if ((eq = strchr(s, '=')))
1403                     substring = 0;
1404                 else if ((eq = strchr(s, '~')))
1405                     substring = 1;
1406                 if (eq)
1407                     len = eq - s;
1408                 else
1409                     len = strlen(s);
1410                 if (len == strlen((const char *)type) &&
1411                     !memcmp((const char *) type, s, len))
1412                 {
1413                     xmlChar *value = xmlNodeGetContent(n);
1414                     if (value && *value)
1415                     {
1416                         if (!eq ||
1417                             (substring && strstr((const char *) value, eq+1)) ||
1418                             (!substring && !strcmp((const char *) value, eq + 1)))
1419                             match = 1;
1420                     }
1421                     xmlFree(value);
1422                 }
1423                 xmlFree(type);
1424             }
1425         }
1426     }
1427     return match;
1428 }
1429
1430
1431 static int ingest_to_cluster(struct client *cl,
1432                              xmlDoc *xdoc,
1433                              xmlNode *root,
1434                              int record_no,
1435                              const char *mergekey_norm);
1436
1437 /** \brief ingest XML record
1438     \param cl client holds the result set for record
1439     \param rec record buffer (0 terminated)
1440     \param record_no record position (1, 2, ..)
1441     \param nmem working NMEM
1442     \retval 0 OK
1443     \retval -1 failure
1444     \retval -2 Filtered
1445 */
1446 int ingest_record(struct client *cl, const char *rec,
1447                   int record_no, NMEM nmem)
1448 {
1449     struct session *se = client_get_session(cl);
1450     int ret = 0;
1451     struct session_database *sdb = client_get_database(cl);
1452     struct conf_service *service = se->service;
1453     xmlDoc *xdoc = normalize_record(se, sdb, service, rec, nmem);
1454     xmlNode *root;
1455     const char *mergekey_norm;
1456
1457     if (!xdoc)
1458         return -1;
1459
1460     root = xmlDocGetRootElement(xdoc);
1461
1462     if (!check_record_filter(root, sdb))
1463     {
1464         session_log(se, YLOG_LOG, "Filtered out record no %d from %s",
1465                     record_no, sdb->database->id);
1466         xmlFreeDoc(xdoc);
1467         return -2;
1468     }
1469
1470     mergekey_norm = get_mergekey(xdoc, cl, record_no, service, nmem);
1471     if (!mergekey_norm)
1472     {
1473         session_log(se, YLOG_WARN, "Got no mergekey");
1474         xmlFreeDoc(xdoc);
1475         return -1;
1476     }
1477     session_enter(se);
1478     if (client_get_session(cl) == se)
1479         ret = ingest_to_cluster(cl, xdoc, root, record_no, mergekey_norm);
1480     session_leave(se);
1481
1482     xmlFreeDoc(xdoc);
1483     return ret;
1484 }
1485
1486 static int ingest_to_cluster(struct client *cl,
1487                              xmlDoc *xdoc,
1488                              xmlNode *root,
1489                              int record_no,
1490                              const char *mergekey_norm)
1491 {
1492     xmlNode *n;
1493     xmlChar *type = 0;
1494     xmlChar *value = 0;
1495     struct session_database *sdb = client_get_database(cl);
1496     struct session *se = client_get_session(cl);
1497     struct conf_service *service = se->service;
1498     struct record *record = record_create(se->nmem,
1499                                           service->num_metadata,
1500                                           service->num_sortkeys, cl,
1501                                           record_no);
1502     struct record_cluster *cluster = reclist_insert(se->reclist,
1503                                                     service,
1504                                                     record,
1505                                                     mergekey_norm,
1506                                                     &se->total_merged);
1507
1508     const char *use_term_factor_str = session_setting_oneval(sdb, PZ_TERMLIST_TERM_FACTOR);
1509     int use_term_factor = 0;
1510     int term_factor = 1;
1511     if (use_term_factor_str && use_term_factor_str[0] != 0)
1512        use_term_factor =  atoi(use_term_factor_str);
1513     if (use_term_factor) {
1514         int maxrecs = client_get_maxrecs(cl);
1515         int hits = (int) client_get_hits(cl);
1516         term_factor = MAX(hits, maxrecs) /  MAX(1, maxrecs);
1517         assert(term_factor >= 1);
1518         yaz_log(YLOG_DEBUG, "Using term factor: %d (%d / %d)", term_factor, MAX(hits, maxrecs), MAX(1, maxrecs));
1519     }
1520
1521     if (!cluster)
1522         return -1;
1523     if (global_parameters.dump_records)
1524         session_log(se, YLOG_LOG, "Cluster id %s from %s (#%d)", cluster->recid,
1525                     sdb->database->id, record_no);
1526     relevance_newrec(se->relevance, cluster);
1527
1528     // now parsing XML record and adding data to cluster or record metadata
1529     for (n = root->children; n; n = n->next)
1530     {
1531         pp2_charset_token_t prt;
1532         if (type)
1533             xmlFree(type);
1534         if (value)
1535             xmlFree(value);
1536         type = value = 0;
1537
1538         if (n->type != XML_ELEMENT_NODE)
1539             continue;
1540         if (!strcmp((const char *) n->name, "metadata"))
1541         {
1542             struct conf_metadata *ser_md = 0;
1543             struct conf_sortkey *ser_sk = 0;
1544             struct record_metadata **wheretoput = 0;
1545             struct record_metadata *rec_md = 0;
1546             int md_field_id = -1;
1547             int sk_field_id = -1;
1548
1549             type = xmlGetProp(n, (xmlChar *) "type");
1550             value = xmlNodeListGetString(xdoc, n->children, 1);
1551
1552             if (!type || !value || !*value)
1553                 continue;
1554
1555             md_field_id
1556                 = conf_service_metadata_field_id(service, (const char *) type);
1557             if (md_field_id < 0)
1558             {
1559                 if (se->number_of_warnings_unknown_metadata == 0)
1560                 {
1561                     session_log(se, YLOG_WARN,
1562                             "Ignoring unknown metadata element: %s", type);
1563                 }
1564                 se->number_of_warnings_unknown_metadata++;
1565                 continue;
1566             }
1567
1568             ser_md = &service->metadata[md_field_id];
1569
1570             if (ser_md->sortkey_offset >= 0){
1571                 sk_field_id = ser_md->sortkey_offset;
1572                 ser_sk = &service->sortkeys[sk_field_id];
1573             }
1574
1575             // non-merged metadata
1576             rec_md = record_metadata_init(se->nmem, (const char *) value,
1577                                           ser_md->type, n->properties);
1578             if (!rec_md)
1579             {
1580                 session_log(se, YLOG_WARN, "bad metadata data '%s' "
1581                             "for element '%s'", value, type);
1582                 continue;
1583             }
1584             wheretoput = &record->metadata[md_field_id];
1585             while (*wheretoput)
1586                 wheretoput = &(*wheretoput)->next;
1587             *wheretoput = rec_md;
1588
1589             // merged metadata
1590             rec_md = record_metadata_init(se->nmem, (const char *) value,
1591                                           ser_md->type, 0);
1592             wheretoput = &cluster->metadata[md_field_id];
1593
1594             // and polulate with data:
1595             // assign cluster or record based on merge action
1596             if (ser_md->merge == Metadata_merge_unique)
1597             {
1598                 while (*wheretoput)
1599                 {
1600                     if (!strcmp((const char *) (*wheretoput)->data.text.disp,
1601                                 rec_md->data.text.disp))
1602                         break;
1603                     wheretoput = &(*wheretoput)->next;
1604                 }
1605                 if (!*wheretoput)
1606                     *wheretoput = rec_md;
1607             }
1608             else if (ser_md->merge == Metadata_merge_longest)
1609             {
1610                 if (!*wheretoput
1611                     || strlen(rec_md->data.text.disp)
1612                     > strlen((*wheretoput)->data.text.disp))
1613                 {
1614                     *wheretoput = rec_md;
1615                     if (ser_sk)
1616                     {
1617                         const char *sort_str = 0;
1618                         int skip_article =
1619                             ser_sk->type == Metadata_sortkey_skiparticle;
1620
1621                         if (!cluster->sortkeys[sk_field_id])
1622                             cluster->sortkeys[sk_field_id] =
1623                                 nmem_malloc(se->nmem,
1624                                             sizeof(union data_types));
1625
1626                         prt =
1627                             pp2_charset_token_create(service->charsets, "sort");
1628
1629                         pp2_charset_token_first(prt, rec_md->data.text.disp,
1630                                                 skip_article);
1631
1632                         pp2_charset_token_next(prt);
1633
1634                         sort_str = pp2_get_sort(prt);
1635
1636                         cluster->sortkeys[sk_field_id]->text.disp =
1637                             rec_md->data.text.disp;
1638                         if (!sort_str)
1639                         {
1640                             sort_str = rec_md->data.text.disp;
1641                             session_log(se, YLOG_WARN,
1642                                     "Could not make sortkey. Bug #1858");
1643                         }
1644                         cluster->sortkeys[sk_field_id]->text.sort =
1645                             nmem_strdup(se->nmem, sort_str);
1646                         pp2_charset_token_destroy(prt);
1647                     }
1648                 }
1649             }
1650             else if (ser_md->merge == Metadata_merge_all)
1651             {
1652                 while (*wheretoput)
1653                     wheretoput = &(*wheretoput)->next;
1654                 *wheretoput = rec_md;
1655             }
1656             else if (ser_md->merge == Metadata_merge_range)
1657             {
1658                 if (!*wheretoput)
1659                 {
1660                     *wheretoput = rec_md;
1661                     if (ser_sk)
1662                         cluster->sortkeys[sk_field_id]
1663                             = &rec_md->data;
1664                 }
1665                 else
1666                 {
1667                     int this_min = rec_md->data.number.min;
1668                     int this_max = rec_md->data.number.max;
1669                     if (this_min < (*wheretoput)->data.number.min)
1670                         (*wheretoput)->data.number.min = this_min;
1671                     if (this_max > (*wheretoput)->data.number.max)
1672                         (*wheretoput)->data.number.max = this_max;
1673                 }
1674             }
1675
1676
1677             // ranking of _all_ fields enabled ...
1678             if (ser_md->rank)
1679                 relevance_countwords(se->relevance, cluster,
1680                                      (char *) value, ser_md->rank,
1681                                      ser_md->name);
1682
1683             // construct facets ... unless the client already has reported them
1684             if (ser_md->termlist && !client_has_facet(cl, (char *) type))
1685             {
1686                 if (ser_md->type == Metadata_type_year)
1687                 {
1688                     char year[64];
1689                     sprintf(year, "%d", rec_md->data.number.max);
1690
1691                     add_facet(se, (char *) type, year, term_factor);
1692                     if (rec_md->data.number.max != rec_md->data.number.min)
1693                     {
1694                         sprintf(year, "%d", rec_md->data.number.min);
1695                         add_facet(se, (char *) type, year, term_factor);
1696                     }
1697                 }
1698                 else
1699                     add_facet(se, (char *) type, (char *) value, term_factor);
1700             }
1701
1702             // cleaning up
1703             xmlFree(type);
1704             xmlFree(value);
1705             type = value = 0;
1706         }
1707         else
1708         {
1709             if (se->number_of_warnings_unknown_elements == 0)
1710                 session_log(se, YLOG_WARN,
1711                         "Unexpected element in internal record: %s", n->name);
1712             se->number_of_warnings_unknown_elements++;
1713         }
1714     }
1715     if (type)
1716         xmlFree(type);
1717     if (value)
1718         xmlFree(value);
1719
1720     relevance_donerecord(se->relevance, cluster);
1721     se->total_records++;
1722
1723     return 0;
1724 }
1725
1726 void session_log(struct session *s, int level, const char *fmt, ...)
1727 {
1728     char buf[1024];
1729     va_list ap;
1730     va_start(ap, fmt);
1731
1732     yaz_vsnprintf(buf, sizeof(buf)-30, fmt, ap);
1733     yaz_log(level, "Session (%u): %s", s->session_id, buf);
1734
1735     va_end(ap);
1736 }
1737
1738 /*
1739  * Local variables:
1740  * c-basic-offset: 4
1741  * c-file-style: "Stroustrup"
1742  * indent-tabs-mode: nil
1743  * End:
1744  * vim: shiftwidth=4 tabstop=8 expandtab
1745  */
1746