src/session.c

   1 /* This file is part of Pazpar2.
   2    Copyright (C) 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 0
  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                                    xmlNode *root,
 301                                    struct conf_service *service)
 302 {
 303     int i;
 304
 305     for (i = 0; i < service->num_metadata; i++)
 306     {
 307         struct conf_metadata *md = &service->metadata[i];
 308         int offset;
 309
 310         if (md->setting == Metadata_setting_postproc &&
 311             (offset = settings_lookup_offset(service, md->name)) >= 0)
 312         {
 313             const char *val = session_setting_oneval(sdb, offset);
 314             if (val)
 315             {
 316                 xmlNode *n = xmlNewTextChild(root, 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     }
 342     return rdoc;
 343 }
 344
 345 void session_settings_dump(struct session *se,
 346                            struct session_database *db,
 347                            WRBUF w)
 348 {
 349     if (db->settings)
 350     {
 351         int i, num = db->num_settings;
 352         for (i = 0; i < num; i++)
 353         {
 354             struct setting *s = db->settings[i];
 355             for (;s ; s = s->next)
 356             {
 357                 wrbuf_puts(w, "<set name=\"");
 358                 wrbuf_xmlputs(w, s->name);
 359                 wrbuf_puts(w, "\" value=\"");
 360                 wrbuf_xmlputs(w, s->value);
 361                 wrbuf_puts(w, "\"/>");
 362             }
 363             if (db->settings[i])
 364                 wrbuf_puts(w, "\n");
 365         }
 366     }
 367 }
 368
 369 // Retrieve first defined value for 'name' for given database.
 370 // Will be extended to take into account user associated with session
 371 const char *session_setting_oneval(struct session_database *db, int offset)
 372 {
 373     if (offset >= db->num_settings || !db->settings[offset])
 374         return "";
 375     return db->settings[offset]->value;
 376 }
 377
 378 // Prepare XSLT stylesheets for record normalization
 379 // Structures are allocated on the session_wide nmem to avoid having
 380 // to recompute this for every search. This would lead
 381 // to leaking if a single session was to repeatedly change the PZ_XSLT
 382 // setting. However, this is not a realistic use scenario.
 383 static int prepare_map(struct session *se, struct session_database *sdb)
 384 {
 385     if (sdb->settings && !sdb->map)
 386     {
 387         const char *s;
 388
 389         if (sdb->settings[PZ_XSLT] &&
 390             (s = session_setting_oneval(sdb, PZ_XSLT)))
 391         {
 392             char auto_stylesheet[256];
 393
 394             if (!strcmp(s, "auto"))
 395             {
 396                 const char *request_syntax = session_setting_oneval(
 397                     sdb, PZ_REQUESTSYNTAX);
 398                 if (request_syntax)
 399                 {
 400                     char *cp;
 401                     yaz_snprintf(auto_stylesheet, sizeof(auto_stylesheet),
 402                                  "%s.xsl", request_syntax);
 403                     for (cp = auto_stylesheet; *cp; cp++)
 404                     {
 405                         /* deliberately only consider ASCII */
 406                         if (*cp > 32 && *cp < 127)
 407                             *cp = tolower(*cp);
 408                     }
 409                     s = auto_stylesheet;
 410                 }
 411                 else
 412                 {
 413                     session_log(se, YLOG_WARN,
 414                                 "No pz:requestsyntax for auto stylesheet");
 415                 }
 416             }
 417             sdb->map = normalize_cache_get(se->normalize_cache,
 418                                            se->service, s);
 419             if (!sdb->map)
 420                 return -1;
 421         }
 422     }
 423     return 0;
 424 }
 425
 426 // called if watch should be removed because http_channel is to be destroyed
 427 static void session_watch_cancel(void *data, struct http_channel *c,
 428                                  void *data2)
 429 {
 430     struct session_watchentry *ent = data;
 431
 432     ent->fun = 0;
 433     ent->data = 0;
 434     ent->obs = 0;
 435 }
 436
 437 // set watch. Returns 0=OK, -1 if watch is already set
 438 int session_set_watch(struct session *s, int what,
 439                       session_watchfun fun, void *data,
 440                       struct http_channel *chan)
 441 {
 442     int ret;
 443     session_enter(s, "session_set_watch");
 444     if (s->watchlist[what].fun)
 445         ret = -1;
 446     else
 447     {
 448
 449         s->watchlist[what].fun = fun;
 450         s->watchlist[what].data = data;
 451         s->watchlist[what].obs = http_add_observer(chan, &s->watchlist[what],
 452                                                    session_watch_cancel);
 453         ret = 0;
 454     }
 455     session_leave(s, "session_set_watch");
 456     return ret;
 457 }
 458
 459 void session_alert_watch(struct session *s, int what)
 460 {
 461     assert(s);
 462     session_enter(s, "session_alert_watch");
 463     if (s->watchlist[what].fun)
 464     {
 465         /* our watch is no longer associated with http_channel */
 466         void *data;
 467         session_watchfun fun;
 468
 469         http_remove_observer(s->watchlist[what].obs);
 470         fun  = s->watchlist[what].fun;
 471         data = s->watchlist[what].data;
 472
 473         /* reset watch before fun is invoked - in case fun wants to set
 474            it again */
 475         s->watchlist[what].fun = 0;
 476         s->watchlist[what].data = 0;
 477         s->watchlist[what].obs = 0;
 478
 479         session_leave(s, "session_alert_watch");
 480         session_log(s, YLOG_DEBUG,
 481                     "Alert Watch: %d calling function: %p", what, fun);
 482         fun(data);
 483     }
 484     else
 485         session_leave(s,"session_alert_watch");
 486 }
 487
 488 //callback for grep_databases
 489 static void select_targets_callback(struct session *se,
 490                                     struct session_database *db)
 491 {
 492     struct client *cl;
 493     struct client_list *l;
 494
 495     for (l = se->clients_cached; l; l = l->next)
 496         if (client_get_database(l->client) == db)
 497             break;
 498
 499     if (l)
 500         cl = l->client;
 501     else
 502     {
 503         cl = client_create(db->database->id);
 504         client_set_database(cl, db);
 505
 506         l = xmalloc(sizeof(*l));
 507         l->client = cl;
 508         l->next = se->clients_cached;
 509         se->clients_cached = l;
 510     }
 511     /* set session always. If may be 0 if client is not active */
 512     client_set_session(cl, se);
 513
 514     l = xmalloc(sizeof(*l));
 515     l->client = cl;
 516     l->next = se->clients_active;
 517     se->clients_active = l;
 518 }
 519
 520 static void session_reset_active_clients(struct session *se,
 521                                          struct client_list *new_list)
 522 {
 523     struct client_list *l;
 524
 525     session_enter(se, "session_reset_active_clients");
 526     l = se->clients_active;
 527     se->clients_active = new_list;
 528     session_leave(se, "session_reset_active_clients");
 529
 530     while (l)
 531     {
 532         struct client_list *l_next = l->next;
 533
 534         client_lock(l->client);
 535         client_set_session(l->client, 0); /* mark client inactive */
 536         client_unlock(l->client);
 537
 538         xfree(l);
 539         l = l_next;
 540     }
 541 }
 542
 543 static void session_remove_cached_clients(struct session *se)
 544 {
 545     struct client_list *l;
 546
 547     session_reset_active_clients(se, 0);
 548
 549     session_enter(se, "session_remove_cached_clients");
 550     l = se->clients_cached;
 551     se->clients_cached = 0;
 552     session_leave(se, "session_remove_cached_clients");
 553
 554     while (l)
 555     {
 556         struct client_list *l_next = l->next;
 557         client_lock(l->client);
 558         client_set_session(l->client, 0);
 559         client_set_database(l->client, 0);
 560         client_unlock(l->client);
 561         client_destroy(l->client);
 562         xfree(l);
 563         l = l_next;
 564     }
 565 }
 566
 567 // Associates a set of clients with a session;
 568 // Note: Session-databases represent databases with per-session
 569 // setting overrides
 570 static int select_targets(struct session *se, const char *filter)
 571 {
 572     return session_grep_databases(se, filter, select_targets_callback);
 573 }
 574
 575 int session_active_clients(struct session *s)
 576 {
 577     struct client_list *l;
 578     int res = 0;
 579
 580     for (l = s->clients_active; l; l = l->next)
 581         if (client_is_active(l->client))
 582             res++;
 583
 584     return res;
 585 }
 586
 587 int session_is_preferred_clients_ready(struct session *s)
 588 {
 589     struct client_list *l;
 590     int res = 0;
 591
 592     for (l = s->clients_active; l; l = l->next)
 593         if (client_is_active_preferred(l->client))
 594             res++;
 595     session_log(s, YLOG_DEBUG, "Has %d active preferred clients.", res);
 596     return res == 0;
 597 }
 598
 599 static void session_clear_set(struct session *se, struct reclist_sortparms *sp)
 600 {
 601     reclist_destroy(se->reclist);
 602     if (nmem_total(se->nmem))
 603         session_log(se, YLOG_DEBUG, "NMEN operation usage %zd",
 604                     nmem_total(se->nmem));
 605     nmem_reset(se->nmem);
 606     se->total_records = se->total_merged = 0;
 607     se->num_termlists = 0;
 608
 609     relevance_clear(se->relevance);
 610
 611     /* reset list of sorted results and clear to relevance search */
 612     se->sorted_results = nmem_malloc(se->nmem, sizeof(*se->sorted_results));
 613     se->sorted_results->name = nmem_strdup(se->nmem, sp->name);
 614     se->sorted_results->increasing = sp->increasing;
 615     se->sorted_results->type = sp->type;
 616     se->sorted_results->next = 0;
 617
 618     session_log(se, YLOG_DEBUG, "clear_set session_sort: field=%s increasing=%d type=%d configured",
 619             sp->name, sp->increasing, sp->type);
 620
 621     se->reclist = reclist_create(se->nmem);
 622 }
 623
 624 void session_sort(struct session *se, struct reclist_sortparms *sp,
 625                   const char *mergekey, const char *rank)
 626 {
 627     struct client_list *l;
 628     const char *field = sp->name;
 629     int increasing = sp->increasing;
 630     int type  = sp->type;
 631     int clients_research = 0;
 632
 633     session_enter(se, "session_sort");
 634     session_log(se, YLOG_DEBUG, "session_sort field=%s increasing=%d type=%d",
 635                 field, increasing, type);
 636
 637     if (rank && (!se->rank || strcmp(se->rank, rank)))
 638     {
 639         /* new rank must research/reingest anyway */
 640         assert(rank);
 641         xfree(se->rank);
 642         se->rank = *rank ? xstrdup(rank) : 0;
 643         clients_research = 1;
 644         session_log(se, YLOG_DEBUG, "session_sort: new rank = %s",
 645                     rank);
 646     }
 647     if (mergekey && (!se->mergekey || strcmp(se->mergekey, mergekey)))
 648     {
 649         /* new mergekey must research/reingest anyway */
 650         assert(mergekey);
 651         xfree(se->mergekey);
 652         se->mergekey = *mergekey ? xstrdup(mergekey) : 0;
 653         clients_research = 1;
 654         session_log(se, YLOG_DEBUG, "session_sort: new mergekey = %s",
 655                     mergekey);
 656     }
 657     if (clients_research == 0)
 658     {
 659         struct reclist_sortparms *sr;
 660         for (sr = se->sorted_results; sr; sr = sr->next)
 661             if (!reclist_sortparms_cmp(sr, sp))
 662                 break;
 663         if (sr)
 664         {
 665             session_log(se, YLOG_DEBUG, "session_sort: field=%s increasing=%d type=%d already fetched",
 666                         field, increasing, type);
 667             session_leave(se, "session_sort");
 668             return;
 669         }
 670     }
 671     session_log(se, YLOG_DEBUG, "session_sort: field=%s increasing=%d type=%d must fetch",
 672                 field, increasing, type);
 673
 674     // We need to reset reclist on every sort that changes the records, not just for position
 675     // So if just one client requires new searching, we need to clear set.
 676     // Ask each of the client if sorting requires re-search due to native sort
 677     // If it does it will require us to
 678     for (l = se->clients_active; l; l = l->next)
 679     {
 680         struct client *cl = l->client;
 681         // Assume no re-search is required.
 682         client_parse_init(cl, 1);
 683         clients_research += client_parse_sort(cl, sp);
 684     }
 685     if (!clients_research || se->clients_starting)
 686     {
 687         // A new sorting based on same record set
 688         struct reclist_sortparms *sr = nmem_malloc(se->nmem, sizeof(*sr));
 689         sr->name = nmem_strdup(se->nmem, field);
 690         sr->increasing = increasing;
 691         sr->type = type;
 692         sr->next = se->sorted_results;
 693         se->sorted_results = sr;
 694         session_log(se, YLOG_DEBUG, "session_sort: no research/ingesting done");
 695         session_leave(se, "session_sort");
 696     }
 697     else
 698     {
 699         se->clients_starting = 1;
 700         session_log(se, YLOG_DEBUG,
 701                     "session_sort: reset results due to %d clients researching",
 702                     clients_research);
 703         session_clear_set(se, sp);
 704         session_log(se, YLOG_DEBUG, "Re- search/ingesting for clients due to change in sort order");
 705
 706         session_leave(se, "session_sort");
 707         for (l = se->clients_active; l; l = l->next)
 708         {
 709             struct client *cl = l->client;
 710             if (client_get_state(cl) == Client_Connecting ||
 711                 client_get_state(cl) == Client_Idle ||
 712                 client_get_state(cl) == Client_Working) {
 713                 client_start_search(cl);
 714             }
 715             else
 716             {
 717                 session_log(se, YLOG_DEBUG,
 718                             "session_sort: %s: No re-start/ingest in show. "
 719                             "Wrong client state: %d",
 720                             client_get_id(cl), client_get_state(cl));
 721             }
 722         }
 723         session_enter(se, "session_sort");
 724         se->clients_starting = 0;
 725         session_leave(se, "session_sort");
 726     }
 727 }
 728
 729 void session_stop(struct session *se)
 730 {
 731     struct client_list *l;
 732     session_enter(se, "session_stop1");
 733     if (se->clients_starting)
 734     {
 735         session_leave(se, "session_stop1");
 736         return;
 737     }
 738     se->clients_starting = 1;
 739     session_leave(se, "session_stop1");
 740
 741     session_alert_watch(se, SESSION_WATCH_SHOW);
 742     session_alert_watch(se, SESSION_WATCH_BYTARGET);
 743     session_alert_watch(se, SESSION_WATCH_TERMLIST);
 744     session_alert_watch(se, SESSION_WATCH_SHOW_PREF);
 745
 746     for (l = se->clients_active; l; l = l->next)
 747     {
 748         struct client *cl = l->client;
 749         client_stop(cl);
 750     }
 751     session_enter(se, "session_stop2");
 752     se->clients_starting = 0;
 753     session_leave(se, "session_stop2");
 754 }
 755
 756 enum pazpar2_error_code session_search(struct session *se,
 757                                        const char *query,
 758                                        const char *startrecs,
 759                                        const char *maxrecs,
 760                                        const char *filter,
 761                                        const char *limit,
 762                                        const char **addinfo,
 763                                        const char **addinfo2,
 764                                        struct reclist_sortparms *sp,
 765                                        const char *mergekey,
 766                                        const char *rank)
 767 {
 768     int live_channels = 0;
 769     int no_working = 0;
 770     int no_failed_query = 0;
 771     int no_failed_limit = 0;
 772     struct client_list *l;
 773
 774     session_log(se, YLOG_DEBUG, "Search");
 775
 776     *addinfo = 0;
 777
 778     session_enter(se, "session_search0");
 779     if (se->clients_starting)
 780     {
 781         session_leave(se, "session_search0");
 782         return PAZPAR2_NO_ERROR;
 783     }
 784     se->clients_starting = 1;
 785     session_leave(se, "session_search0");
 786
 787     if (se->settings_modified) {
 788         session_remove_cached_clients(se);
 789     }
 790     else
 791         session_reset_active_clients(se, 0);
 792
 793     session_enter(se, "session_search");
 794     se->settings_modified = 0;
 795
 796     if (mergekey)
 797     {
 798         xfree(se->mergekey);
 799         se->mergekey = *mergekey ? xstrdup(mergekey) : 0;
 800     }
 801     if (rank)
 802     {
 803         xfree(se->rank);
 804         se->rank = *rank ? xstrdup(rank) : 0;
 805     }
 806
 807     session_clear_set(se, sp);
 808     relevance_destroy(&se->relevance);
 809
 810     live_channels = select_targets(se, filter);
 811     if (!live_channels)
 812     {
 813         session_leave(se, "session_search");
 814         se->clients_starting = 0;
 815         return PAZPAR2_NO_TARGETS;
 816     }
 817
 818     facet_limits_destroy(se->facet_limits);
 819     se->facet_limits = facet_limits_create(limit);
 820     if (!se->facet_limits)
 821     {
 822         *addinfo = "limit";
 823         session_leave(se, "session_search");
 824         se->clients_starting = 0;
 825         return PAZPAR2_MALFORMED_PARAMETER_VALUE;
 826     }
 827
 828     session_leave(se, "session_search");
 829
 830     session_alert_watch(se, SESSION_WATCH_SHOW);
 831     session_alert_watch(se, SESSION_WATCH_BYTARGET);
 832     session_alert_watch(se, SESSION_WATCH_TERMLIST);
 833     session_alert_watch(se, SESSION_WATCH_SHOW_PREF);
 834
 835     for (l = se->clients_active; l; l = l->next)
 836     {
 837         int parse_ret;
 838         struct client *cl = l->client;
 839         client_parse_init(cl, 1);
 840         if (prepare_map(se, client_get_database(cl)) < 0)
 841             continue;
 842
 843         parse_ret = client_parse_query(cl, query, se->facet_limits, addinfo2);
 844         if (parse_ret == -1)
 845             no_failed_query++;
 846         else if (parse_ret == -2)
 847             no_failed_limit++;
 848         else if (parse_ret < 0)
 849             no_working++; /* other error, such as bad CCL map */
 850         else
 851         {
 852             client_parse_range(cl, startrecs, maxrecs);
 853             client_parse_sort(cl, sp);
 854             client_start_search(cl);
 855             no_working++;
 856         }
 857     }
 858     session_enter(se, "session_search2");
 859     se->clients_starting = 0;
 860     session_leave(se, "session_search2");
 861     if (no_working == 0)
 862     {
 863         if (no_failed_query > 0)
 864         {
 865             *addinfo = "query";
 866             return PAZPAR2_MALFORMED_PARAMETER_VALUE;
 867         }
 868         else if (no_failed_limit > 0)
 869         {
 870             *addinfo = "limit";
 871             return PAZPAR2_MALFORMED_PARAMETER_VALUE;
 872         }
 873         else
 874             return PAZPAR2_NO_TARGETS;
 875     }
 876     return PAZPAR2_NO_ERROR;
 877 }
 878
 879 // Creates a new session_database object for a database
 880 static void session_init_databases_fun(void *context, struct database *db)
 881 {
 882     struct session *se = (struct session *) context;
 883     struct session_database *new = nmem_malloc(se->session_nmem, sizeof(*new));
 884     int i;
 885
 886     new->database = db;
 887
 888     new->map = 0;
 889     assert(db->settings);
 890     new->settings = nmem_malloc(se->session_nmem,
 891                                 sizeof(struct settings *) * db->num_settings);
 892     new->num_settings = db->num_settings;
 893     for (i = 0; i < db->num_settings; i++)
 894     {
 895         struct setting *setting = db->settings[i];
 896         new->settings[i] = setting;
 897     }
 898     new->next = se->databases;
 899     se->databases = new;
 900 }
 901
 902 // Doesn't free memory associated with sdb -- nmem takes care of that
 903 static void session_database_destroy(struct session_database *sdb)
 904 {
 905     sdb->map = 0;
 906 }
 907
 908 // Initialize session_database list -- this represents this session's view
 909 // of the database list -- subject to modification by the settings ws command
 910 void session_init_databases(struct session *se)
 911 {
 912     se->databases = 0;
 913     predef_grep_databases(se, se->service, session_init_databases_fun);
 914 }
 915
 916 // Probably session_init_databases_fun should be refactored instead of
 917 // called here.
 918 static struct session_database *load_session_database(struct session *se,
 919                                                       const char *id)
 920 {
 921     struct database *db = new_database_inherit_settings(id, se->session_nmem, se->service->settings);
 922     session_init_databases_fun((void*) se, db);
 923
 924     // New sdb is head of se->databases list
 925     return se->databases;
 926 }
 927
 928 // Find an existing session database. If not found, load it
 929 static struct session_database *find_session_database(struct session *se,
 930                                                       const char *id)
 931 {
 932     struct session_database *sdb;
 933
 934     for (sdb = se->databases; sdb; sdb = sdb->next)
 935         if (!strcmp(sdb->database->id, id))
 936             return sdb;
 937     return load_session_database(se, id);
 938 }
 939
 940 // Apply a session override to a database
 941 void session_apply_setting(struct session *se, const char *dbname,
 942                            const char *name, const char *value)
 943 {
 944     session_enter(se, "session_apply_setting");
 945     {
 946         struct session_database *sdb = find_session_database(se, dbname);
 947         struct conf_service *service = se->service;
 948         struct setting *s;
 949         int offset = settings_create_offset(service, name);
 950
 951         expand_settings_array(&sdb->settings, &sdb->num_settings, offset,
 952                               se->session_nmem);
 953         // Force later recompute of settings-driven data structures
 954         // (happens when a search starts and client connections are prepared)
 955         if (offset == PZ_XSLT)
 956             sdb->map = 0;
 957         se->settings_modified = 1;
 958         for (s = sdb->settings[offset]; s; s = s->next)
 959             if (!strcmp(s->name, name) &&
 960                 dbname && s->target && !strcmp(dbname, s->target))
 961                 break;
 962         if (!s)
 963         {
 964             s = nmem_malloc(se->session_nmem, sizeof(*s));
 965             s->precedence = 0;
 966             s->target = nmem_strdup(se->session_nmem, dbname);
 967             s->name = nmem_strdup(se->session_nmem, name);
 968             s->next = sdb->settings[offset];
 969             sdb->settings[offset] = s;
 970         }
 971         s->value = nmem_strdup(se->session_nmem, value);
 972     }
 973     session_leave(se, "session_apply_setting");
 974 }
 975
 976 void session_destroy(struct session *se)
 977 {
 978     struct session_database *sdb;
 979     session_log(se, YLOG_LOG, "destroy");
 980     session_use(-1);
 981     session_remove_cached_clients(se);
 982
 983     for (sdb = se->databases; sdb; sdb = sdb->next)
 984         session_database_destroy(sdb);
 985     normalize_cache_destroy(se->normalize_cache);
 986     relevance_destroy(&se->relevance);
 987     reclist_destroy(se->reclist);
 988     xfree(se->mergekey);
 989     xfree(se->rank);
 990     if (nmem_total(se->nmem))
 991         session_log(se, YLOG_DEBUG, "NMEN operation usage %zd", nmem_total(se->nmem));
 992     if (nmem_total(se->session_nmem))
 993         session_log(se, YLOG_DEBUG, "NMEN session usage %zd", nmem_total(se->session_nmem));
 994     facet_limits_destroy(se->facet_limits);
 995     nmem_destroy(se->nmem);
 996     service_destroy(se->service);
 997     yaz_mutex_destroy(&se->session_mutex);
 998 }
 999
1000 size_t session_get_memory_status(struct session *session) {
1001     size_t session_nmem;
1002     if (session == 0)
1003         return 0;
1004     session_enter(session, "session_get_memory_status");
1005     session_nmem = nmem_total(session->nmem);
1006     session_leave(session, "session_get_memory_status");
1007     return session_nmem;
1008 }
1009
1010
1011 struct session *new_session(NMEM nmem, struct conf_service *service,
1012                             unsigned session_id)
1013 {
1014     int i;
1015     struct session *session = nmem_malloc(nmem, sizeof(*session));
1016
1017     char tmp_str[50];
1018
1019     sprintf(tmp_str, "session#%u", session_id);
1020
1021     session->session_id = session_id;
1022     session_log(session, YLOG_DEBUG, "New");
1023     session->service = service;
1024     session->relevance = 0;
1025     session->total_records = 0;
1026     session->number_of_warnings_unknown_elements = 0;
1027     session->number_of_warnings_unknown_metadata = 0;
1028     session->num_termlists = 0;
1029     session->reclist = reclist_create(nmem);
1030     session->clients_active = 0;
1031     session->clients_cached = 0;
1032     session->settings_modified = 0;
1033     session->session_nmem = nmem;
1034     session->nmem = nmem_create();
1035     session->databases = 0;
1036     session->sorted_results = 0;
1037     session->facet_limits = 0;
1038     session->mergekey = 0;
1039     session->rank = 0;
1040     session->clients_starting = 0;
1041
1042     for (i = 0; i <= SESSION_WATCH_MAX; i++)
1043     {
1044         session->watchlist[i].data = 0;
1045         session->watchlist[i].fun = 0;
1046     }
1047     session->normalize_cache = normalize_cache_create();
1048     session->session_mutex = 0;
1049     pazpar2_mutex_create(&session->session_mutex, tmp_str);
1050     session_log(session, YLOG_LOG, "create");
1051
1052     session_use(1);
1053     return session;
1054 }
1055
1056 const char * client_get_suggestions_xml(struct client *cl, WRBUF wrbuf);
1057
1058 static struct hitsbytarget *hitsbytarget_nb(struct session *se,
1059                                             int *count, NMEM nmem)
1060 {
1061     struct hitsbytarget *res = 0;
1062     struct client_list *l;
1063     size_t sz = 0;
1064
1065     for (l = se->clients_active; l; l = l->next)
1066         sz++;
1067
1068     res = nmem_malloc(nmem, sizeof(*res) * sz);
1069     *count = 0;
1070     for (l = se->clients_active; l; l = l->next)
1071     {
1072         struct client *cl = l->client;
1073         WRBUF w = wrbuf_alloc();
1074         const char *name = session_setting_oneval(client_get_database(cl),
1075                                                   PZ_NAME);
1076         res[*count].id = client_get_id(cl);
1077         res[*count].name = *name ? name : "Unknown";
1078         res[*count].hits = client_get_hits(cl);
1079         res[*count].approximation = client_get_approximation(cl);
1080         res[*count].records = client_get_num_records(cl,
1081                                                      &res[*count].filtered,
1082                                                      0, 0);
1083         res[*count].diagnostic =
1084             client_get_diagnostic(cl, &res[*count].message,
1085                                   &res[*count].addinfo);
1086         res[*count].state = client_get_state_str(cl);
1087         res[*count].connected  = client_get_connection(cl) ? 1 : 0;
1088         session_settings_dump(se, client_get_database(cl), w);
1089         res[*count].settings_xml = nmem_strdup(nmem, wrbuf_cstr(w));
1090         wrbuf_rewind(w);
1091         wrbuf_puts(w, "");
1092         res[*count].suggestions_xml = nmem_strdup(nmem, client_get_suggestions_xml(cl, w));
1093         wrbuf_destroy(w);
1094         (*count)++;
1095     }
1096     return res;
1097 }
1098
1099 struct hitsbytarget *get_hitsbytarget(struct session *se, int *count, NMEM nmem)
1100 {
1101     struct hitsbytarget *p;
1102     session_enter(se, "get_hitsbytarget");
1103     p = hitsbytarget_nb(se, count, nmem);
1104     session_leave(se, "get_hitsbytarget");
1105     return p;
1106 }
1107
1108 // Compares two hitsbytarget nodes by hitcount
1109 static int cmp_ht(const void *p1, const void *p2)
1110 {
1111     const struct hitsbytarget *h1 = p1;
1112     const struct hitsbytarget *h2 = p2;
1113     return h2->hits - h1->hits;
1114 }
1115
1116 // Compares two hitsbytarget nodes by hitcount
1117 static int cmp_ht_approx(const void *p1, const void *p2)
1118 {
1119     const struct hitsbytarget *h1 = p1;
1120     const struct hitsbytarget *h2 = p2;
1121     return h2->approximation - h1->approximation;
1122 }
1123
1124 static int targets_termlist_nb(WRBUF wrbuf, struct session *se, int num,
1125                                NMEM nmem, int version)
1126 {
1127     struct hitsbytarget *ht;
1128     int count, i;
1129
1130     ht = hitsbytarget_nb(se, &count, nmem);
1131     if (version >= 2)
1132         qsort(ht, count, sizeof(struct hitsbytarget), cmp_ht_approx);
1133     else
1134         qsort(ht, count, sizeof(struct hitsbytarget), cmp_ht);
1135     for (i = 0; i < count && i < num && ht[i].hits > 0; i++)
1136     {
1137
1138         // do only print terms which have display names
1139
1140         wrbuf_puts(wrbuf, "<term>\n");
1141
1142         wrbuf_puts(wrbuf, "<id>");
1143         wrbuf_xmlputs(wrbuf, ht[i].id);
1144         wrbuf_puts(wrbuf, "</id>\n");
1145
1146         wrbuf_puts(wrbuf, "<name>");
1147         if (!ht[i].name || !ht[i].name[0])
1148             wrbuf_xmlputs(wrbuf, "NO TARGET NAME");
1149         else
1150             wrbuf_xmlputs(wrbuf, ht[i].name);
1151         wrbuf_puts(wrbuf, "</name>\n");
1152
1153         wrbuf_printf(wrbuf, "<frequency>" ODR_INT_PRINTF "</frequency>\n",
1154                      ht[i].hits);
1155
1156         if (version >= 2) {
1157             // Should not print if we know it isn't a approximation.
1158             wrbuf_printf(wrbuf, "<approximation>" ODR_INT_PRINTF "</approximation>\n", ht[i].approximation);
1159             wrbuf_printf(wrbuf, "<records>%d</records>\n", ht[i].records - ht[i].filtered);
1160             wrbuf_printf(wrbuf, "<filtered>%d</filtered>\n", ht[i].filtered);
1161         }
1162
1163         wrbuf_puts(wrbuf, "<state>");
1164         wrbuf_xmlputs(wrbuf, ht[i].state);
1165         wrbuf_puts(wrbuf, "</state>\n");
1166
1167         wrbuf_printf(wrbuf, "<diagnostic>%d</diagnostic>\n",
1168                      ht[i].diagnostic);
1169         wrbuf_puts(wrbuf, "</term>\n");
1170     }
1171     return count;
1172 }
1173
1174 void perform_termlist(struct http_channel *c, struct session *se,
1175                       const char *name, int num, int version)
1176 {
1177     int i, j;
1178     NMEM nmem_tmp = nmem_create();
1179     char **names;
1180     int num_names = 0;
1181
1182     if (!name)
1183         name = "*";
1184
1185     nmem_strsplit(nmem_tmp, ",", name, &names, &num_names);
1186
1187     session_enter(se, "perform_termlist");
1188
1189     for (j = 0; j < num_names; j++)
1190     {
1191         const char *tname;
1192         int must_generate_empty = 1; /* bug 5350 */
1193
1194         for (i = 0; i < se->num_termlists; i++)
1195         {
1196             tname = se->termlists[i].name;
1197             if (!strcmp(names[j], tname) || !strcmp(names[j], "*"))
1198             {
1199                 struct termlist_score **p = 0;
1200                 int len;
1201
1202                 wrbuf_puts(c->wrbuf, "<list name=\"");
1203                 wrbuf_xmlputs(c->wrbuf, tname);
1204                 wrbuf_puts(c->wrbuf, "\">\n");
1205                 must_generate_empty = 0;
1206
1207                 p = termlist_highscore(se->termlists[i].termlist, &len,
1208                                        nmem_tmp);
1209                 if (p)
1210                 {
1211                     int i;
1212                     for (i = 0; i < len && i < num; i++)
1213                     {
1214                         // prevent sending empty term elements
1215                         if (!p[i]->display_term || !p[i]->display_term[0])
1216                             continue;
1217
1218                         wrbuf_puts(c->wrbuf, "<term>");
1219                         wrbuf_puts(c->wrbuf, "<name>");
1220                         wrbuf_xmlputs(c->wrbuf, p[i]->display_term);
1221                         wrbuf_puts(c->wrbuf, "</name>");
1222
1223                         wrbuf_printf(c->wrbuf,
1224                                      "<frequency>%d</frequency>",
1225                                      p[i]->frequency);
1226                         wrbuf_puts(c->wrbuf, "</term>\n");
1227                     }
1228                 }
1229                 wrbuf_puts(c->wrbuf, "</list>\n");
1230             }
1231         }
1232         tname = "xtargets";
1233         if (!strcmp(names[j], tname) || !strcmp(names[j], "*"))
1234         {
1235             wrbuf_puts(c->wrbuf, "<list name=\"");
1236             wrbuf_xmlputs(c->wrbuf, tname);
1237             wrbuf_puts(c->wrbuf, "\">\n");
1238
1239             targets_termlist_nb(c->wrbuf, se, num, c->nmem, version);
1240             wrbuf_puts(c->wrbuf, "</list>\n");
1241             must_generate_empty = 0;
1242         }
1243         if (must_generate_empty)
1244         {
1245             wrbuf_puts(c->wrbuf, "<list name=\"");
1246             wrbuf_xmlputs(c->wrbuf, names[j]);
1247             wrbuf_puts(c->wrbuf, "\"/>\n");
1248         }
1249     }
1250     session_leave(se, "perform_termlist");
1251     nmem_destroy(nmem_tmp);
1252 }
1253
1254 #ifdef MISSING_HEADERS
1255 void report_nmem_stats(void)
1256 {
1257     size_t in_use, is_free;
1258
1259     nmem_get_memory_in_use(&in_use);
1260     nmem_get_memory_free(&is_free);
1261
1262     yaz_log(YLOG_LOG, "nmem stat: use=%ld free=%ld",
1263             (long) in_use, (long) is_free);
1264 }
1265 #endif
1266
1267 struct record_cluster *show_single_start(struct session *se, const char *id,
1268                                          struct record_cluster **prev_r,
1269                                          struct record_cluster **next_r)
1270 {
1271     struct record_cluster *r = 0;
1272
1273     session_enter(se, "show_single_start");
1274     *prev_r = 0;
1275     *next_r = 0;
1276     reclist_limit(se->reclist, se, 1);
1277
1278     reclist_enter(se->reclist);
1279     while ((r = reclist_read_record(se->reclist)))
1280     {
1281         if (!strcmp(r->recid, id))
1282         {
1283             *next_r = reclist_read_record(se->reclist);
1284             break;
1285         }
1286         *prev_r = r;
1287     }
1288     reclist_leave(se->reclist);
1289     if (!r)
1290         session_leave(se, "show_single_start");
1291     return r;
1292 }
1293
1294 void show_single_stop(struct session *se, struct record_cluster *rec)
1295 {
1296     session_leave(se, "show_single_stop");
1297 }
1298
1299
1300 int session_fetch_more(struct session *se)
1301 {
1302     struct client_list *l;
1303     int ret = 0;
1304
1305     for (l = se->clients_active; l; l = l->next)
1306     {
1307         struct client *cl = l->client;
1308         if (client_get_state(cl) == Client_Idle)
1309         {
1310             if (client_fetch_more(cl))
1311             {
1312                 session_log(se, YLOG_LOG, "%s: more to fetch",
1313                             client_get_id(cl));
1314                 ret = 1;
1315             }
1316             else
1317             {
1318                 int filtered;
1319                 int ingest_failures;
1320                 int record_failures;
1321                 int num = client_get_num_records(
1322                     cl, &filtered, &ingest_failures, &record_failures);
1323
1324                 session_log(se, YLOG_LOG, "%s: hits=" ODR_INT_PRINTF
1325                             " fetched=%d filtered=%d",
1326                             client_get_id(cl),
1327                             client_get_hits(cl),
1328                             num, filtered);
1329                 if (ingest_failures || record_failures)
1330                 {
1331                     session_log(se, YLOG_WARN, "%s:"
1332                                 " ingest failures=%d record failures=%d",
1333                                 client_get_id(cl),
1334                                 ingest_failures, record_failures);
1335                 }
1336             }
1337         }
1338         else
1339         {
1340             session_log(se, YLOG_LOG, "%s: no fetch due to state=%s",
1341                         client_get_id(cl), client_get_state_str(cl));
1342         }
1343
1344     }
1345     return ret;
1346 }
1347
1348 struct record_cluster **show_range_start(struct session *se,
1349                                          struct reclist_sortparms *sp,
1350                                          int start, int *num, int *total,
1351                                          Odr_int *sumhits, Odr_int *approx_hits,
1352                                          void (*show_records_ready)(void *data),
1353                                          struct http_channel *chan)
1354 {
1355     struct record_cluster **recs = 0;
1356     struct reclist_sortparms *spp;
1357     struct client_list *l;
1358     int i;
1359 #if USE_TIMING
1360     yaz_timing_t t = yaz_timing_create();
1361 #endif
1362     session_enter(se, "show_range_start");
1363     *sumhits = 0;
1364     *approx_hits = 0;
1365     *total = 0;
1366     reclist_limit(se->reclist, se, 0);
1367     if (se->relevance)
1368     {
1369         for (spp = sp; spp; spp = spp->next)
1370             if (spp->type == Metadata_type_relevance)
1371             {
1372                 relevance_prepare_read(se->relevance, se->reclist);
1373                 break;
1374             }
1375         for (l = se->clients_active; l; l = l->next) {
1376             *sumhits += client_get_hits(l->client);
1377             *approx_hits += client_get_approximation(l->client);
1378         }
1379     }
1380     reclist_sort(se->reclist, sp);
1381
1382     reclist_enter(se->reclist);
1383     *total = reclist_get_num_records(se->reclist);
1384
1385     for (l = se->clients_active; l; l = l->next)
1386         client_update_show_stat(l->client, 0);
1387
1388     for (i = 0; i < start; i++)
1389     {
1390         struct record_cluster *r = reclist_read_record(se->reclist);
1391         if (!r)
1392         {
1393             *num = 0;
1394             break;
1395         }
1396         else
1397         {
1398             struct record *rec = r->records;
1399             for (;rec; rec = rec->next)
1400                 client_update_show_stat(rec->client, 1);
1401         }
1402     }
1403     recs = nmem_malloc(se->nmem, (*num > 0 ? *num : 1) * sizeof(*recs));
1404     for (i = 0; i < *num; i++)
1405     {
1406         struct record_cluster *r = reclist_read_record(se->reclist);
1407         if (!r)
1408         {
1409             *num = i;
1410             break;
1411         }
1412         else
1413         {
1414             struct record *rec = r->records;
1415             for (;rec; rec = rec->next)
1416                 client_update_show_stat(rec->client, 1);
1417             recs[i] = r;
1418         }
1419     }
1420     reclist_leave(se->reclist);
1421 #if USE_TIMING
1422     yaz_timing_stop(t);
1423     session_log(se, YLOG_LOG, "show %6.5f %3.2f %3.2f",
1424             yaz_timing_get_real(t), yaz_timing_get_user(t),
1425             yaz_timing_get_sys(t));
1426     yaz_timing_destroy(&t);
1427 #endif
1428
1429     if (!session_fetch_more(se))
1430         session_log(se, YLOG_LOG, "can not fetch more");
1431     else
1432     {
1433         show_range_stop(se, recs);
1434         session_log(se, YLOG_LOG, "fetching more in progress");
1435         if (session_set_watch(se, SESSION_WATCH_SHOW,
1436                               show_records_ready, chan, chan))
1437         {
1438             session_log(se, YLOG_WARN, "Ignoring show block");
1439             session_enter(se, "show_range_start");
1440         }
1441         else
1442         {
1443             session_log(se, YLOG_LOG, "session watch OK");
1444             return 0;
1445         }
1446     }
1447     return recs;
1448 }
1449
1450 void show_range_stop(struct session *se, struct record_cluster **recs)
1451 {
1452     session_leave(se, "show_range_stop");
1453 }
1454
1455 void statistics(struct session *se, struct statistics *stat)
1456 {
1457     struct client_list *l;
1458     int count = 0;
1459
1460     memset(stat, 0, sizeof(*stat));
1461     stat->num_hits = 0;
1462     for (l = se->clients_active; l; l = l->next)
1463     {
1464         struct client *cl = l->client;
1465         if (!client_get_connection(cl))
1466             stat->num_no_connection++;
1467         stat->num_hits += client_get_hits(cl);
1468         switch (client_get_state(cl))
1469         {
1470         case Client_Connecting: stat->num_connecting++; break;
1471         case Client_Working: stat->num_working++; break;
1472         case Client_Idle: stat->num_idle++; break;
1473         case Client_Failed: stat->num_failed++; break;
1474         case Client_Error: stat->num_error++; break;
1475         default: break;
1476         }
1477         count++;
1478     }
1479     stat->num_records = se->total_records;
1480
1481     stat->num_clients = count;
1482 }
1483
1484 static struct record_metadata *record_metadata_init(
1485     NMEM nmem, const char *value, enum conf_metadata_type type,
1486     struct _xmlAttr *attr)
1487 {
1488     struct record_metadata *rec_md = record_metadata_create(nmem);
1489     struct record_metadata_attr **attrp = &rec_md->attributes;
1490
1491     for (; attr; attr = attr->next)
1492     {
1493         if (attr->children && attr->children->content)
1494         {
1495             if (strcmp((const char *) attr->name, "type")
1496                 && strcmp((const char *) attr->name, "empty"))
1497             {  /* skip the "type" + "empty" attribute..
1498                   The "Type" is already part of the element in output
1499                   (md-%s) and so repeating it here is redundant */
1500                 *attrp = nmem_malloc(nmem, sizeof(**attrp));
1501                 (*attrp)->name =
1502                     nmem_strdup(nmem, (const char *) attr->name);
1503                 (*attrp)->value =
1504                     nmem_strdup(nmem, (const char *) attr->children->content);
1505                 attrp = &(*attrp)->next;
1506             }
1507         }
1508     }
1509     *attrp = 0;
1510
1511     switch (type)
1512     {
1513     case Metadata_type_generic:
1514     case Metadata_type_skiparticle:
1515         if (strstr(value, "://")) /* looks like a URL */
1516             rec_md->data.text.disp = nmem_strdup(nmem, value);
1517         else
1518             rec_md->data.text.disp =
1519                 normalize7bit_generic(nmem_strdup(nmem, value), " ,/.:([");
1520         rec_md->data.text.sort = 0;
1521         rec_md->data.text.snippet = 0;
1522         break;
1523     case Metadata_type_year:
1524     case Metadata_type_date:
1525     {
1526         int first, last;
1527         int longdate = 0;
1528
1529         if (type == Metadata_type_date)
1530             longdate = 1;
1531         if (extract7bit_dates((char *) value, &first, &last, longdate) < 0)
1532             return 0;
1533
1534         rec_md->data.number.min = first;
1535         rec_md->data.number.max = last;
1536     }
1537     break;
1538     case Metadata_type_float:
1539         rec_md->data.fnumber = atof(value);
1540         break;
1541     case Metadata_type_relevance:
1542     case Metadata_type_position:
1543         return 0;
1544     }
1545     return rec_md;
1546 }
1547
1548 static void mergekey_norm_wr(pp2_charset_fact_t charsets,
1549                              WRBUF norm_wr, const char *value)
1550 {
1551     const char *norm_str;
1552     pp2_charset_token_t prt =
1553         pp2_charset_token_create(charsets, "mergekey");
1554
1555     pp2_charset_token_first(prt, value, 0);
1556     while ((norm_str = pp2_charset_token_next(prt)))
1557     {
1558         if (*norm_str)
1559         {
1560             if (wrbuf_len(norm_wr))
1561                 wrbuf_puts(norm_wr, " ");
1562             wrbuf_puts(norm_wr, norm_str);
1563         }
1564     }
1565     pp2_charset_token_destroy(prt);
1566 }
1567
1568 static int get_mergekey_from_doc(xmlDoc *doc, xmlNode *root, const char *name,
1569                                  struct conf_service *service, WRBUF norm_wr)
1570 {
1571     xmlNode *n;
1572     int no_found = 0;
1573     for (n = root->children; n; n = n->next)
1574     {
1575         if (n->type != XML_ELEMENT_NODE)
1576             continue;
1577         if (!strcmp((const char *) n->name, "metadata"))
1578         {
1579             xmlChar *type = xmlGetProp(n, (xmlChar *) "type");
1580             if (type == NULL) {
1581                 yaz_log(YLOG_FATAL, "Missing type attribute on metadata element. Skipping!");
1582             }
1583             else if (!strcmp(name, (const char *) type))
1584             {
1585                 xmlChar *value = xmlNodeListGetString(doc, n->children, 1);
1586                 if (value && *value)
1587                 {
1588                     if (wrbuf_len(norm_wr) > 0)
1589                         wrbuf_puts(norm_wr, " ");
1590                     wrbuf_puts(norm_wr, name);
1591                     mergekey_norm_wr(service->charsets, norm_wr,
1592                                      (const char *) value);
1593                     no_found++;
1594                 }
1595                 if (value)
1596                     xmlFree(value);
1597             }
1598             xmlFree(type);
1599         }
1600     }
1601     return no_found;
1602 }
1603
1604 static const char *get_mergekey(xmlDoc *doc, xmlNode *root,
1605                                 struct client *cl, int record_no,
1606                                 struct conf_service *service, NMEM nmem,
1607                                 const char *session_mergekey)
1608 {
1609     char *mergekey_norm = 0;
1610     WRBUF norm_wr = wrbuf_alloc();
1611     xmlChar *mergekey;
1612
1613     if (session_mergekey)
1614     {
1615         int i, num = 0;
1616         char **values = 0;
1617         nmem_strsplit_escape2(nmem, ",", session_mergekey, &values,
1618                               &num, 1, '\\', 1);
1619
1620         for (i = 0; i < num; i++)
1621             get_mergekey_from_doc(doc, root, values[i], service, norm_wr);
1622     }
1623     else if ((mergekey = xmlGetProp(root, (xmlChar *) "mergekey")))
1624     {
1625         mergekey_norm_wr(service->charsets, norm_wr, (const char *) mergekey);
1626         xmlFree(mergekey);
1627     }
1628     else
1629     {
1630         /* no mergekey defined in XSL. Look for mergekey metadata instead */
1631         int field_id;
1632         for (field_id = 0; field_id < service->num_metadata; field_id++)
1633         {
1634             struct conf_metadata *ser_md = &service->metadata[field_id];
1635             if (ser_md->mergekey != Metadata_mergekey_no)
1636             {
1637                 int r = get_mergekey_from_doc(doc, root, ser_md->name,
1638                                               service, norm_wr);
1639                 if (r == 0 && ser_md->mergekey == Metadata_mergekey_required)
1640                 {
1641                     /* no mergekey on this one and it is required..
1642                        Generate unique key instead */
1643                     wrbuf_rewind(norm_wr);
1644                     break;
1645                 }
1646             }
1647         }
1648     }
1649
1650     /* generate unique key if none is not generated already or is empty */
1651     if (wrbuf_len(norm_wr) == 0)
1652     {
1653         wrbuf_printf(norm_wr, "position: %s-%d",
1654                      client_get_id(cl), record_no);
1655     }
1656     else
1657     {
1658         const char *lead = "content: ";
1659         wrbuf_insert(norm_wr, 0, lead, strlen(lead));
1660     }
1661     if (wrbuf_len(norm_wr) > 0)
1662         mergekey_norm = nmem_strdup(nmem, wrbuf_cstr(norm_wr));
1663     wrbuf_destroy(norm_wr);
1664     return mergekey_norm;
1665 }
1666
1667 /** \brief see if metadata for pz:recordfilter exists
1668     \param root xml root element of normalized record
1669     \param sdb session database for client
1670     \retval 0 if there is no metadata for pz:recordfilter
1671     \retval 1 if there is metadata for pz:recordfilter
1672
1673     If there is no pz:recordfilter defined, this function returns 1
1674     as well.
1675 */
1676
1677 static int check_record_filter(xmlNode *root, struct session_database *sdb)
1678 {
1679     int match = 0;
1680     xmlNode *n;
1681     const char *s;
1682     s = session_setting_oneval(sdb, PZ_RECORDFILTER);
1683
1684     if (!s || !*s)
1685         return 1;
1686
1687     for (n = root->children; n; n = n->next)
1688     {
1689         if (n->type != XML_ELEMENT_NODE)
1690             continue;
1691         if (!strcmp((const char *) n->name, "metadata"))
1692         {
1693             xmlChar *type = xmlGetProp(n, (xmlChar *) "type");
1694             if (type)
1695             {
1696                 size_t len;
1697                 int substring;
1698                 const char *eq;
1699
1700                 if ((eq = strchr(s, '=')))
1701                     substring = 0;
1702                 else if ((eq = strchr(s, '~')))
1703                     substring = 1;
1704                 if (eq)
1705                     len = eq - s;
1706                 else
1707                     len = strlen(s);
1708                 if (len == strlen((const char *)type) &&
1709                     !memcmp((const char *) type, s, len))
1710                 {
1711                     xmlChar *value = xmlNodeGetContent(n);
1712                     if (value && *value)
1713                     {
1714                         if (!eq ||
1715                             (substring && strstr((const char *) value, eq+1)) ||
1716                             (!substring && !strcmp((const char *) value, eq + 1)))
1717                             match = 1;
1718                     }
1719                     xmlFree(value);
1720                 }
1721                 xmlFree(type);
1722             }
1723         }
1724     }
1725     return match;
1726 }
1727
1728 static int ingest_to_cluster(struct client *cl,
1729                              xmlDoc *xdoc,
1730                              xmlNode *root,
1731                              int record_no,
1732                              struct record_metadata_attr *mergekey);
1733
1734 static int ingest_sub_record(struct client *cl, xmlDoc *xdoc, xmlNode *root,
1735                              int record_no, NMEM nmem,
1736                              struct session_database *sdb,
1737                              struct record_metadata_attr *mergekeys)
1738 {
1739     int ret = 0;
1740     struct session *se = client_get_session(cl);
1741
1742     if (!check_record_filter(root, sdb))
1743     {
1744         session_log(se, YLOG_LOG,
1745                     "Filtered out record no %d from %s",
1746                     record_no, sdb->database->id);
1747         return 0;
1748     }
1749     session_enter(se, "ingest_sub_record");
1750     if (client_get_session(cl) == se && se->relevance)
1751         ret = ingest_to_cluster(cl, xdoc, root, record_no, mergekeys);
1752     session_leave(se, "ingest_sub_record");
1753
1754     return ret;
1755 }
1756
1757 /** \brief ingest XML record
1758     \param cl client holds the result set for record
1759     \param rec record buffer (0 terminated)
1760     \param record_no record position (1, 2, ..)
1761     \param nmem working NMEM
1762     \retval 0 OK
1763     \retval -1 failure
1764     \retval -2 Filtered
1765 */
1766 int ingest_record(struct client *cl, const char *rec,
1767                   int record_no, NMEM nmem)
1768 {
1769     struct session *se = client_get_session(cl);
1770     struct session_database *sdb = client_get_database(cl);
1771     struct conf_service *service = se->service;
1772     xmlDoc *xdoc = normalize_record(se, sdb, service, rec, nmem);
1773     int r = ingest_xml_record(cl, xdoc, record_no, nmem, 0);
1774     client_store_xdoc(cl, record_no, xdoc);
1775     return r;
1776 }
1777
1778 int ingest_xml_record(struct client *cl, xmlDoc *xdoc,
1779                       int record_no, NMEM nmem, int cached_copy)
1780 {
1781     struct session *se = client_get_session(cl);
1782     struct session_database *sdb = client_get_database(cl);
1783     struct conf_service *service = se->service;
1784     xmlNode *root;
1785     int r = 0;
1786     if (!xdoc)
1787         return -1;
1788
1789     if (global_parameters.dump_records)
1790     {
1791         session_log(se, YLOG_LOG, "Normalized record from %s",
1792                     sdb->database->id);
1793         log_xml_doc(xdoc);
1794     }
1795
1796     root = xmlDocGetRootElement(xdoc);
1797
1798     if (!strcmp((const char *) root->name, "cluster"))
1799     {
1800         int no_merge_keys = 0;
1801         int no_merge_dups = 0;
1802         xmlNode *sroot;
1803         struct record_metadata_attr *mk = 0;
1804
1805         for (sroot = root->children; sroot; sroot = sroot->next)
1806             if (sroot->type == XML_ELEMENT_NODE &&
1807                 !strcmp((const char *) sroot->name, "record"))
1808             {
1809                 struct record_metadata_attr **mkp;
1810                 const char *mergekey_norm =
1811                     get_mergekey(xdoc, sroot, cl, record_no, service, nmem,
1812                                  se->mergekey);
1813                 if (!mergekey_norm)
1814                 {
1815                     r = -1;
1816                     break;
1817                 }
1818                 for (mkp = &mk; *mkp; mkp = &(*mkp)->next)
1819                     if (!strcmp((*mkp)->value, mergekey_norm))
1820                         break;
1821                 if (!*mkp)
1822                 {
1823                     *mkp = (struct record_metadata_attr*)
1824                         nmem_malloc(nmem, sizeof(**mkp));
1825                     (*mkp)->name = 0;
1826                     (*mkp)->value = nmem_strdup(nmem, mergekey_norm);
1827                     (*mkp)->next = 0;
1828                     no_merge_keys++;
1829                 }
1830                 else
1831                     no_merge_dups++;
1832             }
1833         if (no_merge_keys > 1 || no_merge_dups > 0)
1834         {
1835             yaz_log(YLOG_LOG, "Got %d mergekeys, %d dups for position %d",
1836                     no_merge_keys, no_merge_dups, record_no);
1837         }
1838         for (sroot = root->children; !r && sroot; sroot = sroot->next)
1839             if (sroot->type == XML_ELEMENT_NODE &&
1840                 !strcmp((const char *) sroot->name, "record"))
1841             {
1842                 if (!cached_copy)
1843                     insert_settings_values(sdb, xdoc, root, service);
1844                 r = ingest_sub_record(cl, xdoc, sroot, record_no, nmem, sdb,
1845                                       mk);
1846             }
1847     }
1848     else if (!strcmp((const char *) root->name, "record"))
1849     {
1850         const char *mergekey_norm =
1851             get_mergekey(xdoc, root, cl, record_no, service, nmem,
1852                          se->mergekey);
1853         if (mergekey_norm)
1854         {
1855             struct record_metadata_attr *mk = (struct record_metadata_attr*)
1856                 nmem_malloc(nmem, sizeof(*mk));
1857             mk->name = 0;
1858             mk->value = nmem_strdup(nmem, mergekey_norm);
1859             mk->next = 0;
1860
1861             if (!cached_copy)
1862                 insert_settings_values(sdb, xdoc, root, service);
1863             r = ingest_sub_record(cl, xdoc, root, record_no, nmem, sdb, mk);
1864         }
1865     }
1866     else
1867     {
1868         session_log(se, YLOG_WARN, "Bad pz root element: %s",
1869                     (const char *) root->name);
1870         r = -1;
1871     }
1872     return r;
1873 }
1874
1875
1876 //    struct conf_metadata *ser_md = &service->metadata[md_field_id];
1877 //    struct record_metadata *rec_md = record->metadata[md_field_id];
1878 static int match_metadata_local(struct conf_service *service,
1879                                 struct conf_metadata *ser_md,
1880                                 struct record_metadata *rec_md0,
1881                                 char **values, int num_v)
1882 {
1883     int i;
1884     struct record_metadata *rec_md = rec_md0;
1885     WRBUF val_wr = 0;
1886     WRBUF text_wr = wrbuf_alloc();
1887     for (i = 0; i < num_v; )
1888     {
1889         if (rec_md)
1890         {
1891             if (ser_md->type == Metadata_type_year
1892                 || ser_md->type == Metadata_type_date)
1893             {
1894                 int y = atoi(values[i]);
1895                 if (y >= rec_md->data.number.min
1896                     && y <= rec_md->data.number.max)
1897                     break;
1898             }
1899             else
1900             {
1901                 if (!val_wr)
1902                 {
1903                     val_wr = wrbuf_alloc();
1904                     mergekey_norm_wr(service->charsets, val_wr, values[i]);
1905                 }
1906                 wrbuf_rewind(text_wr);
1907                 mergekey_norm_wr(service->charsets, text_wr,
1908                                  rec_md->data.text.disp);
1909                 if (!strcmp(wrbuf_cstr(val_wr), wrbuf_cstr(text_wr)))
1910                     break;
1911             }
1912             rec_md = rec_md->next;
1913         }
1914         else
1915         {
1916             rec_md = rec_md0;
1917             wrbuf_destroy(val_wr);
1918             val_wr = 0;
1919             i++;
1920         }
1921     }
1922     wrbuf_destroy(val_wr);
1923     wrbuf_destroy(text_wr);
1924     return i < num_v ? 1 : 0;
1925 }
1926
1927 int session_check_cluster_limit(struct session *se, struct record_cluster *rec)
1928 {
1929     int i;
1930     struct conf_service *service = se->service;
1931     int ret = 1;
1932     const char *name;
1933     const char *value;
1934     NMEM nmem_tmp = nmem_create();
1935
1936     for (i = 0; (name = facet_limits_get(se->facet_limits, i, &value)); i++)
1937     {
1938         int j;
1939         for (j = 0; j < service->num_metadata; j++)
1940         {
1941             struct conf_metadata *md = service->metadata + j;
1942             if (!strcmp(md->name, name) && md->limitcluster)
1943             {
1944                 char **values = 0;
1945                 int num = 0;
1946                 int md_field_id =
1947                     conf_service_metadata_field_id(service,
1948                                                    md->limitcluster);
1949
1950                 if (md_field_id < 0)
1951                 {
1952                     ret = 0;
1953                     break;
1954                 }
1955
1956                 nmem_strsplit_escape2(nmem_tmp, "|", value, &values,
1957                                       &num, 1, '\\', 1);
1958
1959                 if (!match_metadata_local(service,
1960                                           &service->metadata[md_field_id],
1961                                           rec->metadata[md_field_id],
1962                                           values, num))
1963                 {
1964                     ret = 0;
1965                     break;
1966                 }
1967             }
1968         }
1969     }
1970     nmem_destroy(nmem_tmp);
1971     return ret;
1972 }
1973
1974 // Skip record on non-zero
1975 static int check_limit_local(struct client *cl,
1976                              struct record *record,
1977                              int record_no)
1978 {
1979     int skip_record = 0;
1980     struct session *se = client_get_session(cl);
1981     struct conf_service *service = se->service;
1982     NMEM nmem_tmp = nmem_create();
1983     struct session_database *sdb = client_get_database(cl);
1984     int l = 0;
1985     while (!skip_record)
1986     {
1987         int md_field_id;
1988         char **values = 0;
1989         int num_v = 0;
1990         const char *name =
1991             client_get_facet_limit_local(cl, sdb, &l, nmem_tmp,
1992                                          &num_v, &values);
1993         if (!name)
1994             break;
1995
1996         if (!strcmp(name, "*"))
1997         {
1998             for (md_field_id = 0; md_field_id < service->num_metadata;
1999                  md_field_id++)
2000             {
2001                 if (match_metadata_local(
2002                         service,
2003                         &service->metadata[md_field_id],
2004                         record->metadata[md_field_id],
2005                         values, num_v))
2006                     break;
2007             }
2008             if (md_field_id == service->num_metadata)
2009                 skip_record = 1;
2010         }
2011         else
2012         {
2013             md_field_id = conf_service_metadata_field_id(service, name);
2014             if (md_field_id < 0)
2015             {
2016                 skip_record = 1;
2017                 break;
2018             }
2019             if (!match_metadata_local(
2020                     service,
2021                     &service->metadata[md_field_id],
2022                     record->metadata[md_field_id],
2023                     values, num_v))
2024             {
2025                 skip_record = 1;
2026             }
2027         }
2028     }
2029     nmem_destroy(nmem_tmp);
2030     return skip_record;
2031 }
2032
2033 static int ingest_to_cluster(struct client *cl,
2034                              xmlDoc *xdoc,
2035                              xmlNode *root,
2036                              int record_no,
2037                              struct record_metadata_attr *merge_keys)
2038 {
2039     xmlNode *n;
2040     xmlChar *type = 0;
2041     xmlChar *value = 0;
2042     struct session *se = client_get_session(cl);
2043     struct conf_service *service = se->service;
2044     int term_factor = 1;
2045     struct record_cluster *cluster;
2046     struct record_metadata **metadata0;
2047     struct session_database *sdb = client_get_database(cl);
2048     NMEM ingest_nmem = 0;
2049     char **rank_values = 0;
2050     int rank_num = 0;
2051     struct record *record = record_create(se->nmem,
2052                                           service->num_metadata,
2053                                           service->num_sortkeys, cl,
2054                                           record_no);
2055
2056     for (n = root->children; n; n = n->next)
2057     {
2058         if (type)
2059             xmlFree(type);
2060         if (value)
2061             xmlFree(value);
2062         type = value = 0;
2063
2064         if (n->type != XML_ELEMENT_NODE)
2065             continue;
2066         if (!strcmp((const char *) n->name, "metadata"))
2067         {
2068             struct conf_metadata *ser_md = 0;
2069             struct record_metadata **wheretoput = 0;
2070             struct record_metadata *rec_md = 0;
2071             int md_field_id = -1;
2072
2073             type = xmlGetProp(n, (xmlChar *) "type");
2074             value = xmlNodeListGetString(xdoc, n->children, 1);
2075             if (!type)
2076                 continue;
2077             if (!value || !*value)
2078             {
2079                 xmlChar *empty = xmlGetProp(n, (xmlChar *) "empty");
2080                 if (!empty)
2081                     continue;
2082                 if (value)
2083                     xmlFree(value);
2084                 value = empty;
2085             }
2086             md_field_id
2087                 = conf_service_metadata_field_id(service, (const char *) type);
2088             if (md_field_id < 0)
2089             {
2090                 if (se->number_of_warnings_unknown_metadata == 0)
2091                 {
2092                     session_log(se, YLOG_WARN,
2093                             "Ignoring unknown metadata element: %s", type);
2094                 }
2095                 se->number_of_warnings_unknown_metadata++;
2096                 continue;
2097             }
2098
2099             ser_md = &service->metadata[md_field_id];
2100
2101             // non-merged metadata
2102             rec_md = record_metadata_init(se->nmem, (const char *) value,
2103                                           ser_md->type, n->properties);
2104             if (!rec_md)
2105             {
2106                 session_log(se, YLOG_WARN, "bad metadata data '%s' "
2107                             "for element '%s'", value, type);
2108                 continue;
2109             }
2110
2111             if (ser_md->type == Metadata_type_generic)
2112             {
2113                 WRBUF w = wrbuf_alloc();
2114                 if (relevance_snippet(se->relevance,
2115                                       (char*) value, ser_md->name, w))
2116                     rec_md->data.text.snippet = nmem_strdup(se->nmem,
2117                                                             wrbuf_cstr(w));
2118                 wrbuf_destroy(w);
2119             }
2120
2121
2122             wheretoput = &record->metadata[md_field_id];
2123             while (*wheretoput)
2124                 wheretoput = &(*wheretoput)->next;
2125             *wheretoput = rec_md;
2126         }
2127     }
2128
2129     if (check_limit_local(cl, record, record_no))
2130     {
2131         if (type)
2132             xmlFree(type);
2133         if (value)
2134             xmlFree(value);
2135         return -2;
2136     }
2137     cluster = reclist_insert(se->reclist, se->relevance, service, record,
2138                              merge_keys, &se->total_merged);
2139     if (!cluster)
2140     {
2141         if (type)
2142             xmlFree(type);
2143         if (value)
2144             xmlFree(value);
2145         return 0; // complete match with existing record
2146     }
2147
2148     {
2149         const char *use_term_factor_str =
2150             session_setting_oneval(sdb, PZ_TERMLIST_TERM_FACTOR);
2151         if (use_term_factor_str && use_term_factor_str[0] == '1')
2152         {
2153             int maxrecs = client_get_maxrecs(cl);
2154             int hits = (int) client_get_hits(cl);
2155             term_factor = MAX(hits, maxrecs) /  MAX(1, maxrecs);
2156             assert(term_factor >= 1);
2157             session_log(se, YLOG_DEBUG, "Using term factor: %d (%d / %d)",
2158                         term_factor, MAX(hits, maxrecs), MAX(1, maxrecs));
2159         }
2160     }
2161
2162     if (global_parameters.dump_records)
2163         session_log(se, YLOG_LOG, "Cluster id %s from %s (#%d)", cluster->recid,
2164                     sdb->database->id, record_no);
2165
2166     // original metadata, to check if first existence of a field
2167     metadata0 = xmalloc(sizeof(*metadata0) * service->num_metadata);
2168     memcpy(metadata0, cluster->metadata,
2169            sizeof(*metadata0) * service->num_metadata);
2170
2171     ingest_nmem = nmem_create();
2172     if (se->rank)
2173     {
2174         yaz_log(YLOG_LOG, "local in sort : %s", se->rank);
2175         nmem_strsplit_escape2(ingest_nmem, ",", se->rank, &rank_values,
2176                               &rank_num, 1, '\\', 1);
2177     }
2178
2179     // now parsing XML record and adding data to cluster or record metadata
2180     for (n = root->children; n; n = n->next)
2181     {
2182         pp2_charset_token_t prt;
2183         if (type)
2184             xmlFree(type);
2185         if (value)
2186             xmlFree(value);
2187         type = value = 0;
2188
2189         if (n->type != XML_ELEMENT_NODE)
2190             continue;
2191         if (!strcmp((const char *) n->name, "metadata"))
2192         {
2193             struct conf_metadata *ser_md = 0;
2194             struct conf_sortkey *ser_sk = 0;
2195             struct record_metadata **wheretoput = 0;
2196             struct record_metadata *rec_md = 0;
2197             int md_field_id = -1;
2198             int sk_field_id = -1;
2199             const char *rank = 0;
2200             xmlChar *xml_rank = 0;
2201
2202             type = xmlGetProp(n, (xmlChar *) "type");
2203             value = xmlNodeListGetString(xdoc, n->children, 1);
2204
2205             if (!type || !value || !*value)
2206                 continue;
2207
2208             md_field_id
2209                 = conf_service_metadata_field_id(service, (const char *) type);
2210             if (md_field_id < 0)
2211                 continue;
2212
2213             ser_md = &service->metadata[md_field_id];
2214
2215             if (ser_md->sortkey_offset >= 0)
2216             {
2217                 sk_field_id = ser_md->sortkey_offset;
2218                 ser_sk = &service->sortkeys[sk_field_id];
2219             }
2220
2221             // merged metadata
2222             rec_md = record_metadata_init(se->nmem, (const char *) value,
2223                                           ser_md->type, 0);
2224
2225             // see if the field was not in cluster already (from beginning)
2226
2227             if (!rec_md)
2228                 continue;
2229
2230             if (rank_num)
2231             {
2232                 int i;
2233                 for (i = 0; i < rank_num; i++)
2234                 {
2235                     const char *val = rank_values[i];
2236                     const char *cp = strchr(val, '=');
2237                     if (!cp)
2238                         continue;
2239                     if ((cp - val) == strlen((const char *) type)
2240                         && !memcmp(val, type, cp - val))
2241                     {
2242                         rank = cp + 1;
2243                         break;
2244                     }
2245                 }
2246             }
2247             else
2248             {
2249                 xml_rank = xmlGetProp(n, (xmlChar *) "rank");
2250                 rank = xml_rank ? (const char *) xml_rank : ser_md->rank;
2251             }
2252
2253             wheretoput = &cluster->metadata[md_field_id];
2254
2255             if (ser_md->merge == Metadata_merge_first)
2256             {
2257                 if (!metadata0[md_field_id])
2258                 {
2259                     while (*wheretoput)
2260                         wheretoput = &(*wheretoput)->next;
2261                     *wheretoput = rec_md;
2262                 }
2263             }
2264             else if (ser_md->merge == Metadata_merge_unique)
2265             {
2266                 while (*wheretoput)
2267                 {
2268                     if (!strcmp((const char *) (*wheretoput)->data.text.disp,
2269                                 rec_md->data.text.disp))
2270                         break;
2271                     wheretoput = &(*wheretoput)->next;
2272                 }
2273                 if (!*wheretoput)
2274                     *wheretoput = rec_md;
2275             }
2276             else if (ser_md->merge == Metadata_merge_longest)
2277             {
2278                 if (!*wheretoput
2279                     || strlen(rec_md->data.text.disp)
2280                     > strlen((*wheretoput)->data.text.disp))
2281                 {
2282                     *wheretoput = rec_md;
2283                     if (ser_sk)
2284                     {
2285                         const char *sort_str = 0;
2286                         int skip_article =
2287                             ser_sk->type == Metadata_type_skiparticle;
2288
2289                         if (!cluster->sortkeys[sk_field_id])
2290                             cluster->sortkeys[sk_field_id] =
2291                                 nmem_malloc(se->nmem,
2292                                             sizeof(union data_types));
2293
2294                         prt =
2295                             pp2_charset_token_create(service->charsets, "sort");
2296
2297                         pp2_charset_token_first(prt, rec_md->data.text.disp,
2298                                                 skip_article);
2299
2300                         pp2_charset_token_next(prt);
2301
2302                         sort_str = pp2_get_sort(prt);
2303
2304                         cluster->sortkeys[sk_field_id]->text.disp =
2305                             rec_md->data.text.disp;
2306                         if (!sort_str)
2307                         {
2308                             sort_str = rec_md->data.text.disp;
2309                             session_log(se, YLOG_WARN,
2310                                     "Could not make sortkey. Bug #1858");
2311                         }
2312                         cluster->sortkeys[sk_field_id]->text.sort =
2313                             nmem_strdup(se->nmem, sort_str);
2314                         pp2_charset_token_destroy(prt);
2315                     }
2316                 }
2317             }
2318             else if (ser_md->merge == Metadata_merge_all)
2319             {
2320                 while (*wheretoput)
2321                     wheretoput = &(*wheretoput)->next;
2322                 *wheretoput = rec_md;
2323             }
2324             else if (ser_md->merge == Metadata_merge_range)
2325             {
2326                 if (!*wheretoput)
2327                 {
2328                     *wheretoput = rec_md;
2329                     if (ser_sk)
2330                         cluster->sortkeys[sk_field_id]
2331                             = &rec_md->data;
2332                 }
2333                 else
2334                 {
2335                     int this_min = rec_md->data.number.min;
2336                     int this_max = rec_md->data.number.max;
2337                     if (this_min < (*wheretoput)->data.number.min)
2338                         (*wheretoput)->data.number.min = this_min;
2339                     if (this_max > (*wheretoput)->data.number.max)
2340                         (*wheretoput)->data.number.max = this_max;
2341                 }
2342             }
2343
2344             // ranking of _all_ fields enabled ...
2345             if (rank)
2346             {
2347                 relevance_countwords(se->relevance, cluster,
2348                                      (char *) value, rank, ser_md->name);
2349             }
2350             // construct facets ... unless the client already has reported them
2351             if (ser_md->termlist && !client_has_facet(cl, (char *) type))
2352             {
2353                 if (ser_md->type == Metadata_type_year)
2354                 {
2355                     char year[64];
2356                     sprintf(year, "%d", rec_md->data.number.max);
2357
2358                     add_facet(se, (char *) type, year, term_factor);
2359                     if (rec_md->data.number.max != rec_md->data.number.min)
2360                     {
2361                         sprintf(year, "%d", rec_md->data.number.min);
2362                         add_facet(se, (char *) type, year, term_factor);
2363                     }
2364                 }
2365                 else
2366                     add_facet(se, (char *) type, (char *) value, term_factor);
2367             }
2368
2369             // cleaning up
2370             if (xml_rank)
2371                 xmlFree(xml_rank);
2372             xmlFree(type);
2373             xmlFree(value);
2374             type = value = 0;
2375         }
2376         else
2377         {
2378             if (se->number_of_warnings_unknown_elements == 0)
2379                 session_log(se, YLOG_WARN,
2380                         "Unexpected element in internal record: %s", n->name);
2381             se->number_of_warnings_unknown_elements++;
2382         }
2383     }
2384     if (type)
2385         xmlFree(type);
2386     if (value)
2387         xmlFree(value);
2388
2389     nmem_destroy(ingest_nmem);
2390     xfree(metadata0);
2391     relevance_donerecord(se->relevance, cluster);
2392     se->total_records++;
2393
2394     return 0;
2395 }
2396
2397 void session_log(struct session *s, int level, const char *fmt, ...)
2398 {
2399     char buf[1024];
2400     va_list ap;
2401     va_start(ap, fmt);
2402
2403     yaz_vsnprintf(buf, sizeof(buf)-30, fmt, ap);
2404     yaz_log(level, "Session %u: %s", s ? s->session_id : 0, buf);
2405
2406     va_end(ap);
2407 }
2408
2409 /*
2410  * Local variables:
2411  * c-basic-offset: 4
2412  * c-file-style: "Stroustrup"
2413  * indent-tabs-mode: nil
2414  * End:
2415  * vim: shiftwidth=4 tabstop=8 expandtab
2416  */
2417