src/session.c

   1 /* This file is part of Pazpar2.
   2    Copyright (C) 2006-2013 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
  60 #define USE_TIMING 1
  61 #if USE_TIMING
  62 #include <yaz/timing.h>
  63 #endif
  64
  65 #include "ppmutex.h"
  66 #include "parameters.h"
  67 #include "session.h"
  68 #include "eventl.h"
  69 #include "http.h"
  70 #include "termlists.h"
  71 #include "reclists.h"
  72 #include "relevance.h"
  73 #include "database.h"
  74 #include "client.h"
  75 #include "settings.h"
  76 #include "normalize7bit.h"
  77
  78 #include <libxml/tree.h>
  79
  80 #define MAX_CHUNK 15
  81
  82 #define MAX(a,b) ((a)>(b)?(a):(b))
  83
  84 // Note: Some things in this structure will eventually move to configuration
  85 struct parameters global_parameters =
  86 {
  87     0,   // dump_records
  88     0,   // debug_mode
  89     0,   // predictable sessions
  90 };
  91
  92 struct client_list {
  93     struct client *client;
  94     struct client_list *next;
  95 };
  96
  97 /* session counting (1) , disable client counting (0) */
  98 static YAZ_MUTEX g_session_mutex = 0;
  99 static int no_sessions = 0;
 100
 101 static int session_use(int delta)
 102 {
 103     int sessions;
 104     if (!g_session_mutex)
 105         yaz_mutex_create(&g_session_mutex);
 106     yaz_mutex_enter(g_session_mutex);
 107     no_sessions += delta;
 108     sessions = no_sessions;
 109     yaz_mutex_leave(g_session_mutex);
 110     yaz_log(YLOG_DEBUG, "%s sessions=%d", delta == 0 ? "" :
 111             (delta > 0 ? "INC" : "DEC"), no_sessions);
 112     return sessions;
 113 }
 114
 115 int sessions_count(void)
 116 {
 117     return session_use(0);
 118 }
 119
 120 static void log_xml_doc(xmlDoc *doc)
 121 {
 122     FILE *lf = yaz_log_file();
 123     xmlChar *result = 0;
 124     int len = 0;
 125 #if LIBXML_VERSION >= 20600
 126     xmlDocDumpFormatMemory(doc, &result, &len, 1);
 127 #else
 128     xmlDocDumpMemory(doc, &result, &len);
 129 #endif
 130     if (lf && len)
 131     {
 132         (void) fwrite(result, 1, len, lf);
 133         fprintf(lf, "\n");
 134     }
 135     xmlFree(result);
 136 }
 137
 138 static void session_enter(struct session *s, const char *caller)
 139 {
 140     if (caller)
 141         session_log(s, YLOG_DEBUG, "Session lock by %s", caller);
 142     yaz_mutex_enter(s->session_mutex);
 143 }
 144
 145 static void session_leave(struct session *s, const char *caller)
 146 {
 147     yaz_mutex_leave(s->session_mutex);
 148     if (caller)
 149         session_log(s, YLOG_DEBUG, "Session unlock by %s", caller);
 150 }
 151
 152 static void session_normalize_facet(struct session *s,
 153                                     const char *type, const char *value,
 154                                     WRBUF display_wrbuf, WRBUF facet_wrbuf)
 155 {
 156     struct conf_service *service = s->service;
 157     pp2_charset_token_t prt;
 158     const char *facet_component;
 159     int i;
 160     const char *icu_chain_id = 0;
 161
 162     for (i = 0; i < service->num_metadata; i++)
 163         if (!strcmp((service->metadata + i)->name, type))
 164             icu_chain_id = (service->metadata + i)->facetrule;
 165     if (!icu_chain_id)
 166         icu_chain_id = "facet";
 167     prt = pp2_charset_token_create(service->charsets, icu_chain_id);
 168     if (!prt)
 169     {
 170         session_log(s, YLOG_FATAL,
 171                     "Unknown ICU chain '%s' for facet of type '%s'",
 172                 icu_chain_id, type);
 173         wrbuf_destroy(facet_wrbuf);
 174         wrbuf_destroy(display_wrbuf);
 175         return;
 176     }
 177     pp2_charset_token_first(prt, value, 0);
 178     while ((facet_component = pp2_charset_token_next(prt)))
 179     {
 180         const char *display_component;
 181         if (*facet_component)
 182         {
 183             if (wrbuf_len(facet_wrbuf))
 184                 wrbuf_puts(facet_wrbuf, " ");
 185             wrbuf_puts(facet_wrbuf, facet_component);
 186         }
 187         display_component = pp2_get_display(prt);
 188         if (display_component)
 189         {
 190             if (wrbuf_len(display_wrbuf))
 191                 wrbuf_puts(display_wrbuf, " ");
 192             wrbuf_puts(display_wrbuf, display_component);
 193         }
 194     }
 195     pp2_charset_token_destroy(prt);
 196 }
 197
 198 void add_facet(struct session *s, const char *type, const char *value, int count)
 199 {
 200     WRBUF facet_wrbuf = wrbuf_alloc();
 201     WRBUF display_wrbuf = wrbuf_alloc();
 202
 203     session_normalize_facet(s, type, value, display_wrbuf, facet_wrbuf);
 204
 205     if (wrbuf_len(facet_wrbuf))
 206     {
 207         int i;
 208         for (i = 0; i < s->num_termlists; i++)
 209             if (!strcmp(s->termlists[i].name, type))
 210                 break;
 211         if (i == s->num_termlists)
 212         {
 213             if (i == SESSION_MAX_TERMLISTS)
 214             {
 215                 session_log(s, YLOG_FATAL, "Too many termlists");
 216                 wrbuf_destroy(facet_wrbuf);
 217                 wrbuf_destroy(display_wrbuf);
 218                 return;
 219             }
 220
 221             s->termlists[i].name = nmem_strdup(s->nmem, type);
 222             s->termlists[i].termlist = termlist_create(s->nmem);
 223             s->num_termlists = i + 1;
 224         }
 225
 226 #if 0
 227         session_log(s, YLOG_LOG, "Facets for %s: %s norm:%s (%d)", type, value, wrbuf_cstr(facet_wrbuf), count);
 228 #endif
 229         termlist_insert(s->termlists[i].termlist, wrbuf_cstr(display_wrbuf),
 230                         wrbuf_cstr(facet_wrbuf), count);
 231     }
 232     wrbuf_destroy(facet_wrbuf);
 233     wrbuf_destroy(display_wrbuf);
 234 }
 235
 236 static xmlDoc *record_to_xml(struct session *se,
 237                              struct session_database *sdb, const char *rec)
 238 {
 239     struct database *db = sdb->database;
 240     xmlDoc *rdoc = 0;
 241
 242     rdoc = xmlParseMemory(rec, strlen(rec));
 243
 244     if (!rdoc)
 245     {
 246         session_log(se, YLOG_WARN, "Non-wellformed XML");
 247         return 0;
 248     }
 249
 250     if (global_parameters.dump_records)
 251     {
 252         session_log(se, YLOG_LOG, "Un-normalized record from %s", db->id);
 253         log_xml_doc(rdoc);
 254     }
 255
 256     return rdoc;
 257 }
 258
 259 #define MAX_XSLT_ARGS 16
 260
 261 // Add static values from session database settings if applicable
 262 static void insert_settings_parameters(struct session_database *sdb,
 263                                        struct conf_service *service,
 264                                        char **parms,
 265                                        NMEM nmem)
 266 {
 267     int i;
 268     int nparms = 0;
 269     int offset = 0;
 270
 271     for (i = 0; i < service->num_metadata; i++)
 272     {
 273         struct conf_metadata *md = &service->metadata[i];
 274         int setting;
 275
 276         if (md->setting == Metadata_setting_parameter &&
 277             (setting = settings_lookup_offset(service, md->name)) >= 0)
 278         {
 279             const char *val = session_setting_oneval(sdb, setting);
 280             if (val && nparms < MAX_XSLT_ARGS)
 281             {
 282                 char *buf;
 283                 int len = strlen(val);
 284                 buf = nmem_malloc(nmem, len + 3);
 285                 buf[0] = '\'';
 286                 strcpy(buf + 1, val);
 287                 buf[len+1] = '\'';
 288                 buf[len+2] = '\0';
 289                 parms[offset++] = md->name;
 290                 parms[offset++] = buf;
 291                 nparms++;
 292             }
 293         }
 294     }
 295     parms[offset] = 0;
 296 }
 297
 298 // Add static values from session database settings if applicable
 299 static void insert_settings_values(struct session_database *sdb, xmlDoc *doc,
 300     struct conf_service *service)
 301 {
 302     int i;
 303
 304     for (i = 0; i < service->num_metadata; i++)
 305     {
 306         struct conf_metadata *md = &service->metadata[i];
 307         int offset;
 308
 309         if (md->setting == Metadata_setting_postproc &&
 310             (offset = settings_lookup_offset(service, md->name)) >= 0)
 311         {
 312             const char *val = session_setting_oneval(sdb, offset);
 313             if (val)
 314             {
 315                 xmlNode *r = xmlDocGetRootElement(doc);
 316                 xmlNode *n = xmlNewTextChild(r, 0, (xmlChar *) "metadata",
 317                                              (xmlChar *) val);
 318                 xmlSetProp(n, (xmlChar *) "type", (xmlChar *) md->name);
 319             }
 320         }
 321     }
 322 }
 323
 324 static xmlDoc *normalize_record(struct session *se,
 325                                 struct session_database *sdb,
 326                                 struct conf_service *service,
 327                                 const char *rec, NMEM nmem)
 328 {
 329     xmlDoc *rdoc = record_to_xml(se, sdb, rec);
 330
 331     if (rdoc)
 332     {
 333         char *parms[MAX_XSLT_ARGS*2+1];
 334
 335         insert_settings_parameters(sdb, service, parms, nmem);
 336
 337         if (normalize_record_transform(sdb->map, &rdoc, (const char **)parms))
 338         {
 339             session_log(se, YLOG_WARN, "Normalize failed");
 340         }
 341         else
 342         {
 343             insert_settings_values(sdb, rdoc, service);
 344
 345             if (global_parameters.dump_records)
 346             {
 347                 session_log(se, YLOG_LOG, "Normalized record from %s",
 348                             sdb->database->id);
 349                 log_xml_doc(rdoc);
 350             }
 351         }
 352     }
 353     return rdoc;
 354 }
 355
 356 void session_settings_dump(struct session *se,
 357                            struct session_database *db,
 358                            WRBUF w)
 359 {
 360     if (db->settings)
 361     {
 362         int i, num = db->num_settings;
 363         for (i = 0; i < num; i++)
 364         {
 365             struct setting *s = db->settings[i];
 366             for (;s ; s = s->next)
 367             {
 368                 wrbuf_puts(w, "<set name=\"");
 369                 wrbuf_xmlputs(w, s->name);
 370                 wrbuf_puts(w, "\" value=\"");
 371                 wrbuf_xmlputs(w, s->value);
 372                 wrbuf_puts(w, "\"/>");
 373             }
 374             if (db->settings[i])
 375                 wrbuf_puts(w, "\n");
 376         }
 377     }
 378 }
 379
 380 // Retrieve first defined value for 'name' for given database.
 381 // Will be extended to take into account user associated with session
 382 const char *session_setting_oneval(struct session_database *db, int offset)
 383 {
 384     if (offset >= db->num_settings || !db->settings[offset])
 385         return "";
 386     return db->settings[offset]->value;
 387 }
 388
 389 // Prepare XSLT stylesheets for record normalization
 390 // Structures are allocated on the session_wide nmem to avoid having
 391 // to recompute this for every search. This would lead
 392 // to leaking if a single session was to repeatedly change the PZ_XSLT
 393 // setting. However, this is not a realistic use scenario.
 394 static int prepare_map(struct session *se, struct session_database *sdb)
 395 {
 396     if (sdb->settings && !sdb->map)
 397     {
 398         const char *s;
 399
 400         if (sdb->settings[PZ_XSLT] &&
 401             (s = session_setting_oneval(sdb, PZ_XSLT)))
 402         {
 403             char auto_stylesheet[256];
 404
 405             if (!strcmp(s, "auto"))
 406             {
 407                 const char *request_syntax = session_setting_oneval(
 408                     sdb, PZ_REQUESTSYNTAX);
 409                 if (request_syntax)
 410                 {
 411                     char *cp;
 412                     yaz_snprintf(auto_stylesheet, sizeof(auto_stylesheet),
 413                                  "%s.xsl", request_syntax);
 414                     for (cp = auto_stylesheet; *cp; cp++)
 415                     {
 416                         /* deliberately only consider ASCII */
 417                         if (*cp > 32 && *cp < 127)
 418                             *cp = tolower(*cp);
 419                     }
 420                     s = auto_stylesheet;
 421                 }
 422                 else
 423                 {
 424                     session_log(se, YLOG_WARN,
 425                                 "No pz:requestsyntax for auto stylesheet");
 426                 }
 427             }
 428             sdb->map = normalize_cache_get(se->normalize_cache,
 429                                            se->service, s);
 430             if (!sdb->map)
 431                 return -1;
 432         }
 433     }
 434     return 0;
 435 }
 436
 437 // called if watch should be removed because http_channel is to be destroyed
 438 static void session_watch_cancel(void *data, struct http_channel *c,
 439                                  void *data2)
 440 {
 441     struct session_watchentry *ent = data;
 442
 443     ent->fun = 0;
 444     ent->data = 0;
 445     ent->obs = 0;
 446 }
 447
 448 // set watch. Returns 0=OK, -1 if watch is already set
 449 int session_set_watch(struct session *s, int what,
 450                       session_watchfun fun, void *data,
 451                       struct http_channel *chan)
 452 {
 453     int ret;
 454     session_enter(s, "session_set_watch");
 455     if (s->watchlist[what].fun)
 456         ret = -1;
 457     else
 458     {
 459
 460         s->watchlist[what].fun = fun;
 461         s->watchlist[what].data = data;
 462         s->watchlist[what].obs = http_add_observer(chan, &s->watchlist[what],
 463                                                    session_watch_cancel);
 464         ret = 0;
 465     }
 466     session_leave(s, "session_set_watch");
 467     return ret;
 468 }
 469
 470 void session_alert_watch(struct session *s, int what)
 471 {
 472     assert(s);
 473     session_enter(s, "session_alert_watch");
 474     if (s->watchlist[what].fun)
 475     {
 476         /* our watch is no longer associated with http_channel */
 477         void *data;
 478         session_watchfun fun;
 479
 480         http_remove_observer(s->watchlist[what].obs);
 481         fun  = s->watchlist[what].fun;
 482         data = s->watchlist[what].data;
 483
 484         /* reset watch before fun is invoked - in case fun wants to set
 485            it again */
 486         s->watchlist[what].fun = 0;
 487         s->watchlist[what].data = 0;
 488         s->watchlist[what].obs = 0;
 489
 490         session_leave(s, "session_alert_watch");
 491         session_log(s, YLOG_DEBUG,
 492                     "Alert Watch: %d calling function: %p", what, fun);
 493         fun(data);
 494     }
 495     else
 496         session_leave(s,"session_alert_watch");
 497 }
 498
 499 //callback for grep_databases
 500 static void select_targets_callback(struct session *se,
 501                                     struct session_database *db)
 502 {
 503     struct client *cl;
 504     struct client_list *l;
 505
 506     for (l = se->clients_cached; l; l = l->next)
 507         if (client_get_database(l->client) == db)
 508             break;
 509
 510     if (l)
 511         cl = l->client;
 512     else
 513     {
 514         cl = client_create(db->database->id);
 515         client_set_database(cl, db);
 516
 517         l = xmalloc(sizeof(*l));
 518         l->client = cl;
 519         l->next = se->clients_cached;
 520         se->clients_cached = l;
 521     }
 522     /* set session always. If may be 0 if client is not active */
 523     client_set_session(cl, se);
 524
 525     l = xmalloc(sizeof(*l));
 526     l->client = cl;
 527     l->next = se->clients_active;
 528     se->clients_active = l;
 529 }
 530
 531 static void session_reset_active_clients(struct session *se,
 532                                          struct client_list *new_list)
 533 {
 534     struct client_list *l;
 535
 536     session_enter(se, "session_reset_active_clients");
 537     l = se->clients_active;
 538     se->clients_active = new_list;
 539     session_leave(se, "session_reset_active_clients");
 540
 541     while (l)
 542     {
 543         struct client_list *l_next = l->next;
 544
 545         client_lock(l->client);
 546         client_set_session(l->client, 0); /* mark client inactive */
 547         client_unlock(l->client);
 548
 549         xfree(l);
 550         l = l_next;
 551     }
 552 }
 553
 554 static void session_remove_cached_clients(struct session *se)
 555 {
 556     struct client_list *l;
 557
 558     session_reset_active_clients(se, 0);
 559
 560     session_enter(se, "session_remove_cached_clients");
 561     l = se->clients_cached;
 562     se->clients_cached = 0;
 563     session_leave(se, "session_remove_cached_clients");
 564
 565     while (l)
 566     {
 567         struct client_list *l_next = l->next;
 568         client_lock(l->client);
 569         client_set_session(l->client, 0);
 570         client_set_database(l->client, 0);
 571         client_unlock(l->client);
 572         client_destroy(l->client);
 573         xfree(l);
 574         l = l_next;
 575     }
 576 }
 577
 578 // Associates a set of clients with a session;
 579 // Note: Session-databases represent databases with per-session
 580 // setting overrides
 581 static int select_targets(struct session *se, const char *filter)
 582 {
 583     return session_grep_databases(se, filter, select_targets_callback);
 584 }
 585
 586 int session_active_clients(struct session *s)
 587 {
 588     struct client_list *l;
 589     int res = 0;
 590
 591     for (l = s->clients_active; l; l = l->next)
 592         if (client_is_active(l->client))
 593             res++;
 594
 595     return res;
 596 }
 597
 598 int session_is_preferred_clients_ready(struct session *s)
 599 {
 600     struct client_list *l;
 601     int res = 0;
 602
 603     for (l = s->clients_active; l; l = l->next)
 604         if (client_is_active_preferred(l->client))
 605             res++;
 606     session_log(s, YLOG_DEBUG, "Has %d active preferred clients.", res);
 607     return res == 0;
 608 }
 609
 610 static void session_clear_set(struct session *se, struct reclist_sortparms *sp)
 611 {
 612     reclist_destroy(se->reclist);
 613     if (nmem_total(se->nmem))
 614         session_log(se, YLOG_DEBUG, "NMEN operation usage %zd",
 615                     nmem_total(se->nmem));
 616     nmem_reset(se->nmem);
 617     se->total_records = se->total_merged = 0;
 618     se->num_termlists = 0;
 619
 620     relevance_clear(se->relevance);
 621
 622     /* reset list of sorted results and clear to relevance search */
 623     se->sorted_results = nmem_malloc(se->nmem, sizeof(*se->sorted_results));
 624     se->sorted_results->name = nmem_strdup(se->nmem, sp->name);
 625     se->sorted_results->increasing = sp->increasing;
 626     se->sorted_results->type = sp->type;
 627     se->sorted_results->next = 0;
 628
 629     session_log(se, YLOG_DEBUG, "clear_set session_sort: field=%s increasing=%d type=%d configured",
 630             sp->name, sp->increasing, sp->type);
 631
 632     se->reclist = reclist_create(se->nmem);
 633 }
 634
 635 static void session_sort_unlocked(struct session *se,
 636                                   struct reclist_sortparms *sp,
 637                                   const char *mergekey,
 638                                   const char *rank)
 639 {
 640     struct client_list *l;
 641     const char *field = sp->name;
 642     int increasing = sp->increasing;
 643     int type  = sp->type;
 644     int clients_research = 0;
 645
 646     session_log(se, YLOG_DEBUG, "session_sort field=%s increasing=%d type=%d",
 647                 field, increasing, type);
 648
 649     if (rank && (!se->rank || strcmp(se->rank, rank)))
 650     {
 651         /* new rank must research/reingest anyway */
 652         assert(rank);
 653         xfree(se->rank);
 654         se->rank = *rank ? xstrdup(rank) : 0;
 655         clients_research = 1;
 656         session_log(se, YLOG_DEBUG, "session_sort: new rank = %s",
 657                     rank);
 658     }
 659     if (mergekey && (!se->mergekey || strcmp(se->mergekey, mergekey)))
 660     {
 661         /* new mergekey must research/reingest anyway */
 662         assert(mergekey);
 663         xfree(se->mergekey);
 664         se->mergekey = *mergekey ? xstrdup(mergekey) : 0;
 665         clients_research = 1;
 666         session_log(se, YLOG_DEBUG, "session_sort: new mergekey = %s",
 667                     mergekey);
 668     }
 669     if (clients_research == 0)
 670     {
 671         struct reclist_sortparms *sr;
 672         for (sr = se->sorted_results; sr; sr = sr->next)
 673             if (!reclist_sortparms_cmp(sr, sp))
 674                 break;
 675         if (sr)
 676         {
 677             session_log(se, YLOG_DEBUG, "session_sort: field=%s increasing=%d type=%d already fetched",
 678                         field, increasing, type);
 679             return;
 680         }
 681     }
 682     session_log(se, YLOG_DEBUG, "session_sort: field=%s increasing=%d type=%d must fetch",
 683                 field, increasing, type);
 684
 685     // We need to reset reclist on every sort that changes the records, not just for position
 686     // So if just one client requires new searching, we need to clear set.
 687     // Ask each of the client if sorting requires re-search due to native sort
 688     // If it does it will require us to
 689     for (l = se->clients_active; l; l = l->next)
 690     {
 691         struct client *cl = l->client;
 692         // Assume no re-search is required.
 693         client_parse_init(cl, 1);
 694         clients_research += client_parse_sort(cl, sp);
 695     }
 696     if (clients_research)
 697     {
 698         session_log(se, YLOG_DEBUG,
 699                     "session_sort: reset results due to %d clients researching",
 700                     clients_research);
 701         session_clear_set(se, sp);
 702     }
 703     else
 704     {
 705         // A new sorting based on same record set
 706         struct reclist_sortparms *sr = nmem_malloc(se->nmem, sizeof(*sr));
 707         sr->name = nmem_strdup(se->nmem, field);
 708         sr->increasing = increasing;
 709         sr->type = type;
 710         sr->next = se->sorted_results;
 711         se->sorted_results = sr;
 712         session_log(se, YLOG_DEBUG, "session_sort: no research/ingesting done");
 713         return ;
 714     }
 715     session_log(se, YLOG_DEBUG, "Re- search/ingesting for clients due to change in sort order");
 716
 717     for (l = se->clients_active; l; l = l->next)
 718     {
 719         struct client *cl = l->client;
 720         if (client_get_state(cl) == Client_Connecting ||
 721             client_get_state(cl) == Client_Idle ||
 722             client_get_state(cl) == Client_Working) {
 723             client_start_search(cl);
 724         }
 725         else
 726         {
 727             session_log(se, YLOG_DEBUG,
 728                         "session_sort: %s: No re-start/ingest in show. "
 729                         "Wrong client state: %d",
 730                         client_get_id(cl), client_get_state(cl));
 731         }
 732     }
 733 }
 734
 735 void session_sort(struct session *se, struct reclist_sortparms *sp,
 736                   const char *mergekey, const char *rank)
 737 {
 738     //session_enter(se, "session_sort");
 739     session_sort_unlocked(se, sp, mergekey, rank);
 740     //session_leave(se, "session_sort");
 741 }
 742
 743
 744 enum pazpar2_error_code session_search(struct session *se,
 745                                        const char *query,
 746                                        const char *startrecs,
 747                                        const char *maxrecs,
 748                                        const char *filter,
 749                                        const char *limit,
 750                                        const char **addinfo,
 751                                        struct reclist_sortparms *sp,
 752                                        const char *mergekey,
 753                                        const char *rank)
 754 {
 755     int live_channels = 0;
 756     int no_working = 0;
 757     int no_failed_query = 0;
 758     int no_failed_limit = 0;
 759     struct client_list *l, *l0;
 760
 761     session_alert_watch(se, SESSION_WATCH_SHOW);
 762     session_alert_watch(se, SESSION_WATCH_BYTARGET);
 763     session_alert_watch(se, SESSION_WATCH_TERMLIST);
 764     session_alert_watch(se, SESSION_WATCH_SHOW_PREF);
 765
 766     session_log(se, YLOG_DEBUG, "Search");
 767
 768     *addinfo = 0;
 769
 770     if (se->settings_modified) {
 771         session_remove_cached_clients(se);
 772     }
 773     else
 774         session_reset_active_clients(se, 0);
 775
 776     session_enter(se, "session_search");
 777     se->settings_modified = 0;
 778
 779     if (mergekey)
 780     {
 781         xfree(se->mergekey);
 782         se->mergekey = *mergekey ? xstrdup(mergekey) : 0;
 783     }
 784     if (rank)
 785     {
 786         xfree(se->rank);
 787         se->rank = *rank ? xstrdup(rank) : 0;
 788     }
 789
 790     session_clear_set(se, sp);
 791     relevance_destroy(&se->relevance);
 792
 793     live_channels = select_targets(se, filter);
 794     if (!live_channels)
 795     {
 796         session_leave(se, "session_search");
 797         return PAZPAR2_NO_TARGETS;
 798     }
 799
 800     facet_limits_destroy(se->facet_limits);
 801     se->facet_limits = facet_limits_create(limit);
 802     if (!se->facet_limits)
 803     {
 804         *addinfo = "limit";
 805         session_leave(se, "session_search");
 806         return PAZPAR2_MALFORMED_PARAMETER_VALUE;
 807     }
 808
 809     l0 = se->clients_active;
 810     se->clients_active = 0;
 811     session_leave(se, "session_search");
 812
 813     for (l = l0; l; l = l->next)
 814     {
 815         int parse_ret;
 816         struct client *cl = l->client;
 817         client_parse_init(cl, 1);
 818         if (prepare_map(se, client_get_database(cl)) < 0)
 819             continue;
 820
 821         parse_ret = client_parse_query(cl, query, se->facet_limits);
 822         if (parse_ret == -1)
 823             no_failed_query++;
 824         else if (parse_ret == -2)
 825             no_failed_limit++;
 826         else if (parse_ret < 0)
 827             no_working++; /* other error, such as bad CCL map */
 828         else
 829         {
 830             client_parse_range(cl, startrecs, maxrecs);
 831             client_parse_sort(cl, sp);
 832             client_start_search(cl);
 833             no_working++;
 834         }
 835     }
 836     session_reset_active_clients(se, l0);
 837
 838     if (no_working == 0)
 839     {
 840         if (no_failed_query > 0)
 841         {
 842             *addinfo = "query";
 843             return PAZPAR2_MALFORMED_PARAMETER_VALUE;
 844         }
 845         else if (no_failed_limit > 0)
 846         {
 847             *addinfo = "limit";
 848             return PAZPAR2_MALFORMED_PARAMETER_VALUE;
 849         }
 850         else
 851             return PAZPAR2_NO_TARGETS;
 852     }
 853     return PAZPAR2_NO_ERROR;
 854 }
 855
 856 // Creates a new session_database object for a database
 857 static void session_init_databases_fun(void *context, struct database *db)
 858 {
 859     struct session *se = (struct session *) context;
 860     struct session_database *new = nmem_malloc(se->session_nmem, sizeof(*new));
 861     int i;
 862
 863     new->database = db;
 864
 865     new->map = 0;
 866     assert(db->settings);
 867     new->settings = nmem_malloc(se->session_nmem,
 868                                 sizeof(struct settings *) * db->num_settings);
 869     new->num_settings = db->num_settings;
 870     for (i = 0; i < db->num_settings; i++)
 871     {
 872         struct setting *setting = db->settings[i];
 873         new->settings[i] = setting;
 874     }
 875     new->next = se->databases;
 876     se->databases = new;
 877 }
 878
 879 // Doesn't free memory associated with sdb -- nmem takes care of that
 880 static void session_database_destroy(struct session_database *sdb)
 881 {
 882     sdb->map = 0;
 883 }
 884
 885 // Initialize session_database list -- this represents this session's view
 886 // of the database list -- subject to modification by the settings ws command
 887 void session_init_databases(struct session *se)
 888 {
 889     se->databases = 0;
 890     predef_grep_databases(se, se->service, session_init_databases_fun);
 891 }
 892
 893 // Probably session_init_databases_fun should be refactored instead of
 894 // called here.
 895 static struct session_database *load_session_database(struct session *se,
 896                                                       char *id)
 897 {
 898     struct database *db = new_database_inherit_settings(id, se->session_nmem, se->service->settings);
 899     session_init_databases_fun((void*) se, db);
 900
 901     // New sdb is head of se->databases list
 902     return se->databases;
 903 }
 904
 905 // Find an existing session database. If not found, load it
 906 static struct session_database *find_session_database(struct session *se,
 907                                                       char *id)
 908 {
 909     struct session_database *sdb;
 910
 911     for (sdb = se->databases; sdb; sdb = sdb->next)
 912         if (!strcmp(sdb->database->id, id))
 913             return sdb;
 914     return load_session_database(se, id);
 915 }
 916
 917 // Apply a session override to a database
 918 void session_apply_setting(struct session *se, char *dbname, char *setting,
 919                            char *value)
 920 {
 921     struct session_database *sdb = find_session_database(se, dbname);
 922     struct conf_service *service = se->service;
 923     struct setting *new = nmem_malloc(se->session_nmem, sizeof(*new));
 924     int offset = settings_create_offset(service, setting);
 925
 926     expand_settings_array(&sdb->settings, &sdb->num_settings, offset,
 927                           se->session_nmem);
 928     new->precedence = 0;
 929     new->target = dbname;
 930     new->name = setting;
 931     new->value = value;
 932     new->next = sdb->settings[offset];
 933     sdb->settings[offset] = new;
 934
 935     se->settings_modified = 1;
 936
 937     // Force later recompute of settings-driven data structures
 938     // (happens when a search starts and client connections are prepared)
 939     switch (offset)
 940     {
 941     case PZ_XSLT:
 942         if (sdb->map)
 943         {
 944             sdb->map = 0;
 945         }
 946         break;
 947     }
 948 }
 949
 950 void session_destroy(struct session *se)
 951 {
 952     struct session_database *sdb;
 953     session_log(se, YLOG_DEBUG, "Destroying");
 954     session_use(-1);
 955     session_remove_cached_clients(se);
 956
 957     for (sdb = se->databases; sdb; sdb = sdb->next)
 958         session_database_destroy(sdb);
 959     normalize_cache_destroy(se->normalize_cache);
 960     relevance_destroy(&se->relevance);
 961     reclist_destroy(se->reclist);
 962     xfree(se->mergekey);
 963     xfree(se->rank);
 964     if (nmem_total(se->nmem))
 965         session_log(se, YLOG_DEBUG, "NMEN operation usage %zd", nmem_total(se->nmem));
 966     if (nmem_total(se->session_nmem))
 967         session_log(se, YLOG_DEBUG, "NMEN session usage %zd", nmem_total(se->session_nmem));
 968     facet_limits_destroy(se->facet_limits);
 969     nmem_destroy(se->nmem);
 970     service_destroy(se->service);
 971     yaz_mutex_destroy(&se->session_mutex);
 972 }
 973
 974 size_t session_get_memory_status(struct session *session) {
 975     size_t session_nmem;
 976     if (session == 0)
 977         return 0;
 978     session_enter(session, "session_get_memory_status");
 979     session_nmem = nmem_total(session->nmem);
 980     session_leave(session, "session_get_memory_status");
 981     return session_nmem;
 982 }
 983
 984
 985 struct session *new_session(NMEM nmem, struct conf_service *service,
 986                             unsigned session_id)
 987 {
 988     int i;
 989     struct session *session = nmem_malloc(nmem, sizeof(*session));
 990
 991     char tmp_str[50];
 992
 993     sprintf(tmp_str, "session#%u", session_id);
 994
 995     session->session_id = session_id;
 996     session_log(session, YLOG_DEBUG, "New");
 997     session->service = service;
 998     session->relevance = 0;
 999     session->total_records = 0;
1000     session->number_of_warnings_unknown_elements = 0;
1001     session->number_of_warnings_unknown_metadata = 0;
1002     session->num_termlists = 0;
1003     session->reclist = reclist_create(nmem);
1004     session->clients_active = 0;
1005     session->clients_cached = 0;
1006     session->settings_modified = 0;
1007     session->session_nmem = nmem;
1008     session->nmem = nmem_create();
1009     session->databases = 0;
1010     session->sorted_results = 0;
1011     session->facet_limits = 0;
1012     session->mergekey = 0;
1013     session->rank = 0;
1014
1015     for (i = 0; i <= SESSION_WATCH_MAX; i++)
1016     {
1017         session->watchlist[i].data = 0;
1018         session->watchlist[i].fun = 0;
1019     }
1020     session->normalize_cache = normalize_cache_create();
1021     session->session_mutex = 0;
1022     pazpar2_mutex_create(&session->session_mutex, tmp_str);
1023     session_use(1);
1024     return session;
1025 }
1026
1027 const char * client_get_suggestions_xml(struct client *cl, WRBUF wrbuf);
1028
1029 static struct hitsbytarget *hitsbytarget_nb(struct session *se,
1030                                             int *count, NMEM nmem)
1031 {
1032     struct hitsbytarget *res = 0;
1033     struct client_list *l;
1034     size_t sz = 0;
1035
1036     for (l = se->clients_active; l; l = l->next)
1037         sz++;
1038
1039     res = nmem_malloc(nmem, sizeof(*res) * sz);
1040     *count = 0;
1041     for (l = se->clients_active; l; l = l->next)
1042     {
1043         struct client *cl = l->client;
1044         WRBUF w = wrbuf_alloc();
1045         const char *name = session_setting_oneval(client_get_database(cl),
1046                                                   PZ_NAME);
1047
1048         res[*count].id = client_get_id(cl);
1049         res[*count].name = *name ? name : "Unknown";
1050         res[*count].hits = client_get_hits(cl);
1051         res[*count].approximation = client_get_approximation(cl);
1052         res[*count].records = client_get_num_records(cl);
1053         res[*count].filtered = client_get_num_records_filtered(cl);
1054         res[*count].diagnostic =
1055             client_get_diagnostic(cl, &res[*count].message,
1056                                   &res[*count].addinfo);
1057         res[*count].state = client_get_state_str(cl);
1058         res[*count].connected  = client_get_connection(cl) ? 1 : 0;
1059         session_settings_dump(se, client_get_database(cl), w);
1060         res[*count].settings_xml = nmem_strdup(nmem, wrbuf_cstr(w));
1061         wrbuf_rewind(w);
1062         wrbuf_puts(w, "");
1063         res[*count].suggestions_xml = nmem_strdup(nmem, client_get_suggestions_xml(cl, w));
1064         wrbuf_destroy(w);
1065         (*count)++;
1066     }
1067     return res;
1068 }
1069
1070 struct hitsbytarget *get_hitsbytarget(struct session *se, int *count, NMEM nmem)
1071 {
1072     struct hitsbytarget *p;
1073     session_enter(se, "get_hitsbytarget");
1074     p = hitsbytarget_nb(se, count, nmem);
1075     session_leave(se, "get_hitsbytarget");
1076     return p;
1077 }
1078
1079 // Compares two hitsbytarget nodes by hitcount
1080 static int cmp_ht(const void *p1, const void *p2)
1081 {
1082     const struct hitsbytarget *h1 = p1;
1083     const struct hitsbytarget *h2 = p2;
1084     return h2->hits - h1->hits;
1085 }
1086
1087 // Compares two hitsbytarget nodes by hitcount
1088 static int cmp_ht_approx(const void *p1, const void *p2)
1089 {
1090     const struct hitsbytarget *h1 = p1;
1091     const struct hitsbytarget *h2 = p2;
1092     return h2->approximation - h1->approximation;
1093 }
1094
1095 static int targets_termlist_nb(WRBUF wrbuf, struct session *se, int num,
1096                                NMEM nmem, int version)
1097 {
1098     struct hitsbytarget *ht;
1099     int count, i;
1100
1101     ht = hitsbytarget_nb(se, &count, nmem);
1102     if (version >= 2)
1103         qsort(ht, count, sizeof(struct hitsbytarget), cmp_ht_approx);
1104     else
1105         qsort(ht, count, sizeof(struct hitsbytarget), cmp_ht);
1106     for (i = 0; i < count && i < num && ht[i].hits > 0; i++)
1107     {
1108
1109         // do only print terms which have display names
1110
1111         wrbuf_puts(wrbuf, "<term>\n");
1112
1113         wrbuf_puts(wrbuf, "<id>");
1114         wrbuf_xmlputs(wrbuf, ht[i].id);
1115         wrbuf_puts(wrbuf, "</id>\n");
1116
1117         wrbuf_puts(wrbuf, "<name>");
1118         if (!ht[i].name || !ht[i].name[0])
1119             wrbuf_xmlputs(wrbuf, "NO TARGET NAME");
1120         else
1121             wrbuf_xmlputs(wrbuf, ht[i].name);
1122         wrbuf_puts(wrbuf, "</name>\n");
1123
1124         wrbuf_printf(wrbuf, "<frequency>" ODR_INT_PRINTF "</frequency>\n",
1125                      ht[i].hits);
1126
1127         if (version >= 2) {
1128             // Should not print if we know it isn't a approximation.
1129             wrbuf_printf(wrbuf, "<approximation>" ODR_INT_PRINTF "</approximation>\n", ht[i].approximation);
1130             wrbuf_printf(wrbuf, "<records>%d</records>\n", ht[i].records - ht[i].filtered);
1131             wrbuf_printf(wrbuf, "<filtered>%d</filtered>\n", ht[i].filtered);
1132         }
1133
1134         wrbuf_puts(wrbuf, "<state>");
1135         wrbuf_xmlputs(wrbuf, ht[i].state);
1136         wrbuf_puts(wrbuf, "</state>\n");
1137
1138         wrbuf_printf(wrbuf, "<diagnostic>%d</diagnostic>\n",
1139                      ht[i].diagnostic);
1140         wrbuf_puts(wrbuf, "</term>\n");
1141     }
1142     return count;
1143 }
1144
1145 void perform_termlist(struct http_channel *c, struct session *se,
1146                       const char *name, int num, int version)
1147 {
1148     int i, j;
1149     NMEM nmem_tmp = nmem_create();
1150     char **names;
1151     int num_names = 0;
1152
1153     if (!name)
1154         name = "*";
1155
1156     nmem_strsplit(nmem_tmp, ",", name, &names, &num_names);
1157
1158     session_enter(se, "perform_termlist");
1159
1160     for (j = 0; j < num_names; j++)
1161     {
1162         const char *tname;
1163         int must_generate_empty = 1; /* bug 5350 */
1164
1165         for (i = 0; i < se->num_termlists; i++)
1166         {
1167             tname = se->termlists[i].name;
1168             if (!strcmp(names[j], tname) || !strcmp(names[j], "*"))
1169             {
1170                 struct termlist_score **p = 0;
1171                 int len;
1172
1173                 wrbuf_puts(c->wrbuf, "<list name=\"");
1174                 wrbuf_xmlputs(c->wrbuf, tname);
1175                 wrbuf_puts(c->wrbuf, "\">\n");
1176                 must_generate_empty = 0;
1177
1178                 p = termlist_highscore(se->termlists[i].termlist, &len,
1179                                        nmem_tmp);
1180                 if (p)
1181                 {
1182                     int i;
1183                     for (i = 0; i < len && i < num; i++)
1184                     {
1185                         // prevent sending empty term elements
1186                         if (!p[i]->display_term || !p[i]->display_term[0])
1187                             continue;
1188
1189                         wrbuf_puts(c->wrbuf, "<term>");
1190                         wrbuf_puts(c->wrbuf, "<name>");
1191                         wrbuf_xmlputs(c->wrbuf, p[i]->display_term);
1192                         wrbuf_puts(c->wrbuf, "</name>");
1193
1194                         wrbuf_printf(c->wrbuf,
1195                                      "<frequency>%d</frequency>",
1196                                      p[i]->frequency);
1197                         wrbuf_puts(c->wrbuf, "</term>\n");
1198                     }
1199                 }
1200                 wrbuf_puts(c->wrbuf, "</list>\n");
1201             }
1202         }
1203         tname = "xtargets";
1204         if (!strcmp(names[j], tname) || !strcmp(names[j], "*"))
1205         {
1206             wrbuf_puts(c->wrbuf, "<list name=\"");
1207             wrbuf_xmlputs(c->wrbuf, tname);
1208             wrbuf_puts(c->wrbuf, "\">\n");
1209
1210             targets_termlist_nb(c->wrbuf, se, num, c->nmem, version);
1211             wrbuf_puts(c->wrbuf, "</list>\n");
1212             must_generate_empty = 0;
1213         }
1214         if (must_generate_empty)
1215         {
1216             wrbuf_puts(c->wrbuf, "<list name=\"");
1217             wrbuf_xmlputs(c->wrbuf, names[j]);
1218             wrbuf_puts(c->wrbuf, "\"/>\n");
1219         }
1220     }
1221     session_leave(se, "perform_termlist");
1222     nmem_destroy(nmem_tmp);
1223 }
1224
1225 #ifdef MISSING_HEADERS
1226 void report_nmem_stats(void)
1227 {
1228     size_t in_use, is_free;
1229
1230     nmem_get_memory_in_use(&in_use);
1231     nmem_get_memory_free(&is_free);
1232
1233     yaz_log(YLOG_LOG, "nmem stat: use=%ld free=%ld",
1234             (long) in_use, (long) is_free);
1235 }
1236 #endif
1237
1238 struct record_cluster *show_single_start(struct session *se, const char *id,
1239                                          struct record_cluster **prev_r,
1240                                          struct record_cluster **next_r)
1241 {
1242     struct record_cluster *r = 0;
1243
1244     session_enter(se, "show_single_start");
1245     *prev_r = 0;
1246     *next_r = 0;
1247     reclist_limit(se->reclist, se, 1);
1248
1249     reclist_enter(se->reclist);
1250     while ((r = reclist_read_record(se->reclist)))
1251     {
1252         if (!strcmp(r->recid, id))
1253         {
1254             *next_r = reclist_read_record(se->reclist);
1255             break;
1256         }
1257         *prev_r = r;
1258     }
1259     reclist_leave(se->reclist);
1260     if (!r)
1261         session_leave(se, "show_single_start");
1262     return r;
1263 }
1264
1265 void show_single_stop(struct session *se, struct record_cluster *rec)
1266 {
1267     session_leave(se, "show_single_stop");
1268 }
1269
1270
1271 int session_fetch_more(struct session *se)
1272 {
1273     struct client_list *l;
1274     int ret = 0;
1275
1276     for (l = se->clients_active; l; l = l->next)
1277     {
1278         struct client *cl = l->client;
1279         if (client_get_state(cl) == Client_Idle)
1280         {
1281             if (client_fetch_more(cl))
1282             {
1283                 session_log(se, YLOG_LOG, "%s: more to fetch",
1284                             client_get_id(cl));
1285                 ret = 1;
1286             }
1287             else
1288             {
1289                 session_log(se, YLOG_LOG, "%s: no more to fetch",
1290                             client_get_id(cl));
1291             }
1292         }
1293         else
1294         {
1295             session_log(se, YLOG_LOG, "%s: no fetch due to state=%s",
1296                         client_get_id(cl), client_get_state_str(cl));
1297         }
1298
1299     }
1300     return ret;
1301 }
1302
1303 struct record_cluster **show_range_start(struct session *se,
1304                                          struct reclist_sortparms *sp,
1305                                          int start, int *num, int *total,
1306                                          Odr_int *sumhits, Odr_int *approx_hits,
1307                                          void (*show_records_ready)(void *data),
1308                                          struct http_channel *chan)
1309 {
1310     struct record_cluster **recs = 0;
1311     struct reclist_sortparms *spp;
1312     struct client_list *l;
1313     int i;
1314 #if USE_TIMING
1315     yaz_timing_t t = yaz_timing_create();
1316 #endif
1317     session_enter(se, "show_range_start");
1318     *sumhits = 0;
1319     *approx_hits = 0;
1320     *total = 0;
1321     reclist_limit(se->reclist, se, 0);
1322     if (se->relevance)
1323     {
1324         for (spp = sp; spp; spp = spp->next)
1325             if (spp->type == Metadata_sortkey_relevance)
1326             {
1327                 relevance_prepare_read(se->relevance, se->reclist);
1328                 break;
1329             }
1330         for (l = se->clients_active; l; l = l->next) {
1331             *sumhits += client_get_hits(l->client);
1332             *approx_hits += client_get_approximation(l->client);
1333         }
1334     }
1335     reclist_sort(se->reclist, sp);
1336
1337     reclist_enter(se->reclist);
1338     *total = reclist_get_num_records(se->reclist);
1339
1340     for (l = se->clients_active; l; l = l->next)
1341         client_update_show_stat(l->client, 0);
1342
1343     for (i = 0; i < start; i++)
1344     {
1345         struct record_cluster *r = reclist_read_record(se->reclist);
1346         if (!r)
1347         {
1348             *num = 0;
1349             break;
1350         }
1351         else
1352         {
1353             struct record *rec = r->records;
1354             for (;rec; rec = rec->next)
1355                 client_update_show_stat(rec->client, 1);
1356         }
1357     }
1358     recs = nmem_malloc(se->nmem, (*num > 0 ? *num : 1) * sizeof(*recs));
1359     for (i = 0; i < *num; i++)
1360     {
1361         struct record_cluster *r = reclist_read_record(se->reclist);
1362         if (!r)
1363         {
1364             *num = i;
1365             break;
1366         }
1367         else
1368         {
1369             struct record *rec = r->records;
1370             for (;rec; rec = rec->next)
1371                 client_update_show_stat(rec->client, 1);
1372             recs[i] = r;
1373         }
1374     }
1375     reclist_leave(se->reclist);
1376 #if USE_TIMING
1377     yaz_timing_stop(t);
1378     session_log(se, YLOG_LOG, "show %6.5f %3.2f %3.2f",
1379             yaz_timing_get_real(t), yaz_timing_get_user(t),
1380             yaz_timing_get_sys(t));
1381     yaz_timing_destroy(&t);
1382 #endif
1383
1384     if (!session_fetch_more(se))
1385         session_log(se, YLOG_LOG, "can not fetch more");
1386     else
1387     {
1388         show_range_stop(se, recs);
1389         session_log(se, YLOG_LOG, "fetching more in progress");
1390         if (session_set_watch(se, SESSION_WATCH_SHOW,
1391                               show_records_ready, chan, chan))
1392         {
1393             session_log(se, YLOG_WARN, "Ignoring show block");
1394             session_enter(se, "show_range_start");
1395         }
1396         else
1397         {
1398             session_log(se, YLOG_LOG, "session watch OK");
1399             return 0;
1400         }
1401     }
1402     return recs;
1403 }
1404
1405 void show_range_stop(struct session *se, struct record_cluster **recs)
1406 {
1407     session_leave(se, "show_range_stop");
1408 }
1409
1410 void statistics(struct session *se, struct statistics *stat)
1411 {
1412     struct client_list *l;
1413     int count = 0;
1414
1415     memset(stat, 0, sizeof(*stat));
1416     stat->num_hits = 0;
1417     for (l = se->clients_active; l; l = l->next)
1418     {
1419         struct client *cl = l->client;
1420         if (!client_get_connection(cl))
1421             stat->num_no_connection++;
1422         stat->num_hits += client_get_hits(cl);
1423         switch (client_get_state(cl))
1424         {
1425         case Client_Connecting: stat->num_connecting++; break;
1426         case Client_Working: stat->num_working++; break;
1427         case Client_Idle: stat->num_idle++; break;
1428         case Client_Failed: stat->num_failed++; break;
1429         case Client_Error: stat->num_error++; break;
1430         default: break;
1431         }
1432         count++;
1433     }
1434     stat->num_records = se->total_records;
1435
1436     stat->num_clients = count;
1437 }
1438
1439 static struct record_metadata *record_metadata_init(
1440     NMEM nmem, const char *value, enum conf_metadata_type type,
1441     struct _xmlAttr *attr)
1442 {
1443     struct record_metadata *rec_md = record_metadata_create(nmem);
1444     struct record_metadata_attr **attrp = &rec_md->attributes;
1445
1446     for (; attr; attr = attr->next)
1447     {
1448         if (attr->children && attr->children->content)
1449         {
1450             if (strcmp((const char *) attr->name, "type")
1451                 && strcmp((const char *) attr->name, "empty"))
1452             {  /* skip the "type" + "empty" attribute..
1453                   The "Type" is already part of the element in output
1454                   (md-%s) and so repeating it here is redundant */
1455                 *attrp = nmem_malloc(nmem, sizeof(**attrp));
1456                 (*attrp)->name =
1457                     nmem_strdup(nmem, (const char *) attr->name);
1458                 (*attrp)->value =
1459                     nmem_strdup(nmem, (const char *) attr->children->content);
1460                 attrp = &(*attrp)->next;
1461             }
1462         }
1463     }
1464     *attrp = 0;
1465
1466     if (type == Metadata_type_generic)
1467     {
1468         char *p = nmem_strdup(nmem, value);
1469
1470         p = normalize7bit_generic(p, " ,/.:([");
1471
1472         rec_md->data.text.disp = p;
1473         rec_md->data.text.sort = 0;
1474         rec_md->data.text.snippet = 0;
1475     }
1476     else if (type == Metadata_type_year || type == Metadata_type_date)
1477     {
1478         int first, last;
1479         int longdate = 0;
1480
1481         if (type == Metadata_type_date)
1482             longdate = 1;
1483         if (extract7bit_dates((char *) value, &first, &last, longdate) < 0)
1484             return 0;
1485
1486         rec_md->data.number.min = first;
1487         rec_md->data.number.max = last;
1488     }
1489     else
1490         return 0;
1491     return rec_md;
1492 }
1493
1494 static void mergekey_norm_wr(pp2_charset_fact_t charsets,
1495                              WRBUF norm_wr, const char *value)
1496 {
1497     const char *norm_str;
1498     pp2_charset_token_t prt =
1499         pp2_charset_token_create(charsets, "mergekey");
1500
1501     pp2_charset_token_first(prt, value, 0);
1502     while ((norm_str = pp2_charset_token_next(prt)))
1503     {
1504         if (*norm_str)
1505         {
1506             if (wrbuf_len(norm_wr))
1507                 wrbuf_puts(norm_wr, " ");
1508             wrbuf_puts(norm_wr, norm_str);
1509         }
1510     }
1511     pp2_charset_token_destroy(prt);
1512 }
1513
1514 static int get_mergekey_from_doc(xmlDoc *doc, xmlNode *root, const char *name,
1515                                  struct conf_service *service, WRBUF norm_wr)
1516 {
1517     xmlNode *n;
1518     int no_found = 0;
1519     for (n = root->children; n; n = n->next)
1520     {
1521         if (n->type != XML_ELEMENT_NODE)
1522             continue;
1523         if (!strcmp((const char *) n->name, "metadata"))
1524         {
1525             xmlChar *type = xmlGetProp(n, (xmlChar *) "type");
1526             if (type == NULL) {
1527                 yaz_log(YLOG_FATAL, "Missing type attribute on metadata element. Skipping!");
1528             }
1529             else if (!strcmp(name, (const char *) type))
1530             {
1531                 xmlChar *value = xmlNodeListGetString(doc, n->children, 1);
1532                 if (value && *value)
1533                 {
1534                     if (wrbuf_len(norm_wr) > 0)
1535                         wrbuf_puts(norm_wr, " ");
1536                     wrbuf_puts(norm_wr, name);
1537                     mergekey_norm_wr(service->charsets, norm_wr,
1538                                      (const char *) value);
1539                     no_found++;
1540                 }
1541                 if (value)
1542                     xmlFree(value);
1543             }
1544             xmlFree(type);
1545         }
1546     }
1547     return no_found;
1548 }
1549
1550 static const char *get_mergekey(xmlDoc *doc, struct client *cl, int record_no,
1551                                 struct conf_service *service, NMEM nmem,
1552                                 const char *session_mergekey)
1553 {
1554     char *mergekey_norm = 0;
1555     xmlNode *root = xmlDocGetRootElement(doc);
1556     WRBUF norm_wr = wrbuf_alloc();
1557     xmlChar *mergekey;
1558
1559     if (session_mergekey)
1560     {
1561         int i, num = 0;
1562         char **values = 0;
1563         nmem_strsplit_escape2(nmem, ",", session_mergekey, &values,
1564                               &num, 1, '\\', 1);
1565
1566         for (i = 0; i < num; i++)
1567             get_mergekey_from_doc(doc, root, values[i], service, norm_wr);
1568     }
1569     else if ((mergekey = xmlGetProp(root, (xmlChar *) "mergekey")))
1570     {
1571         mergekey_norm_wr(service->charsets, norm_wr, (const char *) mergekey);
1572         xmlFree(mergekey);
1573     }
1574     else
1575     {
1576         /* no mergekey defined in XSL. Look for mergekey metadata instead */
1577         int field_id;
1578         for (field_id = 0; field_id < service->num_metadata; field_id++)
1579         {
1580             struct conf_metadata *ser_md = &service->metadata[field_id];
1581             if (ser_md->mergekey != Metadata_mergekey_no)
1582             {
1583                 int r = get_mergekey_from_doc(doc, root, ser_md->name,
1584                                               service, norm_wr);
1585                 if (r == 0 && ser_md->mergekey == Metadata_mergekey_required)
1586                 {
1587                     /* no mergekey on this one and it is required..
1588                        Generate unique key instead */
1589                     wrbuf_rewind(norm_wr);
1590                     break;
1591                 }
1592             }
1593         }
1594     }
1595
1596     /* generate unique key if none is not generated already or is empty */
1597     if (wrbuf_len(norm_wr) == 0)
1598     {
1599         wrbuf_printf(norm_wr, "position: %s-%d",
1600                      client_get_id(cl), record_no);
1601     }
1602     else
1603     {
1604         const char *lead = "content: ";
1605         wrbuf_insert(norm_wr, 0, lead, strlen(lead));
1606     }
1607     if (wrbuf_len(norm_wr) > 0)
1608         mergekey_norm = nmem_strdup(nmem, wrbuf_cstr(norm_wr));
1609     wrbuf_destroy(norm_wr);
1610     return mergekey_norm;
1611 }
1612
1613 /** \brief see if metadata for pz:recordfilter exists
1614     \param root xml root element of normalized record
1615     \param sdb session database for client
1616     \retval 0 if there is no metadata for pz:recordfilter
1617     \retval 1 if there is metadata for pz:recordfilter
1618
1619     If there is no pz:recordfilter defined, this function returns 1
1620     as well.
1621 */
1622
1623 static int check_record_filter(xmlNode *root, struct session_database *sdb)
1624 {
1625     int match = 0;
1626     xmlNode *n;
1627     const char *s;
1628     s = session_setting_oneval(sdb, PZ_RECORDFILTER);
1629
1630     if (!s || !*s)
1631         return 1;
1632
1633     for (n = root->children; n; n = n->next)
1634     {
1635         if (n->type != XML_ELEMENT_NODE)
1636             continue;
1637         if (!strcmp((const char *) n->name, "metadata"))
1638         {
1639             xmlChar *type = xmlGetProp(n, (xmlChar *) "type");
1640             if (type)
1641             {
1642                 size_t len;
1643                 int substring;
1644                 const char *eq;
1645
1646                 if ((eq = strchr(s, '=')))
1647                     substring = 0;
1648                 else if ((eq = strchr(s, '~')))
1649                     substring = 1;
1650                 if (eq)
1651                     len = eq - s;
1652                 else
1653                     len = strlen(s);
1654                 if (len == strlen((const char *)type) &&
1655                     !memcmp((const char *) type, s, len))
1656                 {
1657                     xmlChar *value = xmlNodeGetContent(n);
1658                     if (value && *value)
1659                     {
1660                         if (!eq ||
1661                             (substring && strstr((const char *) value, eq+1)) ||
1662                             (!substring && !strcmp((const char *) value, eq + 1)))
1663                             match = 1;
1664                     }
1665                     xmlFree(value);
1666                 }
1667                 xmlFree(type);
1668             }
1669         }
1670     }
1671     return match;
1672 }
1673
1674
1675 static int ingest_to_cluster(struct client *cl,
1676                              xmlDoc *xdoc,
1677                              xmlNode *root,
1678                              int record_no,
1679                              const char *mergekey_norm);
1680
1681 /** \brief ingest XML record
1682     \param cl client holds the result set for record
1683     \param rec record buffer (0 terminated)
1684     \param record_no record position (1, 2, ..)
1685     \param nmem working NMEM
1686     \retval 0 OK
1687     \retval -1 failure
1688     \retval -2 Filtered
1689 */
1690 int ingest_record(struct client *cl, const char *rec,
1691                   int record_no, NMEM nmem)
1692 {
1693     struct session *se = client_get_session(cl);
1694     int ret = 0;
1695     struct session_database *sdb = client_get_database(cl);
1696     struct conf_service *service = se->service;
1697     xmlDoc *xdoc = normalize_record(se, sdb, service, rec, nmem);
1698     xmlNode *root;
1699     const char *mergekey_norm;
1700
1701     if (!xdoc)
1702         return -1;
1703
1704     root = xmlDocGetRootElement(xdoc);
1705
1706     if (!check_record_filter(root, sdb))
1707     {
1708         session_log(se, YLOG_LOG, "Filtered out record no %d from %s", record_no, sdb->database->id);
1709         xmlFreeDoc(xdoc);
1710         return -2;
1711     }
1712
1713     mergekey_norm = get_mergekey(xdoc, cl, record_no, service, nmem,
1714         se->mergekey);
1715     if (!mergekey_norm)
1716     {
1717         session_log(se, YLOG_WARN, "Got no mergekey");
1718         xmlFreeDoc(xdoc);
1719         return -1;
1720     }
1721     session_enter(se, "ingest_record");
1722     if (client_get_session(cl) == se && se->relevance)
1723         ret = ingest_to_cluster(cl, xdoc, root, record_no, mergekey_norm);
1724     session_leave(se, "ingest_record");
1725
1726     xmlFreeDoc(xdoc);
1727     return ret;
1728 }
1729
1730 //    struct conf_metadata *ser_md = &service->metadata[md_field_id];
1731 //    struct record_metadata *rec_md = record->metadata[md_field_id];
1732 static int match_metadata_local(struct conf_service *service,
1733                                 struct conf_metadata *ser_md,
1734                                 struct record_metadata *rec_md0,
1735                                 char **values, int num_v)
1736 {
1737     int i;
1738     struct record_metadata *rec_md = rec_md0;
1739     WRBUF val_wr = 0;
1740     WRBUF text_wr = wrbuf_alloc();
1741     for (i = 0; i < num_v; )
1742     {
1743         if (rec_md)
1744         {
1745             if (ser_md->type == Metadata_type_year
1746                 || ser_md->type == Metadata_type_date)
1747             {
1748                 int y = atoi(values[i]);
1749                 if (y >= rec_md->data.number.min
1750                     && y <= rec_md->data.number.max)
1751                     break;
1752             }
1753             else
1754             {
1755                 if (!val_wr)
1756                 {
1757                     val_wr = wrbuf_alloc();
1758                     mergekey_norm_wr(service->charsets, val_wr, values[i]);
1759                 }
1760                 wrbuf_rewind(text_wr);
1761                 mergekey_norm_wr(service->charsets, text_wr,
1762                                  rec_md->data.text.disp);
1763                 if (!strcmp(wrbuf_cstr(val_wr), wrbuf_cstr(text_wr)))
1764                     break;
1765             }
1766             rec_md = rec_md->next;
1767         }
1768         else
1769         {
1770             rec_md = rec_md0;
1771             wrbuf_destroy(val_wr);
1772             val_wr = 0;
1773             i++;
1774         }
1775     }
1776     wrbuf_destroy(val_wr);
1777     wrbuf_destroy(text_wr);
1778     return i < num_v ? 1 : 0;
1779 }
1780
1781 int session_check_cluster_limit(struct session *se, struct record_cluster *rec)
1782 {
1783     int i;
1784     struct conf_service *service = se->service;
1785     int ret = 1;
1786     const char *name;
1787     const char *value;
1788     NMEM nmem_tmp = nmem_create();
1789
1790     for (i = 0; (name = facet_limits_get(se->facet_limits, i, &value)); i++)
1791     {
1792         int j;
1793         for (j = 0; j < service->num_metadata; j++)
1794         {
1795             struct conf_metadata *md = service->metadata + j;
1796             if (!strcmp(md->name, name) && md->limitcluster)
1797             {
1798                 char **values = 0;
1799                 int num = 0;
1800                 int md_field_id =
1801                     conf_service_metadata_field_id(service,
1802                                                    md->limitcluster);
1803
1804                 if (md_field_id < 0)
1805                 {
1806                     ret = 0;
1807                     break;
1808                 }
1809
1810                 nmem_strsplit_escape2(nmem_tmp, "|", value, &values,
1811                                       &num, 1, '\\', 1);
1812
1813                 if (!match_metadata_local(service,
1814                                           &service->metadata[md_field_id],
1815                                           rec->metadata[md_field_id],
1816                                           values, num))
1817                 {
1818                     ret = 0;
1819                     break;
1820                 }
1821             }
1822         }
1823     }
1824     nmem_destroy(nmem_tmp);
1825     return ret;
1826 }
1827
1828 // Skip record on non-zero
1829 static int check_limit_local(struct client *cl,
1830                              struct record *record,
1831                              int record_no)
1832 {
1833     int skip_record = 0;
1834     struct session *se = client_get_session(cl);
1835     struct conf_service *service = se->service;
1836     NMEM nmem_tmp = nmem_create();
1837     struct session_database *sdb = client_get_database(cl);
1838     int l = 0;
1839     while (!skip_record)
1840     {
1841         int md_field_id;
1842         char **values = 0;
1843         int num_v = 0;
1844         const char *name =
1845             client_get_facet_limit_local(cl, sdb, &l, nmem_tmp,
1846                                          &num_v, &values);
1847         if (!name)
1848             break;
1849
1850         if (!strcmp(name, "*"))
1851         {
1852             for (md_field_id = 0; md_field_id < service->num_metadata;
1853                  md_field_id++)
1854             {
1855                 if (match_metadata_local(
1856                         service,
1857                         &service->metadata[md_field_id],
1858                         record->metadata[md_field_id],
1859                         values, num_v))
1860                     break;
1861             }
1862             if (md_field_id == service->num_metadata)
1863                 skip_record = 1;
1864         }
1865         else
1866         {
1867             md_field_id = conf_service_metadata_field_id(service, name);
1868             if (md_field_id < 0)
1869             {
1870                 skip_record = 1;
1871                 break;
1872             }
1873             if (!match_metadata_local(
1874                     service,
1875                     &service->metadata[md_field_id],
1876                     record->metadata[md_field_id],
1877                     values, num_v))
1878             {
1879                 skip_record = 1;
1880             }
1881         }
1882     }
1883     nmem_destroy(nmem_tmp);
1884     return skip_record;
1885 }
1886
1887 static int ingest_to_cluster(struct client *cl,
1888                              xmlDoc *xdoc,
1889                              xmlNode *root,
1890                              int record_no,
1891                              const char *mergekey_norm)
1892 {
1893     xmlNode *n;
1894     xmlChar *type = 0;
1895     xmlChar *value = 0;
1896     struct session *se = client_get_session(cl);
1897     struct conf_service *service = se->service;
1898     int term_factor = 1;
1899     struct record_cluster *cluster;
1900     struct record_metadata **metadata0;
1901     struct session_database *sdb = client_get_database(cl);
1902     NMEM ingest_nmem = 0;
1903     char **rank_values = 0;
1904     int rank_num = 0;
1905     struct record *record = record_create(se->nmem,
1906                                           service->num_metadata,
1907                                           service->num_sortkeys, cl,
1908                                           record_no);
1909
1910     for (n = root->children; n; n = n->next)
1911     {
1912         if (type)
1913             xmlFree(type);
1914         if (value)
1915             xmlFree(value);
1916         type = value = 0;
1917
1918         if (n->type != XML_ELEMENT_NODE)
1919             continue;
1920         if (!strcmp((const char *) n->name, "metadata"))
1921         {
1922             struct conf_metadata *ser_md = 0;
1923             struct record_metadata **wheretoput = 0;
1924             struct record_metadata *rec_md = 0;
1925             int md_field_id = -1;
1926
1927             type = xmlGetProp(n, (xmlChar *) "type");
1928             value = xmlNodeListGetString(xdoc, n->children, 1);
1929             if (!type)
1930                 continue;
1931             if (!value || !*value)
1932             {
1933                 xmlChar *empty = xmlGetProp(n, (xmlChar *) "empty");
1934                 if (!empty)
1935                     continue;
1936                 if (value)
1937                     xmlFree(value);
1938                 value = empty;
1939             }
1940             md_field_id
1941                 = conf_service_metadata_field_id(service, (const char *) type);
1942             if (md_field_id < 0)
1943             {
1944                 if (se->number_of_warnings_unknown_metadata == 0)
1945                 {
1946                     session_log(se, YLOG_WARN,
1947                             "Ignoring unknown metadata element: %s", type);
1948                 }
1949                 se->number_of_warnings_unknown_metadata++;
1950                 continue;
1951             }
1952
1953             ser_md = &service->metadata[md_field_id];
1954
1955             // non-merged metadata
1956             rec_md = record_metadata_init(se->nmem, (const char *) value,
1957                                           ser_md->type, n->properties);
1958             if (!rec_md)
1959             {
1960                 session_log(se, YLOG_WARN, "bad metadata data '%s' "
1961                             "for element '%s'", value, type);
1962                 continue;
1963             }
1964
1965             if (ser_md->type == Metadata_type_generic)
1966             {
1967                 WRBUF w = wrbuf_alloc();
1968                 if (relevance_snippet(se->relevance,
1969                                       (char*) value, ser_md->name, w))
1970                     rec_md->data.text.snippet = nmem_strdup(se->nmem,
1971                                                             wrbuf_cstr(w));
1972                 wrbuf_destroy(w);
1973             }
1974
1975
1976             wheretoput = &record->metadata[md_field_id];
1977             while (*wheretoput)
1978                 wheretoput = &(*wheretoput)->next;
1979             *wheretoput = rec_md;
1980         }
1981     }
1982
1983     if (check_limit_local(cl, record, record_no))
1984     {
1985         session_log(se, YLOG_LOG, "Facet filtered out record no %d from %s",
1986                     record_no, sdb->database->id);
1987         if (type)
1988             xmlFree(type);
1989         if (value)
1990             xmlFree(value);
1991         return -2;
1992     }
1993     cluster = reclist_insert(se->reclist, service, record,
1994                              mergekey_norm, &se->total_merged);
1995     if (!cluster)
1996         return 0; // complete match with existing record
1997
1998     {
1999         const char *use_term_factor_str =
2000             session_setting_oneval(sdb, PZ_TERMLIST_TERM_FACTOR);
2001         if (use_term_factor_str && use_term_factor_str[0] == '1')
2002         {
2003             int maxrecs = client_get_maxrecs(cl);
2004             int hits = (int) client_get_hits(cl);
2005             term_factor = MAX(hits, maxrecs) /  MAX(1, maxrecs);
2006             assert(term_factor >= 1);
2007             session_log(se, YLOG_DEBUG, "Using term factor: %d (%d / %d)",
2008                         term_factor, MAX(hits, maxrecs), MAX(1, maxrecs));
2009         }
2010     }
2011
2012     if (global_parameters.dump_records)
2013         session_log(se, YLOG_LOG, "Cluster id %s from %s (#%d)", cluster->recid,
2014                     sdb->database->id, record_no);
2015
2016
2017     relevance_newrec(se->relevance, cluster);
2018
2019     // original metadata, to check if first existence of a field
2020     metadata0 = xmalloc(sizeof(*metadata0) * service->num_metadata);
2021     memcpy(metadata0, cluster->metadata,
2022            sizeof(*metadata0) * service->num_metadata);
2023
2024     ingest_nmem = nmem_create();
2025     if (se->rank)
2026     {
2027         yaz_log(YLOG_LOG, "local in sort : %s", se->rank);
2028         nmem_strsplit_escape2(ingest_nmem, ",", se->rank, &rank_values,
2029                               &rank_num, 1, '\\', 1);
2030     }
2031
2032     // now parsing XML record and adding data to cluster or record metadata
2033     for (n = root->children; n; n = n->next)
2034     {
2035         pp2_charset_token_t prt;
2036         if (type)
2037             xmlFree(type);
2038         if (value)
2039             xmlFree(value);
2040         type = value = 0;
2041
2042         if (n->type != XML_ELEMENT_NODE)
2043             continue;
2044         if (!strcmp((const char *) n->name, "metadata"))
2045         {
2046             struct conf_metadata *ser_md = 0;
2047             struct conf_sortkey *ser_sk = 0;
2048             struct record_metadata **wheretoput = 0;
2049             struct record_metadata *rec_md = 0;
2050             int md_field_id = -1;
2051             int sk_field_id = -1;
2052             const char *rank = 0;
2053             xmlChar *xml_rank = 0;
2054
2055             type = xmlGetProp(n, (xmlChar *) "type");
2056             value = xmlNodeListGetString(xdoc, n->children, 1);
2057
2058             if (!type || !value || !*value)
2059                 continue;
2060
2061             md_field_id
2062                 = conf_service_metadata_field_id(service, (const char *) type);
2063             if (md_field_id < 0)
2064                 continue;
2065
2066             ser_md = &service->metadata[md_field_id];
2067
2068             if (ser_md->sortkey_offset >= 0)
2069             {
2070                 sk_field_id = ser_md->sortkey_offset;
2071                 ser_sk = &service->sortkeys[sk_field_id];
2072             }
2073
2074             // merged metadata
2075             rec_md = record_metadata_init(se->nmem, (const char *) value,
2076                                           ser_md->type, 0);
2077
2078             // see if the field was not in cluster already (from beginning)
2079
2080             if (!rec_md)
2081                 continue;
2082
2083             if (rank_num)
2084             {
2085                 int i;
2086                 for (i = 0; i < rank_num; i++)
2087                 {
2088                     const char *val = rank_values[i];
2089                     const char *cp = strchr(val, '=');
2090                     if (!cp)
2091                         continue;
2092                     if ((cp - val) == strlen((const char *) type)
2093                         && !memcmp(val, type, cp - val))
2094                     {
2095                         rank = cp + 1;
2096                         break;
2097                     }
2098                 }
2099             }
2100             else
2101             {
2102                 xml_rank = xmlGetProp(n, (xmlChar *) "rank");
2103                 rank = xml_rank ? (const char *) xml_rank : ser_md->rank;
2104             }
2105
2106             wheretoput = &cluster->metadata[md_field_id];
2107
2108             if (ser_md->merge == Metadata_merge_first)
2109             {
2110                 if (!metadata0[md_field_id])
2111                 {
2112                     while (*wheretoput)
2113                         wheretoput = &(*wheretoput)->next;
2114                     *wheretoput = rec_md;
2115                 }
2116             }
2117             else if (ser_md->merge == Metadata_merge_unique)
2118             {
2119                 while (*wheretoput)
2120                 {
2121                     if (!strcmp((const char *) (*wheretoput)->data.text.disp,
2122                                 rec_md->data.text.disp))
2123                         break;
2124                     wheretoput = &(*wheretoput)->next;
2125                 }
2126                 if (!*wheretoput)
2127                     *wheretoput = rec_md;
2128             }
2129             else if (ser_md->merge == Metadata_merge_longest)
2130             {
2131                 if (!*wheretoput
2132                     || strlen(rec_md->data.text.disp)
2133                     > strlen((*wheretoput)->data.text.disp))
2134                 {
2135                     *wheretoput = rec_md;
2136                     if (ser_sk)
2137                     {
2138                         const char *sort_str = 0;
2139                         int skip_article =
2140                             ser_sk->type == Metadata_sortkey_skiparticle;
2141
2142                         if (!cluster->sortkeys[sk_field_id])
2143                             cluster->sortkeys[sk_field_id] =
2144                                 nmem_malloc(se->nmem,
2145                                             sizeof(union data_types));
2146
2147                         prt =
2148                             pp2_charset_token_create(service->charsets, "sort");
2149
2150                         pp2_charset_token_first(prt, rec_md->data.text.disp,
2151                                                 skip_article);
2152
2153                         pp2_charset_token_next(prt);
2154
2155                         sort_str = pp2_get_sort(prt);
2156
2157                         cluster->sortkeys[sk_field_id]->text.disp =
2158                             rec_md->data.text.disp;
2159                         if (!sort_str)
2160                         {
2161                             sort_str = rec_md->data.text.disp;
2162                             session_log(se, YLOG_WARN,
2163                                     "Could not make sortkey. Bug #1858");
2164                         }
2165                         cluster->sortkeys[sk_field_id]->text.sort =
2166                             nmem_strdup(se->nmem, sort_str);
2167                         pp2_charset_token_destroy(prt);
2168                     }
2169                 }
2170             }
2171             else if (ser_md->merge == Metadata_merge_all)
2172             {
2173                 while (*wheretoput)
2174                     wheretoput = &(*wheretoput)->next;
2175                 *wheretoput = rec_md;
2176             }
2177             else if (ser_md->merge == Metadata_merge_range)
2178             {
2179                 if (!*wheretoput)
2180                 {
2181                     *wheretoput = rec_md;
2182                     if (ser_sk)
2183                         cluster->sortkeys[sk_field_id]
2184                             = &rec_md->data;
2185                 }
2186                 else
2187                 {
2188                     int this_min = rec_md->data.number.min;
2189                     int this_max = rec_md->data.number.max;
2190                     if (this_min < (*wheretoput)->data.number.min)
2191                         (*wheretoput)->data.number.min = this_min;
2192                     if (this_max > (*wheretoput)->data.number.max)
2193                         (*wheretoput)->data.number.max = this_max;
2194                 }
2195             }
2196
2197             // ranking of _all_ fields enabled ...
2198             if (rank)
2199             {
2200                 relevance_countwords(se->relevance, cluster,
2201                                      (char *) value, rank, ser_md->name);
2202             }
2203             // construct facets ... unless the client already has reported them
2204             if (ser_md->termlist && !client_has_facet(cl, (char *) type))
2205             {
2206                 if (ser_md->type == Metadata_type_year)
2207                 {
2208                     char year[64];
2209                     sprintf(year, "%d", rec_md->data.number.max);
2210
2211                     add_facet(se, (char *) type, year, term_factor);
2212                     if (rec_md->data.number.max != rec_md->data.number.min)
2213                     {
2214                         sprintf(year, "%d", rec_md->data.number.min);
2215                         add_facet(se, (char *) type, year, term_factor);
2216                     }
2217                 }
2218                 else
2219                     add_facet(se, (char *) type, (char *) value, term_factor);
2220             }
2221
2222             // cleaning up
2223             if (xml_rank)
2224                 xmlFree(xml_rank);
2225             xmlFree(type);
2226             xmlFree(value);
2227             type = value = 0;
2228         }
2229         else
2230         {
2231             if (se->number_of_warnings_unknown_elements == 0)
2232                 session_log(se, YLOG_WARN,
2233                         "Unexpected element in internal record: %s", n->name);
2234             se->number_of_warnings_unknown_elements++;
2235         }
2236     }
2237     if (type)
2238         xmlFree(type);
2239     if (value)
2240         xmlFree(value);
2241
2242     nmem_destroy(ingest_nmem);
2243     xfree(metadata0);
2244     relevance_donerecord(se->relevance, cluster);
2245     se->total_records++;
2246
2247     return 0;
2248 }
2249
2250 void session_log(struct session *s, int level, const char *fmt, ...)
2251 {
2252     char buf[1024];
2253     va_list ap;
2254     va_start(ap, fmt);
2255
2256     yaz_vsnprintf(buf, sizeof(buf)-30, fmt, ap);
2257     yaz_log(level, "Session %u: %s", s ? s->session_id : 0, buf);
2258
2259     va_end(ap);
2260 }
2261
2262 /*
2263  * Local variables:
2264  * c-basic-offset: 4
2265  * c-file-style: "Stroustrup"
2266  * indent-tabs-mode: nil
2267  * End:
2268  * vim: shiftwidth=4 tabstop=8 expandtab
2269  */
2270