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 #include <yaz/xml_get.h>
  60
  61 #define USE_TIMING 0
  62 #if USE_TIMING
  63 #include <yaz/timing.h>
  64 #endif
  65
  66 #include "ppmutex.h"
  67 #include "parameters.h"
  68 #include "session.h"
  69 #include "eventl.h"
  70 #include "http.h"
  71 #include "termlists.h"
  72 #include "reclists.h"
  73 #include "relevance.h"
  74 #include "database.h"
  75 #include "client.h"
  76 #include "settings.h"
  77 #include "normalize7bit.h"
  78
  79 #include <libxml/tree.h>
  80
  81 #define MAX_CHUNK 15
  82
  83 #define MAX(a,b) ((a)>(b)?(a):(b))
  84
  85 // Note: Some things in this structure will eventually move to configuration
  86 struct parameters global_parameters =
  87 {
  88     0,   // dump_records
  89     0,   // debug_mode
  90     0,   // predictable sessions
  91 };
  92
  93 struct client_list {
  94     struct client *client;
  95     struct client_list *next;
  96 };
  97
  98 /* session counting (1) , disable client counting (0) */
  99 static YAZ_MUTEX g_session_mutex = 0;
 100 static int no_sessions = 0;
 101
 102 static int session_use(int delta)
 103 {
 104     int sessions;
 105     if (!g_session_mutex)
 106         yaz_mutex_create(&g_session_mutex);
 107     yaz_mutex_enter(g_session_mutex);
 108     no_sessions += delta;
 109     sessions = no_sessions;
 110     yaz_mutex_leave(g_session_mutex);
 111     yaz_log(YLOG_DEBUG, "%s sessions=%d", delta == 0 ? "" :
 112             (delta > 0 ? "INC" : "DEC"), no_sessions);
 113     return sessions;
 114 }
 115
 116 int sessions_count(void)
 117 {
 118     return session_use(0);
 119 }
 120
 121 static void log_xml_doc(xmlDoc *doc)
 122 {
 123     FILE *lf = yaz_log_file();
 124     xmlChar *result = 0;
 125     int len = 0;
 126 #if LIBXML_VERSION >= 20600
 127     xmlDocDumpFormatMemory(doc, &result, &len, 1);
 128 #else
 129     xmlDocDumpMemory(doc, &result, &len);
 130 #endif
 131     if (lf && len)
 132     {
 133         (void) fwrite(result, 1, len, lf);
 134         fprintf(lf, "\n");
 135     }
 136     xmlFree(result);
 137 }
 138
 139 static void session_enter(struct session *s, const char *caller)
 140 {
 141     if (caller)
 142         session_log(s, YLOG_DEBUG, "Session lock by %s", caller);
 143     yaz_mutex_enter(s->session_mutex);
 144 }
 145
 146 static void session_leave(struct session *s, const char *caller)
 147 {
 148     yaz_mutex_leave(s->session_mutex);
 149     if (caller)
 150         session_log(s, YLOG_DEBUG, "Session unlock by %s", caller);
 151 }
 152
 153 static int run_icu(struct session *s, const char *icu_chain_id,
 154                    const char *value,
 155                    WRBUF norm_wr, WRBUF disp_wr)
 156 {
 157     const char *facet_component;
 158     struct conf_service *service = s->service;
 159     pp2_charset_token_t prt =
 160         pp2_charset_token_create(service->charsets, icu_chain_id);
 161     if (!prt)
 162     {
 163         session_log(s, YLOG_FATAL,
 164                     "Unknown ICU chain '%s'", icu_chain_id);
 165         return 0;
 166     }
 167     pp2_charset_token_first(prt, value, 0);
 168     while ((facet_component = pp2_charset_token_next(prt)))
 169     {
 170         const char *display_component;
 171         if (*facet_component)
 172         {
 173             if (wrbuf_len(norm_wr))
 174                 wrbuf_puts(norm_wr, " ");
 175             wrbuf_puts(norm_wr, facet_component);
 176         }
 177         display_component = pp2_get_display(prt);
 178         if (display_component)
 179         {
 180             if (wrbuf_len(disp_wr))
 181                 wrbuf_puts(disp_wr, " ");
 182             wrbuf_puts(disp_wr, display_component);
 183         }
 184     }
 185     pp2_charset_token_destroy(prt);
 186     return 1;
 187 }
 188
 189 static void session_normalize_facet(struct session *s,
 190                                     const char *type, const char *value,
 191                                     WRBUF display_wrbuf, WRBUF facet_wrbuf)
 192 {
 193     struct conf_service *service = s->service;
 194     int i;
 195     const char *icu_chain_id = 0;
 196
 197     for (i = 0; i < service->num_metadata; i++)
 198         if (!strcmp((service->metadata + i)->name, type))
 199             icu_chain_id = (service->metadata + i)->facetrule;
 200     if (!icu_chain_id)
 201         icu_chain_id = "facet";
 202
 203     run_icu(s, icu_chain_id, value, facet_wrbuf, display_wrbuf);
 204 }
 205
 206 struct facet_id {
 207     char *client_id;
 208     char *type;
 209     char *id;
 210     char *term;
 211     struct facet_id *next;
 212 };
 213
 214 static void session_add_id_facet(struct session *s, struct client *cl,
 215                                  const char *type,
 216                                  const char *id,
 217                                  size_t id_len,
 218                                  const char *term)
 219 {
 220     struct facet_id *t = nmem_malloc(s->session_nmem, sizeof(*t));
 221
 222     t->client_id = nmem_strdup(s->session_nmem, client_get_id(cl));
 223     t->type = nmem_strdup(s->session_nmem, type);
 224     t->id = nmem_strdupn(s->session_nmem, id, id_len);
 225     t->term = nmem_strdup(s->session_nmem, term);
 226     t->next = s->facet_id_list;
 227     s->facet_id_list = t;
 228 }
 229
 230
 231 const char *session_lookup_id_facet(struct session *s, struct client *cl,
 232                                     const char *type,
 233                                     const char *term)
 234 {
 235     struct facet_id *t = s->facet_id_list;
 236     for (; t; t = t->next)
 237         if (!strcmp(client_get_id(cl), t->client_id) &&
 238             !strcmp(t->type, type) && !strcmp(t->term, term))
 239         {
 240             return t->id;
 241         }
 242     return 0;
 243 }
 244
 245 void add_facet(struct session *s, const char *type, const char *value, int count, struct client *cl)
 246 {
 247     WRBUF facet_wrbuf = wrbuf_alloc();
 248     WRBUF display_wrbuf = wrbuf_alloc();
 249     const char *id = 0;
 250     size_t id_len = 0;
 251
 252     /* inspect pz:facetmap:split:name ?? */
 253     if (!strncmp(type, "split:", 6))
 254     {
 255         const char *cp = strchr(value, ':');
 256         if (cp)
 257         {
 258             id = value;
 259             id_len = cp - value;
 260             value = cp + 1;
 261         }
 262         type += 6;
 263     }
 264
 265     session_normalize_facet(s, type, value, display_wrbuf, facet_wrbuf);
 266     if (wrbuf_len(facet_wrbuf))
 267     {
 268         struct named_termlist **tp = &s->termlists;
 269         for (; (*tp); tp = &(*tp)->next)
 270             if (!strcmp((*tp)->name, type))
 271                 break;
 272         if (!*tp)
 273         {
 274             *tp = nmem_malloc(s->nmem, sizeof(**tp));
 275             (*tp)->name = nmem_strdup(s->nmem, type);
 276             (*tp)->termlist = termlist_create(s->nmem);
 277             (*tp)->next = 0;
 278         }
 279         termlist_insert((*tp)->termlist, wrbuf_cstr(display_wrbuf),
 280                         wrbuf_cstr(facet_wrbuf), id, id_len, count);
 281         if (id)
 282             session_add_id_facet(s, cl, type, id, id_len,
 283                                  wrbuf_cstr(display_wrbuf));
 284     }
 285     wrbuf_destroy(facet_wrbuf);
 286     wrbuf_destroy(display_wrbuf);
 287 }
 288
 289 static xmlDoc *record_to_xml(struct session *se,
 290                              struct session_database *sdb, const char *rec)
 291 {
 292     struct database *db = sdb->database;
 293     xmlDoc *rdoc = 0;
 294
 295     rdoc = xmlParseMemory(rec, strlen(rec));
 296
 297     if (!rdoc)
 298     {
 299         session_log(se, YLOG_WARN, "Non-wellformed XML");
 300         return 0;
 301     }
 302
 303     if (global_parameters.dump_records)
 304     {
 305         session_log(se, YLOG_LOG, "Un-normalized record from %s", db->id);
 306         log_xml_doc(rdoc);
 307     }
 308
 309     return rdoc;
 310 }
 311
 312 #define MAX_XSLT_ARGS 16
 313
 314 // Add static values from session database settings if applicable
 315 static void insert_settings_parameters(struct session_database *sdb,
 316                                        struct conf_service *service,
 317                                        char **parms,
 318                                        NMEM nmem)
 319 {
 320     int i;
 321     int nparms = 0;
 322     int offset = 0;
 323
 324     for (i = 0; i < service->num_metadata; i++)
 325     {
 326         struct conf_metadata *md = &service->metadata[i];
 327         int setting;
 328
 329         if (md->setting == Metadata_setting_parameter &&
 330             (setting = settings_lookup_offset(service, md->name)) >= 0)
 331         {
 332             const char *val = session_setting_oneval(sdb, setting);
 333             if (val && nparms < MAX_XSLT_ARGS)
 334             {
 335                 char *buf;
 336                 int len = strlen(val);
 337                 buf = nmem_malloc(nmem, len + 3);
 338                 buf[0] = '\'';
 339                 strcpy(buf + 1, val);
 340                 buf[len+1] = '\'';
 341                 buf[len+2] = '\0';
 342                 parms[offset++] = md->name;
 343                 parms[offset++] = buf;
 344                 nparms++;
 345             }
 346         }
 347     }
 348     parms[offset] = 0;
 349 }
 350
 351 // Add static values from session database settings if applicable
 352 static void insert_settings_values(struct session_database *sdb, xmlDoc *doc,
 353                                    xmlNode *root,
 354                                    struct conf_service *service)
 355 {
 356     int i;
 357
 358     for (i = 0; i < service->num_metadata; i++)
 359     {
 360         struct conf_metadata *md = &service->metadata[i];
 361         int offset;
 362
 363         if (md->setting == Metadata_setting_postproc &&
 364             (offset = settings_lookup_offset(service, md->name)) >= 0)
 365         {
 366             const char *val = session_setting_oneval(sdb, offset);
 367             if (val)
 368             {
 369                 xmlNode *n = xmlNewTextChild(root, 0, (xmlChar *) "metadata",
 370                                              (xmlChar *) val);
 371                 xmlSetProp(n, (xmlChar *) "type", (xmlChar *) md->name);
 372             }
 373         }
 374     }
 375 }
 376
 377 static xmlDoc *normalize_record(struct session *se,
 378                                 struct session_database *sdb,
 379                                 struct conf_service *service,
 380                                 const char *rec, NMEM nmem)
 381 {
 382     xmlDoc *rdoc = record_to_xml(se, sdb, rec);
 383
 384     if (rdoc)
 385     {
 386         char *parms[MAX_XSLT_ARGS*2+1];
 387
 388         insert_settings_parameters(sdb, service, parms, nmem);
 389
 390         if (normalize_record_transform(sdb->map, &rdoc, (const char **)parms))
 391         {
 392             session_log(se, YLOG_WARN, "Normalize failed");
 393         }
 394     }
 395     return rdoc;
 396 }
 397
 398 void session_settings_dump(struct session *se,
 399                            struct session_database *db,
 400                            WRBUF w)
 401 {
 402     if (db->settings)
 403     {
 404         int i, num = db->num_settings;
 405         for (i = 0; i < num; i++)
 406         {
 407             struct setting *s = db->settings[i];
 408             for (;s ; s = s->next)
 409             {
 410                 wrbuf_puts(w, "<set name=\"");
 411                 wrbuf_xmlputs(w, s->name);
 412                 wrbuf_puts(w, "\" value=\"");
 413                 wrbuf_xmlputs(w, s->value);
 414                 wrbuf_puts(w, "\"/>");
 415             }
 416             if (db->settings[i])
 417                 wrbuf_puts(w, "\n");
 418         }
 419     }
 420 }
 421
 422 // Retrieve first defined value for 'name' for given database.
 423 // Will be extended to take into account user associated with session
 424 const char *session_setting_oneval(struct session_database *db, int offset)
 425 {
 426     if (offset >= db->num_settings || !db->settings[offset])
 427         return "";
 428     return db->settings[offset]->value;
 429 }
 430
 431 // Prepare XSLT stylesheets for record normalization
 432 // Structures are allocated on the session_wide nmem to avoid having
 433 // to recompute this for every search. This would lead
 434 // to leaking if a single session was to repeatedly change the PZ_XSLT
 435 // setting. However, this is not a realistic use scenario.
 436 static int prepare_map(struct session *se, struct session_database *sdb)
 437 {
 438     if (sdb->settings && !sdb->map)
 439     {
 440         const char *s;
 441
 442         if (sdb->settings[PZ_XSLT] &&
 443             (s = session_setting_oneval(sdb, PZ_XSLT)))
 444         {
 445             char auto_stylesheet[256];
 446
 447             if (!strcmp(s, "auto"))
 448             {
 449                 const char *request_syntax = session_setting_oneval(
 450                     sdb, PZ_REQUESTSYNTAX);
 451                 if (request_syntax)
 452                 {
 453                     char *cp;
 454                     yaz_snprintf(auto_stylesheet, sizeof(auto_stylesheet),
 455                                  "%s.xsl", request_syntax);
 456                     for (cp = auto_stylesheet; *cp; cp++)
 457                     {
 458                         /* deliberately only consider ASCII */
 459                         if (*cp > 32 && *cp < 127)
 460                             *cp = tolower(*cp);
 461                     }
 462                     s = auto_stylesheet;
 463                 }
 464                 else
 465                 {
 466                     session_log(se, YLOG_WARN,
 467                                 "No pz:requestsyntax for auto stylesheet");
 468                 }
 469             }
 470             sdb->map = normalize_cache_get(se->normalize_cache,
 471                                            se->service, s);
 472             if (!sdb->map)
 473                 return -1;
 474         }
 475     }
 476     return 0;
 477 }
 478
 479 // called if watch should be removed because http_channel is to be destroyed
 480 static void session_watch_cancel(void *data, struct http_channel *c,
 481                                  void *data2)
 482 {
 483     struct session_watchentry *ent = data;
 484
 485     ent->fun = 0;
 486     ent->data = 0;
 487     ent->obs = 0;
 488 }
 489
 490 // set watch. Returns 0=OK, -1 if watch is already set
 491 int session_set_watch(struct session *s, int what,
 492                       session_watchfun fun, void *data,
 493                       struct http_channel *chan)
 494 {
 495     int ret;
 496     session_enter(s, "session_set_watch");
 497     if (s->watchlist[what].fun)
 498         ret = -1;
 499     else
 500     {
 501
 502         s->watchlist[what].fun = fun;
 503         s->watchlist[what].data = data;
 504         s->watchlist[what].obs = http_add_observer(chan, &s->watchlist[what],
 505                                                    session_watch_cancel);
 506         ret = 0;
 507     }
 508     session_leave(s, "session_set_watch");
 509     return ret;
 510 }
 511
 512 void session_alert_watch(struct session *s, int what)
 513 {
 514     assert(s);
 515     session_enter(s, "session_alert_watch");
 516     if (s->watchlist[what].fun)
 517     {
 518         /* our watch is no longer associated with http_channel */
 519         void *data;
 520         session_watchfun fun;
 521
 522         http_remove_observer(s->watchlist[what].obs);
 523         fun  = s->watchlist[what].fun;
 524         data = s->watchlist[what].data;
 525
 526         /* reset watch before fun is invoked - in case fun wants to set
 527            it again */
 528         s->watchlist[what].fun = 0;
 529         s->watchlist[what].data = 0;
 530         s->watchlist[what].obs = 0;
 531
 532         session_leave(s, "session_alert_watch");
 533         session_log(s, YLOG_DEBUG,
 534                     "Alert Watch: %d calling function: %p", what, fun);
 535         fun(data);
 536     }
 537     else
 538         session_leave(s,"session_alert_watch");
 539 }
 540
 541 //callback for grep_databases
 542 static void select_targets_callback(struct session *se,
 543                                     struct session_database *db)
 544 {
 545     struct client *cl;
 546     struct client_list *l;
 547
 548     for (l = se->clients_cached; l; l = l->next)
 549         if (client_get_database(l->client) == db)
 550             break;
 551
 552     if (l)
 553         cl = l->client;
 554     else
 555     {
 556         cl = client_create(db->database->id);
 557         client_set_database(cl, db);
 558
 559         l = xmalloc(sizeof(*l));
 560         l->client = cl;
 561         l->next = se->clients_cached;
 562         se->clients_cached = l;
 563     }
 564     /* set session always. If may be 0 if client is not active */
 565     client_set_session(cl, se);
 566
 567     l = xmalloc(sizeof(*l));
 568     l->client = cl;
 569     l->next = se->clients_active;
 570     se->clients_active = l;
 571 }
 572
 573 static void session_reset_active_clients(struct session *se,
 574                                          struct client_list *new_list)
 575 {
 576     struct client_list *l;
 577
 578     session_enter(se, "session_reset_active_clients");
 579     l = se->clients_active;
 580     se->clients_active = new_list;
 581     session_leave(se, "session_reset_active_clients");
 582
 583     while (l)
 584     {
 585         struct client_list *l_next = l->next;
 586
 587         client_lock(l->client);
 588         client_set_session(l->client, 0); /* mark client inactive */
 589         client_unlock(l->client);
 590
 591         xfree(l);
 592         l = l_next;
 593     }
 594 }
 595
 596 static void session_remove_cached_clients(struct session *se)
 597 {
 598     struct client_list *l;
 599
 600     session_reset_active_clients(se, 0);
 601
 602     session_enter(se, "session_remove_cached_clients");
 603     l = se->clients_cached;
 604     se->clients_cached = 0;
 605     session_leave(se, "session_remove_cached_clients");
 606
 607     while (l)
 608     {
 609         struct client_list *l_next = l->next;
 610         client_lock(l->client);
 611         client_set_session(l->client, 0);
 612         client_set_database(l->client, 0);
 613         client_unlock(l->client);
 614         client_destroy(l->client);
 615         xfree(l);
 616         l = l_next;
 617     }
 618 }
 619
 620 // Associates a set of clients with a session;
 621 // Note: Session-databases represent databases with per-session
 622 // setting overrides
 623 static int select_targets(struct session *se, const char *filter)
 624 {
 625     return session_grep_databases(se, filter, select_targets_callback);
 626 }
 627
 628 int session_active_clients(struct session *s)
 629 {
 630     struct client_list *l;
 631     int res = 0;
 632
 633     for (l = s->clients_active; l; l = l->next)
 634         if (client_is_active(l->client))
 635             res++;
 636
 637     return res;
 638 }
 639
 640 int session_is_preferred_clients_ready(struct session *s)
 641 {
 642     struct client_list *l;
 643     int res = 0;
 644
 645     for (l = s->clients_active; l; l = l->next)
 646         if (client_is_active_preferred(l->client))
 647             res++;
 648     session_log(s, YLOG_DEBUG, "Has %d active preferred clients.", res);
 649     return res == 0;
 650 }
 651
 652 static void session_clear_set(struct session *se, struct reclist_sortparms *sp)
 653 {
 654     reclist_destroy(se->reclist);
 655     if (nmem_total(se->nmem))
 656         session_log(se, YLOG_DEBUG, "NMEN operation usage %zd",
 657                     nmem_total(se->nmem));
 658     nmem_reset(se->nmem);
 659     se->total_records = se->total_merged = 0;
 660     se->termlists = 0;
 661     relevance_clear(se->relevance);
 662
 663     /* reset list of sorted results and clear to relevance search */
 664     se->sorted_results = nmem_malloc(se->nmem, sizeof(*se->sorted_results));
 665     se->sorted_results->name = nmem_strdup(se->nmem, sp->name);
 666     se->sorted_results->increasing = sp->increasing;
 667     se->sorted_results->type = sp->type;
 668     se->sorted_results->next = 0;
 669
 670     session_log(se, YLOG_DEBUG, "clear_set session_sort: field=%s increasing=%d type=%d configured",
 671             sp->name, sp->increasing, sp->type);
 672
 673     se->reclist = reclist_create(se->nmem);
 674 }
 675
 676 void session_sort(struct session *se, struct reclist_sortparms *sp,
 677                   const char *mergekey, const char *rank)
 678 {
 679     struct client_list *l;
 680     const char *field = sp->name;
 681     int increasing = sp->increasing;
 682     int type  = sp->type;
 683     int clients_research = 0;
 684
 685     session_enter(se, "session_sort");
 686     session_log(se, YLOG_DEBUG, "session_sort field=%s increasing=%d type=%d",
 687                 field, increasing, type);
 688
 689     if (rank && (!se->rank || strcmp(se->rank, rank)))
 690     {
 691         /* new rank must research/reingest anyway */
 692         assert(rank);
 693         xfree(se->rank);
 694         se->rank = *rank ? xstrdup(rank) : 0;
 695         clients_research = 1;
 696         session_log(se, YLOG_DEBUG, "session_sort: new rank = %s",
 697                     rank);
 698     }
 699     if (mergekey && (!se->mergekey || strcmp(se->mergekey, mergekey)))
 700     {
 701         /* new mergekey must research/reingest anyway */
 702         assert(mergekey);
 703         xfree(se->mergekey);
 704         se->mergekey = *mergekey ? xstrdup(mergekey) : 0;
 705         clients_research = 1;
 706         session_log(se, YLOG_DEBUG, "session_sort: new mergekey = %s",
 707                     mergekey);
 708     }
 709     if (clients_research == 0)
 710     {
 711         struct reclist_sortparms *sr;
 712         for (sr = se->sorted_results; sr; sr = sr->next)
 713             if (!reclist_sortparms_cmp(sr, sp))
 714                 break;
 715         if (sr)
 716         {
 717             session_log(se, YLOG_DEBUG, "session_sort: field=%s increasing=%d type=%d already fetched",
 718                         field, increasing, type);
 719             session_leave(se, "session_sort");
 720             return;
 721         }
 722     }
 723     session_log(se, YLOG_DEBUG, "session_sort: field=%s increasing=%d type=%d must fetch",
 724                 field, increasing, type);
 725
 726     // We need to reset reclist on every sort that changes the records, not just for position
 727     // So if just one client requires new searching, we need to clear set.
 728     // Ask each of the client if sorting requires re-search due to native sort
 729     // If it does it will require us to
 730     for (l = se->clients_active; l; l = l->next)
 731     {
 732         struct client *cl = l->client;
 733         // Assume no re-search is required.
 734         client_parse_init(cl, 1);
 735         clients_research += client_parse_sort(cl, sp);
 736     }
 737     if (!clients_research || se->clients_starting)
 738     {
 739         // A new sorting based on same record set
 740         struct reclist_sortparms *sr = nmem_malloc(se->nmem, sizeof(*sr));
 741         sr->name = nmem_strdup(se->nmem, field);
 742         sr->increasing = increasing;
 743         sr->type = type;
 744         sr->next = se->sorted_results;
 745         se->sorted_results = sr;
 746         session_log(se, YLOG_DEBUG, "session_sort: no research/ingesting done");
 747         session_leave(se, "session_sort");
 748     }
 749     else
 750     {
 751         se->clients_starting = 1;
 752         session_log(se, YLOG_DEBUG,
 753                     "session_sort: reset results due to %d clients researching",
 754                     clients_research);
 755         session_clear_set(se, sp);
 756         session_log(se, YLOG_DEBUG, "Re- search/ingesting for clients due to change in sort order");
 757
 758         session_leave(se, "session_sort");
 759         for (l = se->clients_active; l; l = l->next)
 760         {
 761             struct client *cl = l->client;
 762             if (client_get_state(cl) == Client_Connecting ||
 763                 client_get_state(cl) == Client_Idle ||
 764                 client_get_state(cl) == Client_Working) {
 765                 client_start_search(cl);
 766             }
 767             else
 768             {
 769                 session_log(se, YLOG_DEBUG,
 770                             "session_sort: %s: No re-start/ingest in show. "
 771                             "Wrong client state: %d",
 772                             client_get_id(cl), client_get_state(cl));
 773             }
 774         }
 775         session_enter(se, "session_sort");
 776         se->clients_starting = 0;
 777         session_leave(se, "session_sort");
 778     }
 779 }
 780
 781 void session_stop(struct session *se)
 782 {
 783     struct client_list *l;
 784     session_enter(se, "session_stop1");
 785     if (se->clients_starting)
 786     {
 787         session_leave(se, "session_stop1");
 788         return;
 789     }
 790     se->clients_starting = 1;
 791     session_leave(se, "session_stop1");
 792
 793     session_alert_watch(se, SESSION_WATCH_SHOW);
 794     session_alert_watch(se, SESSION_WATCH_BYTARGET);
 795     session_alert_watch(se, SESSION_WATCH_TERMLIST);
 796     session_alert_watch(se, SESSION_WATCH_SHOW_PREF);
 797
 798     for (l = se->clients_active; l; l = l->next)
 799     {
 800         struct client *cl = l->client;
 801         client_stop(cl);
 802     }
 803     session_enter(se, "session_stop2");
 804     se->clients_starting = 0;
 805     session_leave(se, "session_stop2");
 806 }
 807
 808 enum pazpar2_error_code session_search(struct session *se,
 809                                        const char *query,
 810                                        const char *startrecs,
 811                                        const char *maxrecs,
 812                                        const char *filter,
 813                                        const char *limit,
 814                                        const char **addinfo,
 815                                        const char **addinfo2,
 816                                        struct reclist_sortparms *sp,
 817                                        const char *mergekey,
 818                                        const char *rank)
 819 {
 820     int live_channels = 0;
 821     int no_working = 0;
 822     int no_failed_query = 0;
 823     int no_failed_limit = 0;
 824     struct client_list *l;
 825
 826     session_log(se, YLOG_DEBUG, "Search");
 827
 828     *addinfo = 0;
 829
 830     session_enter(se, "session_search0");
 831     if (se->clients_starting)
 832     {
 833         session_leave(se, "session_search0");
 834         return PAZPAR2_NO_ERROR;
 835     }
 836     se->clients_starting = 1;
 837     session_leave(se, "session_search0");
 838
 839     if (se->settings_modified) {
 840         session_remove_cached_clients(se);
 841     }
 842     else
 843         session_reset_active_clients(se, 0);
 844
 845     session_enter(se, "session_search");
 846     se->settings_modified = 0;
 847
 848     if (mergekey)
 849     {
 850         xfree(se->mergekey);
 851         se->mergekey = *mergekey ? xstrdup(mergekey) : 0;
 852     }
 853     if (rank)
 854     {
 855         xfree(se->rank);
 856         se->rank = *rank ? xstrdup(rank) : 0;
 857     }
 858
 859     session_clear_set(se, sp);
 860     relevance_destroy(&se->relevance);
 861
 862     live_channels = select_targets(se, filter);
 863     if (!live_channels)
 864     {
 865         session_leave(se, "session_search");
 866         se->clients_starting = 0;
 867         return PAZPAR2_NO_TARGETS;
 868     }
 869
 870     facet_limits_destroy(se->facet_limits);
 871     se->facet_limits = facet_limits_create(limit);
 872     if (!se->facet_limits)
 873     {
 874         *addinfo = "limit";
 875         session_leave(se, "session_search");
 876         se->clients_starting = 0;
 877         return PAZPAR2_MALFORMED_PARAMETER_VALUE;
 878     }
 879
 880     session_leave(se, "session_search");
 881
 882     session_alert_watch(se, SESSION_WATCH_SHOW);
 883     session_alert_watch(se, SESSION_WATCH_BYTARGET);
 884     session_alert_watch(se, SESSION_WATCH_TERMLIST);
 885     session_alert_watch(se, SESSION_WATCH_SHOW_PREF);
 886
 887     for (l = se->clients_active; l; l = l->next)
 888     {
 889         int parse_ret;
 890         struct client *cl = l->client;
 891         client_parse_init(cl, 1);
 892         if (prepare_map(se, client_get_database(cl)) < 0)
 893             continue;
 894
 895         parse_ret = client_parse_query(cl, query, se->facet_limits, addinfo2);
 896         if (parse_ret == -1)
 897             no_failed_query++;
 898         else if (parse_ret == -2)
 899             no_failed_limit++;
 900         else if (parse_ret < 0)
 901             no_working++; /* other error, such as bad CCL map */
 902         else
 903         {
 904             client_parse_range(cl, startrecs, maxrecs);
 905             client_parse_sort(cl, sp);
 906             client_start_search(cl);
 907             no_working++;
 908         }
 909     }
 910     session_enter(se, "session_search2");
 911     se->clients_starting = 0;
 912     session_leave(se, "session_search2");
 913     if (no_working == 0)
 914     {
 915         if (no_failed_query > 0)
 916         {
 917             *addinfo = "query";
 918             return PAZPAR2_MALFORMED_PARAMETER_VALUE;
 919         }
 920         else if (no_failed_limit > 0)
 921         {
 922             *addinfo = "limit";
 923             return PAZPAR2_MALFORMED_PARAMETER_VALUE;
 924         }
 925         else
 926             return PAZPAR2_NO_TARGETS;
 927     }
 928     return PAZPAR2_NO_ERROR;
 929 }
 930
 931 // Creates a new session_database object for a database
 932 static void session_init_databases_fun(void *context, struct database *db)
 933 {
 934     struct session *se = (struct session *) context;
 935     struct session_database *new = nmem_malloc(se->session_nmem, sizeof(*new));
 936     int i;
 937
 938     new->database = db;
 939
 940     new->map = 0;
 941     assert(db->settings);
 942     new->settings = nmem_malloc(se->session_nmem,
 943                                 sizeof(struct settings *) * db->num_settings);
 944     new->num_settings = db->num_settings;
 945     for (i = 0; i < db->num_settings; i++)
 946     {
 947         struct setting *setting = db->settings[i];
 948         new->settings[i] = setting;
 949     }
 950     new->next = se->databases;
 951     se->databases = new;
 952 }
 953
 954 // Doesn't free memory associated with sdb -- nmem takes care of that
 955 static void session_database_destroy(struct session_database *sdb)
 956 {
 957     sdb->map = 0;
 958 }
 959
 960 // Initialize session_database list -- this represents this session's view
 961 // of the database list -- subject to modification by the settings ws command
 962 void session_init_databases(struct session *se)
 963 {
 964     se->databases = 0;
 965     predef_grep_databases(se, se->service, session_init_databases_fun);
 966 }
 967
 968 // Probably session_init_databases_fun should be refactored instead of
 969 // called here.
 970 static struct session_database *load_session_database(struct session *se,
 971                                                       const char *id)
 972 {
 973     struct database *db = new_database_inherit_settings(id, se->session_nmem, se->service->settings);
 974     session_init_databases_fun((void*) se, db);
 975
 976     // New sdb is head of se->databases list
 977     return se->databases;
 978 }
 979
 980 // Find an existing session database. If not found, load it
 981 static struct session_database *find_session_database(struct session *se,
 982                                                       const char *id)
 983 {
 984     struct session_database *sdb;
 985
 986     for (sdb = se->databases; sdb; sdb = sdb->next)
 987         if (!strcmp(sdb->database->id, id))
 988             return sdb;
 989     return load_session_database(se, id);
 990 }
 991
 992 // Apply a session override to a database
 993 void session_apply_setting(struct session *se, const char *dbname,
 994                            const char *name, const char *value)
 995 {
 996     session_enter(se, "session_apply_setting");
 997     {
 998         struct session_database *sdb = find_session_database(se, dbname);
 999         struct conf_service *service = se->service;
1000         struct setting *s;
1001         int offset = settings_create_offset(service, name);
1002
1003         expand_settings_array(&sdb->settings, &sdb->num_settings, offset,
1004                               se->session_nmem);
1005         // Force later recompute of settings-driven data structures
1006         // (happens when a search starts and client connections are prepared)
1007         if (offset == PZ_XSLT)
1008             sdb->map = 0;
1009         se->settings_modified = 1;
1010         for (s = sdb->settings[offset]; s; s = s->next)
1011             if (!strcmp(s->name, name) &&
1012                 dbname && s->target && !strcmp(dbname, s->target))
1013                 break;
1014         if (!s)
1015         {
1016             s = nmem_malloc(se->session_nmem, sizeof(*s));
1017             s->precedence = 0;
1018             s->target = nmem_strdup(se->session_nmem, dbname);
1019             s->name = nmem_strdup(se->session_nmem, name);
1020             s->next = sdb->settings[offset];
1021             sdb->settings[offset] = s;
1022         }
1023         s->value = nmem_strdup(se->session_nmem, value);
1024     }
1025     session_leave(se, "session_apply_setting");
1026 }
1027
1028 void session_destroy(struct session *se)
1029 {
1030     struct session_database *sdb;
1031     session_log(se, YLOG_LOG, "destroy");
1032     session_use(-1);
1033     session_remove_cached_clients(se);
1034
1035     for (sdb = se->databases; sdb; sdb = sdb->next)
1036         session_database_destroy(sdb);
1037     normalize_cache_destroy(se->normalize_cache);
1038     relevance_destroy(&se->relevance);
1039     reclist_destroy(se->reclist);
1040     xfree(se->mergekey);
1041     xfree(se->rank);
1042     if (nmem_total(se->nmem))
1043         session_log(se, YLOG_DEBUG, "NMEN operation usage %zd", nmem_total(se->nmem));
1044     if (nmem_total(se->session_nmem))
1045         session_log(se, YLOG_DEBUG, "NMEN session usage %zd", nmem_total(se->session_nmem));
1046     facet_limits_destroy(se->facet_limits);
1047     nmem_destroy(se->nmem);
1048     service_destroy(se->service);
1049     yaz_mutex_destroy(&se->session_mutex);
1050 }
1051
1052 size_t session_get_memory_status(struct session *session) {
1053     size_t session_nmem;
1054     if (session == 0)
1055         return 0;
1056     session_enter(session, "session_get_memory_status");
1057     session_nmem = nmem_total(session->nmem);
1058     session_leave(session, "session_get_memory_status");
1059     return session_nmem;
1060 }
1061
1062
1063 struct session *new_session(NMEM nmem, struct conf_service *service,
1064                             unsigned session_id)
1065 {
1066     int i;
1067     struct session *session = nmem_malloc(nmem, sizeof(*session));
1068
1069     char tmp_str[50];
1070
1071     sprintf(tmp_str, "session#%u", session_id);
1072
1073     session->session_id = session_id;
1074     session_log(session, YLOG_DEBUG, "New");
1075     session->service = service;
1076     session->relevance = 0;
1077     session->total_records = 0;
1078     session->number_of_warnings_unknown_elements = 0;
1079     session->number_of_warnings_unknown_metadata = 0;
1080     session->termlists = 0;
1081     session->reclist = reclist_create(nmem);
1082     session->clients_active = 0;
1083     session->clients_cached = 0;
1084     session->settings_modified = 0;
1085     session->session_nmem = nmem;
1086     session->facet_id_list = 0;
1087     session->nmem = nmem_create();
1088     session->databases = 0;
1089     session->sorted_results = 0;
1090     session->facet_limits = 0;
1091     session->mergekey = 0;
1092     session->rank = 0;
1093     session->clients_starting = 0;
1094
1095     for (i = 0; i <= SESSION_WATCH_MAX; i++)
1096     {
1097         session->watchlist[i].data = 0;
1098         session->watchlist[i].fun = 0;
1099     }
1100     session->normalize_cache = normalize_cache_create();
1101     session->session_mutex = 0;
1102     pazpar2_mutex_create(&session->session_mutex, tmp_str);
1103     session_log(session, YLOG_LOG, "create");
1104
1105     session_use(1);
1106     return session;
1107 }
1108
1109 static struct hitsbytarget *hitsbytarget_nb(struct session *se,
1110                                             int *count, NMEM nmem)
1111 {
1112     struct hitsbytarget *res = 0;
1113     struct client_list *l;
1114     size_t sz = 0;
1115
1116     for (l = se->clients_active; l; l = l->next)
1117         sz++;
1118
1119     res = nmem_malloc(nmem, sizeof(*res) * sz);
1120     *count = 0;
1121     for (l = se->clients_active; l; l = l->next)
1122     {
1123         struct client *cl = l->client;
1124         WRBUF w = wrbuf_alloc();
1125         const char *name = session_setting_oneval(client_get_database(cl),
1126                                                   PZ_NAME);
1127         res[*count].id = client_get_id(cl);
1128         res[*count].name = *name ? name : "Unknown";
1129         res[*count].hits = client_get_hits(cl);
1130         res[*count].approximation = client_get_approximation(cl);
1131         res[*count].records = client_get_num_records(cl,
1132                                                      &res[*count].filtered,
1133                                                      0, 0);
1134         res[*count].diagnostic =
1135             client_get_diagnostic(cl, &res[*count].message,
1136                                   &res[*count].addinfo);
1137         res[*count].state = client_get_state_str(cl);
1138         res[*count].connected  = client_get_connection(cl) ? 1 : 0;
1139         session_settings_dump(se, client_get_database(cl), w);
1140         res[*count].settings_xml = nmem_strdup(nmem, wrbuf_cstr(w));
1141         wrbuf_rewind(w);
1142         res[*count].suggestions_xml =
1143             nmem_strdup(nmem, client_get_suggestions_xml(cl, w));
1144
1145         res[*count].query_data =
1146             client_get_query(cl, &res[*count].query_type, nmem);
1147         wrbuf_destroy(w);
1148         (*count)++;
1149     }
1150     return res;
1151 }
1152
1153 struct hitsbytarget *get_hitsbytarget(struct session *se, int *count, NMEM nmem)
1154 {
1155     struct hitsbytarget *p;
1156     session_enter(se, "get_hitsbytarget");
1157     p = hitsbytarget_nb(se, count, nmem);
1158     session_leave(se, "get_hitsbytarget");
1159     return p;
1160 }
1161
1162 // Compares two hitsbytarget nodes by hitcount
1163 static int cmp_ht(const void *p1, const void *p2)
1164 {
1165     const struct hitsbytarget *h1 = p1;
1166     const struct hitsbytarget *h2 = p2;
1167     return h2->hits - h1->hits;
1168 }
1169
1170 // Compares two hitsbytarget nodes by hitcount
1171 static int cmp_ht_approx(const void *p1, const void *p2)
1172 {
1173     const struct hitsbytarget *h1 = p1;
1174     const struct hitsbytarget *h2 = p2;
1175     return h2->approximation - h1->approximation;
1176 }
1177
1178 static int targets_termlist_nb(WRBUF wrbuf, struct session *se, int num,
1179                                NMEM nmem, int version)
1180 {
1181     struct hitsbytarget *ht;
1182     int count, i;
1183
1184     ht = hitsbytarget_nb(se, &count, nmem);
1185     if (version >= 2)
1186         qsort(ht, count, sizeof(struct hitsbytarget), cmp_ht_approx);
1187     else
1188         qsort(ht, count, sizeof(struct hitsbytarget), cmp_ht);
1189     for (i = 0; i < count && i < num && ht[i].hits > 0; i++)
1190     {
1191
1192         // do only print terms which have display names
1193
1194         wrbuf_puts(wrbuf, "<term>\n");
1195
1196         wrbuf_puts(wrbuf, "<id>");
1197         wrbuf_xmlputs(wrbuf, ht[i].id);
1198         wrbuf_puts(wrbuf, "</id>\n");
1199
1200         wrbuf_puts(wrbuf, "<name>");
1201         if (!ht[i].name || !ht[i].name[0])
1202             wrbuf_xmlputs(wrbuf, "NO TARGET NAME");
1203         else
1204             wrbuf_xmlputs(wrbuf, ht[i].name);
1205         wrbuf_puts(wrbuf, "</name>\n");
1206
1207         wrbuf_printf(wrbuf, "<frequency>" ODR_INT_PRINTF "</frequency>\n",
1208                      ht[i].hits);
1209
1210         if (version >= 2) {
1211             // Should not print if we know it isn't a approximation.
1212             wrbuf_printf(wrbuf, "<approximation>" ODR_INT_PRINTF "</approximation>\n", ht[i].approximation);
1213             wrbuf_printf(wrbuf, "<records>%d</records>\n", ht[i].records - ht[i].filtered);
1214             wrbuf_printf(wrbuf, "<filtered>%d</filtered>\n", ht[i].filtered);
1215         }
1216
1217         wrbuf_puts(wrbuf, "<state>");
1218         wrbuf_xmlputs(wrbuf, ht[i].state);
1219         wrbuf_puts(wrbuf, "</state>\n");
1220
1221         wrbuf_printf(wrbuf, "<diagnostic>%d</diagnostic>\n",
1222                      ht[i].diagnostic);
1223         wrbuf_puts(wrbuf, "</term>\n");
1224     }
1225     return count;
1226 }
1227
1228 void perform_termlist(struct http_channel *c, struct session *se,
1229                       const char *name, int num, int version)
1230 {
1231     int j;
1232     NMEM nmem_tmp = nmem_create();
1233     char **names;
1234     int num_names = 0;
1235
1236     if (!name)
1237         name = "*";
1238
1239     nmem_strsplit(nmem_tmp, ",", name, &names, &num_names);
1240
1241     session_enter(se, "perform_termlist");
1242
1243     for (j = 0; j < num_names; j++)
1244     {
1245         const char *tname;
1246         int must_generate_empty = 1; /* bug 5350 */
1247
1248         struct named_termlist *t = se->termlists;
1249         for (; t; t = t->next)
1250         {
1251             tname = t->name;
1252             if (!strcmp(names[j], tname) || !strcmp(names[j], "*"))
1253             {
1254                 struct termlist_score **p = 0;
1255                 int len;
1256
1257                 wrbuf_puts(c->wrbuf, "<list name=\"");
1258                 wrbuf_xmlputs(c->wrbuf, tname);
1259                 wrbuf_puts(c->wrbuf, "\">\n");
1260                 must_generate_empty = 0;
1261
1262                 p = termlist_highscore(t->termlist, &len, nmem_tmp);
1263                 if (p)
1264                 {
1265                     int i;
1266                     for (i = 0; i < len && i < num; i++)
1267                     {
1268                         // prevent sending empty term elements
1269                         if (!p[i]->display_term || !p[i]->display_term[0])
1270                             continue;
1271
1272                         wrbuf_puts(c->wrbuf, "<term>");
1273                         wrbuf_puts(c->wrbuf, "<name>");
1274                         wrbuf_xmlputs(c->wrbuf, p[i]->display_term);
1275                         wrbuf_puts(c->wrbuf, "</name>");
1276                         wrbuf_printf(c->wrbuf,
1277                                      "<frequency>%d</frequency>",
1278                                      p[i]->frequency);
1279                         wrbuf_puts(c->wrbuf, "</term>\n");
1280                     }
1281                 }
1282                 wrbuf_puts(c->wrbuf, "</list>\n");
1283             }
1284         }
1285         tname = "xtargets";
1286         if (!strcmp(names[j], tname) || !strcmp(names[j], "*"))
1287         {
1288             wrbuf_puts(c->wrbuf, "<list name=\"");
1289             wrbuf_xmlputs(c->wrbuf, tname);
1290             wrbuf_puts(c->wrbuf, "\">\n");
1291
1292             targets_termlist_nb(c->wrbuf, se, num, c->nmem, version);
1293             wrbuf_puts(c->wrbuf, "</list>\n");
1294             must_generate_empty = 0;
1295         }
1296         if (must_generate_empty)
1297         {
1298             wrbuf_puts(c->wrbuf, "<list name=\"");
1299             wrbuf_xmlputs(c->wrbuf, names[j]);
1300             wrbuf_puts(c->wrbuf, "\"/>\n");
1301         }
1302     }
1303     session_leave(se, "perform_termlist");
1304     nmem_destroy(nmem_tmp);
1305 }
1306
1307 #ifdef MISSING_HEADERS
1308 void report_nmem_stats(void)
1309 {
1310     size_t in_use, is_free;
1311
1312     nmem_get_memory_in_use(&in_use);
1313     nmem_get_memory_free(&is_free);
1314
1315     yaz_log(YLOG_LOG, "nmem stat: use=%ld free=%ld",
1316             (long) in_use, (long) is_free);
1317 }
1318 #endif
1319
1320 struct record_cluster *show_single_start(struct session *se, const char *id,
1321                                          struct record_cluster **prev_r,
1322                                          struct record_cluster **next_r)
1323 {
1324     struct record_cluster *r = 0;
1325
1326     session_enter(se, "show_single_start");
1327     *prev_r = 0;
1328     *next_r = 0;
1329     reclist_limit(se->reclist, se, 1);
1330
1331     reclist_enter(se->reclist);
1332     while ((r = reclist_read_record(se->reclist)))
1333     {
1334         if (!strcmp(r->recid, id))
1335         {
1336             *next_r = reclist_read_record(se->reclist);
1337             break;
1338         }
1339         *prev_r = r;
1340     }
1341     reclist_leave(se->reclist);
1342     if (!r)
1343         session_leave(se, "show_single_start");
1344     return r;
1345 }
1346
1347 void show_single_stop(struct session *se, struct record_cluster *rec)
1348 {
1349     session_leave(se, "show_single_stop");
1350 }
1351
1352
1353 int session_fetch_more(struct session *se)
1354 {
1355     struct client_list *l;
1356     int ret = 0;
1357
1358     for (l = se->clients_active; l; l = l->next)
1359     {
1360         struct client *cl = l->client;
1361         if (client_get_state(cl) == Client_Idle)
1362         {
1363             if (client_fetch_more(cl))
1364             {
1365                 session_log(se, YLOG_LOG, "%s: more to fetch",
1366                             client_get_id(cl));
1367                 ret = 1;
1368             }
1369             else
1370             {
1371                 int filtered;
1372                 int ingest_failures;
1373                 int record_failures;
1374                 int num = client_get_num_records(
1375                     cl, &filtered, &ingest_failures, &record_failures);
1376
1377                 session_log(se, YLOG_LOG, "%s: hits=" ODR_INT_PRINTF
1378                             " fetched=%d filtered=%d",
1379                             client_get_id(cl),
1380                             client_get_hits(cl),
1381                             num, filtered);
1382                 if (ingest_failures || record_failures)
1383                 {
1384                     session_log(se, YLOG_WARN, "%s:"
1385                                 " ingest failures=%d record failures=%d",
1386                                 client_get_id(cl),
1387                                 ingest_failures, record_failures);
1388                 }
1389             }
1390         }
1391         else
1392         {
1393             session_log(se, YLOG_LOG, "%s: no fetch due to state=%s",
1394                         client_get_id(cl), client_get_state_str(cl));
1395         }
1396
1397     }
1398     return ret;
1399 }
1400
1401 struct record_cluster **show_range_start(struct session *se,
1402                                          struct reclist_sortparms *sp,
1403                                          int start, int *num, int *total,
1404                                          Odr_int *sumhits, Odr_int *approx_hits,
1405                                          void (*show_records_ready)(void *data),
1406                                          struct http_channel *chan)
1407 {
1408     struct record_cluster **recs = 0;
1409     struct reclist_sortparms *spp;
1410     struct client_list *l;
1411     int i;
1412 #if USE_TIMING
1413     yaz_timing_t t = yaz_timing_create();
1414 #endif
1415     session_enter(se, "show_range_start");
1416     *sumhits = 0;
1417     *approx_hits = 0;
1418     *total = 0;
1419     reclist_limit(se->reclist, se, 0);
1420     if (se->relevance)
1421     {
1422         for (spp = sp; spp; spp = spp->next)
1423             if (spp->type == Metadata_type_relevance)
1424             {
1425                 relevance_prepare_read(se->relevance, se->reclist);
1426                 break;
1427             }
1428         for (l = se->clients_active; l; l = l->next) {
1429             *sumhits += client_get_hits(l->client);
1430             *approx_hits += client_get_approximation(l->client);
1431         }
1432     }
1433     reclist_sort(se->reclist, sp);
1434
1435     reclist_enter(se->reclist);
1436     *total = reclist_get_num_records(se->reclist);
1437
1438     for (l = se->clients_active; l; l = l->next)
1439         client_update_show_stat(l->client, 0);
1440
1441     for (i = 0; i < start; i++)
1442     {
1443         struct record_cluster *r = reclist_read_record(se->reclist);
1444         if (!r)
1445         {
1446             *num = 0;
1447             break;
1448         }
1449         else
1450         {
1451             struct record *rec = r->records;
1452             for (;rec; rec = rec->next)
1453                 client_update_show_stat(rec->client, 1);
1454         }
1455     }
1456     recs = nmem_malloc(se->nmem, (*num > 0 ? *num : 1) * sizeof(*recs));
1457     for (i = 0; i < *num; i++)
1458     {
1459         struct record_cluster *r = reclist_read_record(se->reclist);
1460         if (!r)
1461         {
1462             *num = i;
1463             break;
1464         }
1465         else
1466         {
1467             struct record *rec = r->records;
1468             for (;rec; rec = rec->next)
1469                 client_update_show_stat(rec->client, 1);
1470             recs[i] = r;
1471         }
1472     }
1473     reclist_leave(se->reclist);
1474 #if USE_TIMING
1475     yaz_timing_stop(t);
1476     session_log(se, YLOG_LOG, "show %6.5f %3.2f %3.2f",
1477             yaz_timing_get_real(t), yaz_timing_get_user(t),
1478             yaz_timing_get_sys(t));
1479     yaz_timing_destroy(&t);
1480 #endif
1481
1482     if (!session_fetch_more(se))
1483         session_log(se, YLOG_LOG, "can not fetch more");
1484     else
1485     {
1486         show_range_stop(se, recs);
1487         session_log(se, YLOG_LOG, "fetching more in progress");
1488         if (session_set_watch(se, SESSION_WATCH_SHOW,
1489                               show_records_ready, chan, chan))
1490         {
1491             session_log(se, YLOG_WARN, "Ignoring show block");
1492             session_enter(se, "show_range_start");
1493         }
1494         else
1495         {
1496             session_log(se, YLOG_LOG, "session watch OK");
1497             return 0;
1498         }
1499     }
1500     return recs;
1501 }
1502
1503 void show_range_stop(struct session *se, struct record_cluster **recs)
1504 {
1505     session_leave(se, "show_range_stop");
1506 }
1507
1508 void statistics(struct session *se, struct statistics *stat)
1509 {
1510     struct client_list *l;
1511     int count = 0;
1512
1513     memset(stat, 0, sizeof(*stat));
1514     stat->num_hits = 0;
1515     for (l = se->clients_active; l; l = l->next)
1516     {
1517         struct client *cl = l->client;
1518         if (!client_get_connection(cl))
1519             stat->num_no_connection++;
1520         stat->num_hits += client_get_hits(cl);
1521         switch (client_get_state(cl))
1522         {
1523         case Client_Connecting: stat->num_connecting++; break;
1524         case Client_Working: stat->num_working++; break;
1525         case Client_Idle: stat->num_idle++; break;
1526         case Client_Failed: stat->num_failed++; break;
1527         case Client_Error: stat->num_error++; break;
1528         default: break;
1529         }
1530         count++;
1531     }
1532     stat->num_records = se->total_records;
1533
1534     stat->num_clients = count;
1535 }
1536
1537 static struct record_metadata *record_metadata_init(
1538     NMEM nmem, const char *value, const char *norm,
1539     enum conf_metadata_type type,
1540     struct _xmlAttr *attr)
1541 {
1542     struct record_metadata *rec_md = record_metadata_create(nmem);
1543     struct record_metadata_attr **attrp = &rec_md->attributes;
1544
1545     for (; attr; attr = attr->next)
1546     {
1547         if (attr->children && attr->children->content)
1548         {
1549             if (strcmp((const char *) attr->name, "type")
1550                 && strcmp((const char *) attr->name, "empty"))
1551             {  /* skip the "type" + "empty" attribute..
1552                   The "Type" is already part of the element in output
1553                   (md-%s) and so repeating it here is redundant */
1554                 *attrp = nmem_malloc(nmem, sizeof(**attrp));
1555                 (*attrp)->name =
1556                     nmem_strdup(nmem, (const char *) attr->name);
1557                 (*attrp)->value =
1558                     nmem_strdup(nmem, (const char *) attr->children->content);
1559                 attrp = &(*attrp)->next;
1560             }
1561         }
1562     }
1563     *attrp = 0;
1564
1565     switch (type)
1566     {
1567     case Metadata_type_generic:
1568     case Metadata_type_skiparticle:
1569         if (norm)
1570         {
1571             rec_md->data.text.disp = nmem_strdup(nmem, value);
1572             rec_md->data.text.norm = nmem_strdup(nmem, norm);
1573         }
1574         else
1575         {
1576             if (strstr(value, "://")) /* looks like a URL */
1577                 rec_md->data.text.disp = nmem_strdup(nmem, value);
1578             else
1579                 rec_md->data.text.disp =
1580                     normalize7bit_generic(nmem_strdup(nmem, value), " ,/.:([");
1581             rec_md->data.text.norm = rec_md->data.text.disp;
1582         }
1583         rec_md->data.text.sort = 0;
1584         rec_md->data.text.snippet = 0;
1585         break;
1586     case Metadata_type_year:
1587     case Metadata_type_date:
1588     {
1589         int first, last;
1590         int longdate = 0;
1591
1592         if (type == Metadata_type_date)
1593             longdate = 1;
1594         if (extract7bit_dates((char *) value, &first, &last, longdate) < 0)
1595             return 0;
1596
1597         rec_md->data.number.min = first;
1598         rec_md->data.number.max = last;
1599     }
1600     break;
1601     case Metadata_type_float:
1602         rec_md->data.fnumber = atof(value);
1603         break;
1604     case Metadata_type_relevance:
1605     case Metadata_type_position:
1606     case Metadata_type_retrieval:
1607         return 0;
1608     }
1609     return rec_md;
1610 }
1611
1612 static void mergekey_norm_wr(pp2_charset_fact_t charsets,
1613                              WRBUF norm_wr, const char *value)
1614 {
1615     const char *norm_str;
1616     pp2_charset_token_t prt =
1617         pp2_charset_token_create(charsets, "mergekey");
1618
1619     pp2_charset_token_first(prt, value, 0);
1620     while ((norm_str = pp2_charset_token_next(prt)))
1621     {
1622         if (*norm_str)
1623         {
1624             if (wrbuf_len(norm_wr))
1625                 wrbuf_puts(norm_wr, " ");
1626             wrbuf_puts(norm_wr, norm_str);
1627         }
1628     }
1629     pp2_charset_token_destroy(prt);
1630 }
1631
1632 static int get_mergekey_from_doc(xmlDoc *doc, xmlNode *root, const char *name,
1633                                  struct conf_service *service, WRBUF norm_wr)
1634 {
1635     xmlNode *n;
1636     int no_found = 0;
1637     for (n = root->children; n; n = n->next)
1638     {
1639         if (n->type != XML_ELEMENT_NODE)
1640             continue;
1641         if (!strcmp((const char *) n->name, "metadata"))
1642         {
1643             const char *type = yaz_xml_get_prop(n, "type");
1644             if (type == NULL) {
1645                 yaz_log(YLOG_FATAL, "Missing type attribute on metadata element. Skipping!");
1646             }
1647             else if (!strcmp(name, (const char *) type))
1648             {
1649                 xmlChar *value = xmlNodeListGetString(doc, n->children, 1);
1650                 if (value && *value)
1651                 {
1652                     if (wrbuf_len(norm_wr) > 0)
1653                         wrbuf_puts(norm_wr, " ");
1654                     wrbuf_puts(norm_wr, name);
1655                     mergekey_norm_wr(service->charsets, norm_wr,
1656                                      (const char *) value);
1657                     no_found++;
1658                 }
1659                 if (value)
1660                     xmlFree(value);
1661             }
1662         }
1663     }
1664     return no_found;
1665 }
1666
1667 static const char *get_mergekey(xmlDoc *doc, xmlNode *root,
1668                                 struct client *cl, int record_no,
1669                                 struct conf_service *service, NMEM nmem,
1670                                 const char *session_mergekey)
1671 {
1672     char *mergekey_norm = 0;
1673     WRBUF norm_wr = wrbuf_alloc();
1674     const char *mergekey;
1675
1676     if (session_mergekey)
1677     {
1678         int i, num = 0;
1679         char **values = 0;
1680         nmem_strsplit_escape2(nmem, ",", session_mergekey, &values,
1681                               &num, 1, '\\', 1);
1682
1683         for (i = 0; i < num; i++)
1684             get_mergekey_from_doc(doc, root, values[i], service, norm_wr);
1685     }
1686     else if ((mergekey = yaz_xml_get_prop(root, "mergekey")))
1687     {
1688         mergekey_norm_wr(service->charsets, norm_wr, mergekey);
1689     }
1690     else
1691     {
1692         /* no mergekey defined in XSL. Look for mergekey metadata instead */
1693         int field_id;
1694         for (field_id = 0; field_id < service->num_metadata; field_id++)
1695         {
1696             struct conf_metadata *ser_md = &service->metadata[field_id];
1697             if (ser_md->mergekey != Metadata_mergekey_no)
1698             {
1699                 int r = get_mergekey_from_doc(doc, root, ser_md->name,
1700                                               service, norm_wr);
1701                 if (r == 0 && ser_md->mergekey == Metadata_mergekey_required)
1702                 {
1703                     /* no mergekey on this one and it is required..
1704                        Generate unique key instead */
1705                     wrbuf_rewind(norm_wr);
1706                     break;
1707                 }
1708             }
1709         }
1710     }
1711
1712     /* generate unique key if none is not generated already or is empty */
1713     if (wrbuf_len(norm_wr) == 0)
1714     {
1715         wrbuf_printf(norm_wr, "position: %s-%06d",
1716                      client_get_id(cl), record_no);
1717     }
1718     else
1719     {
1720         const char *lead = "content: ";
1721         wrbuf_insert(norm_wr, 0, lead, strlen(lead));
1722     }
1723     if (wrbuf_len(norm_wr) > 0)
1724         mergekey_norm = nmem_strdup(nmem, wrbuf_cstr(norm_wr));
1725     wrbuf_destroy(norm_wr);
1726     return mergekey_norm;
1727 }
1728
1729 /** \brief see if metadata for pz:recordfilter exists
1730     \param root xml root element of normalized record
1731     \param sdb session database for client
1732     \retval 0 if there is no metadata for pz:recordfilter
1733     \retval 1 if there is metadata for pz:recordfilter
1734
1735     If there is no pz:recordfilter defined, this function returns 1
1736     as well.
1737 */
1738
1739 static int check_record_filter(xmlNode *root, struct session_database *sdb)
1740 {
1741     int match = 0;
1742     xmlNode *n;
1743     const char *s;
1744     s = session_setting_oneval(sdb, PZ_RECORDFILTER);
1745
1746     if (!s || !*s)
1747         return 1;
1748
1749     for (n = root->children; n; n = n->next)
1750     {
1751         if (n->type != XML_ELEMENT_NODE)
1752             continue;
1753         if (!strcmp((const char *) n->name, "metadata"))
1754         {
1755             const char *type = yaz_xml_get_prop(n, "type");
1756             if (type)
1757             {
1758                 size_t len;
1759                 int substring;
1760                 const char *eq;
1761
1762                 if ((eq = strchr(s, '=')))
1763                     substring = 0;
1764                 else if ((eq = strchr(s, '~')))
1765                     substring = 1;
1766                 if (eq)
1767                     len = eq - s;
1768                 else
1769                     len = strlen(s);
1770                 if (len == strlen((const char *)type) &&
1771                     !memcmp((const char *) type, s, len))
1772                 {
1773                     xmlChar *value = xmlNodeGetContent(n);
1774                     if (value && *value)
1775                     {
1776                         if (!eq ||
1777                             (substring && strstr((const char *) value, eq+1)) ||
1778                             (!substring && !strcmp((const char *) value, eq + 1)))
1779                             match = 1;
1780                     }
1781                     xmlFree(value);
1782                 }
1783             }
1784         }
1785     }
1786     return match;
1787 }
1788
1789 static int ingest_to_cluster(struct client *cl,
1790                              WRBUF wrbuf_disp,
1791                              WRBUF wrbuf_norm,
1792                              xmlDoc *xdoc,
1793                              xmlNode *root,
1794                              int record_no,
1795                              struct record_metadata_attr *mergekey);
1796
1797 static int ingest_sub_record(struct client *cl, xmlDoc *xdoc, xmlNode *root,
1798                              int record_no, NMEM nmem,
1799                              struct session_database *sdb,
1800                              struct record_metadata_attr *mergekeys)
1801 {
1802     int ret = 0;
1803     struct session *se = client_get_session(cl);
1804     WRBUF wrbuf_disp, wrbuf_norm;
1805
1806     if (!check_record_filter(root, sdb))
1807     {
1808         session_log(se, YLOG_LOG,
1809                     "Filtered out record no %d from %s",
1810                     record_no, sdb->database->id);
1811         return 0;
1812     }
1813     wrbuf_disp = wrbuf_alloc();
1814     wrbuf_norm = wrbuf_alloc();
1815     session_enter(se, "ingest_sub_record");
1816     if (client_get_session(cl) == se && se->relevance)
1817         ret = ingest_to_cluster(cl, wrbuf_disp, wrbuf_norm,
1818                                 xdoc, root, record_no, mergekeys);
1819     session_leave(se, "ingest_sub_record");
1820     wrbuf_destroy(wrbuf_norm);
1821     wrbuf_destroy(wrbuf_disp);
1822     return ret;
1823 }
1824
1825 /** \brief ingest XML record
1826     \param cl client holds the result set for record
1827     \param rec record buffer (0 terminated)
1828     \param record_no record position (1, 2, ..)
1829     \param nmem working NMEM
1830     \retval 0 OK
1831     \retval -1 failure
1832     \retval -2 Filtered
1833 */
1834 int ingest_record(struct client *cl, const char *rec,
1835                   int record_no, NMEM nmem)
1836 {
1837     struct session *se = client_get_session(cl);
1838     struct session_database *sdb = client_get_database(cl);
1839     struct conf_service *service = se->service;
1840     xmlDoc *xdoc = normalize_record(se, sdb, service, rec, nmem);
1841     int r = ingest_xml_record(cl, xdoc, record_no, nmem, 0);
1842     client_store_xdoc(cl, record_no, xdoc);
1843     return r;
1844 }
1845
1846 int ingest_xml_record(struct client *cl, xmlDoc *xdoc,
1847                       int record_no, NMEM nmem, int cached_copy)
1848 {
1849     struct session *se = client_get_session(cl);
1850     struct session_database *sdb = client_get_database(cl);
1851     struct conf_service *service = se->service;
1852     xmlNode *root;
1853     int r = 0;
1854     if (!xdoc)
1855         return -1;
1856
1857     if (global_parameters.dump_records)
1858     {
1859         session_log(se, YLOG_LOG, "Normalized record from %s",
1860                     sdb->database->id);
1861         log_xml_doc(xdoc);
1862     }
1863
1864     root = xmlDocGetRootElement(xdoc);
1865
1866     if (!strcmp((const char *) root->name, "cluster"))
1867     {
1868         int no_merge_keys = 0;
1869         int no_merge_dups = 0;
1870         xmlNode *sroot;
1871         struct record_metadata_attr *mk = 0;
1872
1873         for (sroot = root->children; sroot; sroot = sroot->next)
1874             if (sroot->type == XML_ELEMENT_NODE &&
1875                 !strcmp((const char *) sroot->name, "record"))
1876             {
1877                 struct record_metadata_attr **mkp;
1878                 const char *mergekey_norm =
1879                     get_mergekey(xdoc, sroot, cl, record_no, service, nmem,
1880                                  se->mergekey);
1881                 if (!mergekey_norm)
1882                 {
1883                     r = -1;
1884                     break;
1885                 }
1886                 for (mkp = &mk; *mkp; mkp = &(*mkp)->next)
1887                     if (!strcmp((*mkp)->value, mergekey_norm))
1888                         break;
1889                 if (!*mkp)
1890                 {
1891                     *mkp = (struct record_metadata_attr*)
1892                         nmem_malloc(nmem, sizeof(**mkp));
1893                     (*mkp)->name = 0;
1894                     (*mkp)->value = nmem_strdup(nmem, mergekey_norm);
1895                     (*mkp)->next = 0;
1896                     no_merge_keys++;
1897                 }
1898                 else
1899                     no_merge_dups++;
1900             }
1901         if (no_merge_keys > 1 || no_merge_dups > 0)
1902         {
1903             yaz_log(YLOG_LOG, "Got %d mergekeys, %d dups for position %d",
1904                     no_merge_keys, no_merge_dups, record_no);
1905         }
1906         for (sroot = root->children; !r && sroot; sroot = sroot->next)
1907             if (sroot->type == XML_ELEMENT_NODE &&
1908                 !strcmp((const char *) sroot->name, "record"))
1909             {
1910                 if (!cached_copy)
1911                     insert_settings_values(sdb, xdoc, root, service);
1912                 r = ingest_sub_record(cl, xdoc, sroot, record_no, nmem, sdb,
1913                                       mk);
1914             }
1915     }
1916     else if (!strcmp((const char *) root->name, "record"))
1917     {
1918         const char *mergekey_norm =
1919             get_mergekey(xdoc, root, cl, record_no, service, nmem,
1920                          se->mergekey);
1921         if (mergekey_norm)
1922         {
1923             struct record_metadata_attr *mk = (struct record_metadata_attr*)
1924                 nmem_malloc(nmem, sizeof(*mk));
1925             mk->name = 0;
1926             mk->value = nmem_strdup(nmem, mergekey_norm);
1927             mk->next = 0;
1928
1929             if (!cached_copy)
1930                 insert_settings_values(sdb, xdoc, root, service);
1931             r = ingest_sub_record(cl, xdoc, root, record_no, nmem, sdb, mk);
1932         }
1933     }
1934     else
1935     {
1936         session_log(se, YLOG_WARN, "Bad pz root element: %s",
1937                     (const char *) root->name);
1938         r = -1;
1939     }
1940     return r;
1941 }
1942
1943
1944 //    struct conf_metadata *ser_md = &service->metadata[md_field_id];
1945 //    struct record_metadata *rec_md = record->metadata[md_field_id];
1946 static int match_metadata_local(struct conf_service *service,
1947                                 struct conf_metadata *ser_md,
1948                                 struct record_metadata *rec_md0,
1949                                 char **values, int num_v)
1950 {
1951     int i;
1952     struct record_metadata *rec_md = rec_md0;
1953     WRBUF val_wr = 0;
1954     WRBUF text_wr = wrbuf_alloc();
1955     for (i = 0; i < num_v; )
1956     {
1957         if (rec_md)
1958         {
1959             if (ser_md->type == Metadata_type_year
1960                 || ser_md->type == Metadata_type_date)
1961             {
1962                 int y = atoi(values[i]);
1963                 if (y >= rec_md->data.number.min
1964                     && y <= rec_md->data.number.max)
1965                     break;
1966             }
1967             else
1968             {
1969                 if (!val_wr)
1970                 {
1971                     val_wr = wrbuf_alloc();
1972                     mergekey_norm_wr(service->charsets, val_wr, values[i]);
1973                 }
1974                 wrbuf_rewind(text_wr);
1975                 mergekey_norm_wr(service->charsets, text_wr,
1976                                  rec_md->data.text.disp);
1977                 if (!strcmp(wrbuf_cstr(val_wr), wrbuf_cstr(text_wr)))
1978                     break;
1979             }
1980             rec_md = rec_md->next;
1981         }
1982         else
1983         {
1984             rec_md = rec_md0;
1985             wrbuf_destroy(val_wr);
1986             val_wr = 0;
1987             i++;
1988         }
1989     }
1990     wrbuf_destroy(val_wr);
1991     wrbuf_destroy(text_wr);
1992     return i < num_v ? 1 : 0;
1993 }
1994
1995 int session_check_cluster_limit(struct session *se, struct record_cluster *rec)
1996 {
1997     int i;
1998     struct conf_service *service = se->service;
1999     int ret = 1;
2000     const char *name;
2001     const char *value;
2002     NMEM nmem_tmp = nmem_create();
2003
2004     for (i = 0; (name = facet_limits_get(se->facet_limits, i, &value)); i++)
2005     {
2006         int j;
2007         for (j = 0; j < service->num_metadata; j++)
2008         {
2009             struct conf_metadata *md = service->metadata + j;
2010             if (!strcmp(md->name, name) && md->limitcluster)
2011             {
2012                 char **values = 0;
2013                 int num = 0;
2014                 int md_field_id =
2015                     conf_service_metadata_field_id(service,
2016                                                    md->limitcluster);
2017
2018                 if (md_field_id < 0)
2019                 {
2020                     ret = 0;
2021                     break;
2022                 }
2023
2024                 nmem_strsplit_escape2(nmem_tmp, "|", value, &values,
2025                                       &num, 1, '\\', 1);
2026
2027                 if (!match_metadata_local(service,
2028                                           &service->metadata[md_field_id],
2029                                           rec->metadata[md_field_id],
2030                                           values, num))
2031                 {
2032                     ret = 0;
2033                     break;
2034                 }
2035             }
2036         }
2037     }
2038     nmem_destroy(nmem_tmp);
2039     return ret;
2040 }
2041
2042 // Skip record on non-zero
2043 static int check_limit_local(struct client *cl,
2044                              struct record *record,
2045                              int record_no)
2046 {
2047     int skip_record = 0;
2048     struct session *se = client_get_session(cl);
2049     struct conf_service *service = se->service;
2050     NMEM nmem_tmp = nmem_create();
2051     struct session_database *sdb = client_get_database(cl);
2052     int l = 0;
2053     while (!skip_record)
2054     {
2055         int md_field_id;
2056         char **values = 0;
2057         int num_v = 0;
2058         const char *name =
2059             client_get_facet_limit_local(cl, sdb, &l, nmem_tmp,
2060                                          &num_v, &values);
2061         if (!name)
2062             break;
2063
2064         if (!strcmp(name, "*"))
2065         {
2066             for (md_field_id = 0; md_field_id < service->num_metadata;
2067                  md_field_id++)
2068             {
2069                 if (match_metadata_local(
2070                         service,
2071                         &service->metadata[md_field_id],
2072                         record->metadata[md_field_id],
2073                         values, num_v))
2074                     break;
2075             }
2076             if (md_field_id == service->num_metadata)
2077                 skip_record = 1;
2078         }
2079         else
2080         {
2081             md_field_id = conf_service_metadata_field_id(service, name);
2082             if (md_field_id < 0)
2083             {
2084                 skip_record = 1;
2085                 break;
2086             }
2087             if (!match_metadata_local(
2088                     service,
2089                     &service->metadata[md_field_id],
2090                     record->metadata[md_field_id],
2091                     values, num_v))
2092             {
2093                 skip_record = 1;
2094             }
2095         }
2096     }
2097     nmem_destroy(nmem_tmp);
2098     return skip_record;
2099 }
2100
2101 static int ingest_to_cluster(struct client *cl,
2102                              WRBUF wrbuf_disp,
2103                              WRBUF wrbuf_norm,
2104                              xmlDoc *xdoc,
2105                              xmlNode *root,
2106                              int record_no,
2107                              struct record_metadata_attr *merge_keys)
2108 {
2109     xmlNode *n;
2110     struct session *se = client_get_session(cl);
2111     struct conf_service *service = se->service;
2112     int term_factor = 1;
2113     struct record_cluster *cluster;
2114     struct record_metadata **metadata0;
2115     struct session_database *sdb = client_get_database(cl);
2116     NMEM ingest_nmem = 0;
2117     char **rank_values = 0;
2118     int rank_num = 0;
2119     struct record *record = record_create(se->nmem,
2120                                           service->num_metadata,
2121                                           service->num_sortkeys, cl,
2122                                           record_no);
2123
2124     for (n = root->children; n; n = n->next)
2125     {
2126         if (n->type != XML_ELEMENT_NODE)
2127             continue;
2128         if (!strcmp((const char *) n->name, "metadata"))
2129         {
2130             struct conf_metadata *ser_md = 0;
2131             struct record_metadata **wheretoput = 0;
2132             struct record_metadata *rec_md = 0;
2133             int md_field_id = -1;
2134             xmlChar *value0;
2135             const char *type = yaz_xml_get_prop(n, "type");
2136
2137             if (!type)
2138                 continue;
2139
2140             md_field_id
2141                 = conf_service_metadata_field_id(service, (const char *) type);
2142             if (md_field_id < 0)
2143             {
2144                 if (se->number_of_warnings_unknown_metadata == 0)
2145                 {
2146                     session_log(se, YLOG_WARN,
2147                             "Ignoring unknown metadata element: %s", type);
2148                 }
2149                 se->number_of_warnings_unknown_metadata++;
2150                 continue;
2151             }
2152
2153             wrbuf_rewind(wrbuf_disp);
2154             value0 = xmlNodeListGetString(xdoc, n->children, 1);
2155             if (!value0 || !*value0)
2156             {
2157                 const char *empty = yaz_xml_get_prop(n, "empty");
2158                 if (!empty)
2159                     continue;
2160                 wrbuf_puts(wrbuf_disp, (const char *) empty);
2161             }
2162             else
2163             {
2164                 wrbuf_puts(wrbuf_disp, (const char *) value0);
2165             }
2166             if (value0)
2167                 xmlFree(value0);
2168             ser_md = &service->metadata[md_field_id];
2169
2170             // non-merged metadata
2171             rec_md = record_metadata_init(se->nmem, wrbuf_cstr(wrbuf_disp), 0,
2172                                           ser_md->type, n->properties);
2173             if (!rec_md)
2174             {
2175                 session_log(se, YLOG_WARN, "bad metadata data '%s' "
2176                             "for element '%s'", wrbuf_cstr(wrbuf_disp), type);
2177                 continue;
2178             }
2179
2180             if (ser_md->type == Metadata_type_generic)
2181             {
2182                 WRBUF w = wrbuf_alloc();
2183                 if (relevance_snippet(se->relevance,
2184                                       wrbuf_cstr(wrbuf_disp), ser_md->name, w))
2185                     rec_md->data.text.snippet = nmem_strdup(se->nmem,
2186                                                             wrbuf_cstr(w));
2187                 wrbuf_destroy(w);
2188             }
2189
2190
2191             wheretoput = &record->metadata[md_field_id];
2192             while (*wheretoput)
2193                 wheretoput = &(*wheretoput)->next;
2194             *wheretoput = rec_md;
2195         }
2196     }
2197
2198     if (check_limit_local(cl, record, record_no))
2199     {
2200         return -2;
2201     }
2202     cluster = reclist_insert(se->reclist, se->relevance, service, record,
2203                              merge_keys, &se->total_merged);
2204     if (!cluster)
2205     {
2206         return 0; // complete match with existing record
2207     }
2208
2209     {
2210         const char *use_term_factor_str =
2211             session_setting_oneval(sdb, PZ_TERMLIST_TERM_FACTOR);
2212         if (use_term_factor_str && use_term_factor_str[0] == '1')
2213         {
2214             int maxrecs = client_get_maxrecs(cl);
2215             int hits = (int) client_get_hits(cl);
2216             term_factor = MAX(hits, maxrecs) /  MAX(1, maxrecs);
2217             assert(term_factor >= 1);
2218             session_log(se, YLOG_DEBUG, "Using term factor: %d (%d / %d)",
2219                         term_factor, MAX(hits, maxrecs), MAX(1, maxrecs));
2220         }
2221     }
2222
2223     if (global_parameters.dump_records)
2224         session_log(se, YLOG_LOG, "Cluster id %s from %s (#%d)", cluster->recid,
2225                     sdb->database->id, record_no);
2226
2227     // original metadata, to check if first existence of a field
2228     metadata0 = xmalloc(sizeof(*metadata0) * service->num_metadata);
2229     memcpy(metadata0, cluster->metadata,
2230            sizeof(*metadata0) * service->num_metadata);
2231
2232     ingest_nmem = nmem_create();
2233     if (se->rank)
2234     {
2235         yaz_log(YLOG_LOG, "local in sort : %s", se->rank);
2236         nmem_strsplit_escape2(ingest_nmem, ",", se->rank, &rank_values,
2237                               &rank_num, 1, '\\', 1);
2238     }
2239
2240     // now parsing XML record and adding data to cluster or record metadata
2241     for (n = root->children; n; n = n->next)
2242     {
2243         if (n->type != XML_ELEMENT_NODE)
2244             continue;
2245         if (!strcmp((const char *) n->name, "metadata"))
2246         {
2247             struct conf_metadata *ser_md = 0;
2248             struct conf_sortkey *ser_sk = 0;
2249             struct record_metadata **wheretoput = 0;
2250             struct record_metadata *rec_md = 0;
2251             int md_field_id = -1;
2252             int sk_field_id = -1;
2253             const char *rank = 0;
2254             const char *xml_rank = 0;
2255             const char *type = 0;
2256             xmlChar *value0;
2257
2258             type = yaz_xml_get_prop(n, "type");
2259             if (!type)
2260                 continue;
2261
2262             md_field_id
2263                 = conf_service_metadata_field_id(service, (const char *) type);
2264             if (md_field_id < 0)
2265                 continue;
2266
2267             ser_md = &service->metadata[md_field_id];
2268
2269             if (ser_md->sortkey_offset >= 0)
2270             {
2271                 sk_field_id = ser_md->sortkey_offset;
2272                 ser_sk = &service->sortkeys[sk_field_id];
2273             }
2274
2275             wrbuf_rewind(wrbuf_disp);
2276             wrbuf_rewind(wrbuf_norm);
2277
2278             value0 = xmlNodeListGetString(xdoc, n->children, 1);
2279             if (!value0 || !*value0)
2280             {
2281                 if (value0)
2282                     xmlFree(value0);
2283                 continue;
2284             }
2285
2286             if (ser_md->icurule)
2287             {
2288                 run_icu(se, ser_md->icurule, (const char *) value0,
2289                         wrbuf_norm, wrbuf_disp);
2290                 yaz_log(YLOG_LOG, "run_icu input=%s norm=%s disp=%s",
2291                         (const char *) value0,
2292                         wrbuf_cstr(wrbuf_norm), wrbuf_cstr(wrbuf_disp));
2293                 rec_md = record_metadata_init(se->nmem, wrbuf_cstr(wrbuf_disp),
2294                                               wrbuf_cstr(wrbuf_norm),
2295                                               ser_md->type, 0);
2296             }
2297             else
2298             {
2299                 wrbuf_puts(wrbuf_disp, (const char *) value0);
2300                 rec_md = record_metadata_init(se->nmem, wrbuf_cstr(wrbuf_disp),
2301                                               0,
2302                                               ser_md->type, 0);
2303             }
2304
2305             xmlFree(value0);
2306
2307             // see if the field was not in cluster already (from beginning)
2308             if (!rec_md)
2309                 continue;
2310
2311             if (rank_num)
2312             {
2313                 int i;
2314                 for (i = 0; i < rank_num; i++)
2315                 {
2316                     const char *val = rank_values[i];
2317                     const char *cp = strchr(val, '=');
2318                     if (!cp)
2319                         continue;
2320                     if ((cp - val) == strlen((const char *) type)
2321                         && !memcmp(val, type, cp - val))
2322                     {
2323                         rank = cp + 1;
2324                         break;
2325                     }
2326                 }
2327             }
2328             else
2329             {
2330                 xml_rank = yaz_xml_get_prop(n, "rank");
2331                 rank = xml_rank ? (const char *) xml_rank : ser_md->rank;
2332             }
2333
2334             wheretoput = &cluster->metadata[md_field_id];
2335
2336             if (ser_md->merge == Metadata_merge_first)
2337             {
2338                 if (!metadata0[md_field_id])
2339                 {
2340                     while (*wheretoput)
2341                         wheretoput = &(*wheretoput)->next;
2342                     *wheretoput = rec_md;
2343                 }
2344             }
2345             else if (ser_md->merge == Metadata_merge_unique)
2346             {
2347                 while (*wheretoput)
2348                 {
2349                     if (!strcmp((const char *) (*wheretoput)->data.text.norm,
2350                                 rec_md->data.text.norm))
2351                         break;
2352                     wheretoput = &(*wheretoput)->next;
2353                 }
2354                 if (!*wheretoput)
2355                     *wheretoput = rec_md;
2356             }
2357             else if (ser_md->merge == Metadata_merge_longest)
2358             {
2359                 if (!*wheretoput
2360                     || strlen(rec_md->data.text.norm)
2361                     > strlen((*wheretoput)->data.text.norm))
2362                 {
2363                     *wheretoput = rec_md;
2364                     if (ser_sk)
2365                     {
2366                         pp2_charset_token_t prt;
2367                         const char *sort_str = 0;
2368                         int skip_article =
2369                             ser_sk->type == Metadata_type_skiparticle;
2370
2371                         if (!cluster->sortkeys[sk_field_id])
2372                             cluster->sortkeys[sk_field_id] =
2373                                 nmem_malloc(se->nmem,
2374                                             sizeof(union data_types));
2375
2376                         prt =
2377                             pp2_charset_token_create(service->charsets, "sort");
2378
2379                         pp2_charset_token_first(prt, rec_md->data.text.disp,
2380                                                 skip_article);
2381
2382                         pp2_charset_token_next(prt);
2383
2384                         sort_str = pp2_get_sort(prt);
2385
2386                         cluster->sortkeys[sk_field_id]->text.disp =
2387                             rec_md->data.text.disp;
2388                         if (!sort_str)
2389                         {
2390                             sort_str = rec_md->data.text.disp;
2391                             session_log(se, YLOG_WARN,
2392                                     "Could not make sortkey. Bug #1858");
2393                         }
2394                         cluster->sortkeys[sk_field_id]->text.sort =
2395                             nmem_strdup(se->nmem, sort_str);
2396                         pp2_charset_token_destroy(prt);
2397                     }
2398                 }
2399             }
2400             else if (ser_md->merge == Metadata_merge_all)
2401             {
2402                 while (*wheretoput)
2403                     wheretoput = &(*wheretoput)->next;
2404                 *wheretoput = rec_md;
2405             }
2406             else if (ser_md->merge == Metadata_merge_range)
2407             {
2408                 if (!*wheretoput)
2409                 {
2410                     *wheretoput = rec_md;
2411                     if (ser_sk)
2412                         cluster->sortkeys[sk_field_id]
2413                             = &rec_md->data;
2414                 }
2415                 else
2416                 {
2417                     int this_min = rec_md->data.number.min;
2418                     int this_max = rec_md->data.number.max;
2419                     if (this_min < (*wheretoput)->data.number.min)
2420                         (*wheretoput)->data.number.min = this_min;
2421                     if (this_max > (*wheretoput)->data.number.max)
2422                         (*wheretoput)->data.number.max = this_max;
2423                 }
2424             }
2425
2426             // ranking of _all_ fields enabled ...
2427             if (rank)
2428             {
2429                 relevance_countwords(se->relevance, cluster,
2430                                      wrbuf_cstr(wrbuf_disp),
2431                                      rank, ser_md->name);
2432             }
2433             // construct facets ... unless the client already has reported them
2434             if (ser_md->termlist && !client_has_facet(cl, (char *) type))
2435             {
2436                 if (ser_md->type == Metadata_type_year)
2437                 {
2438                     char year[64];
2439                     sprintf(year, "%d", rec_md->data.number.max);
2440
2441                     add_facet(se, (char *) type, year, term_factor, cl);
2442                     if (rec_md->data.number.max != rec_md->data.number.min)
2443                     {
2444                         sprintf(year, "%d", rec_md->data.number.min);
2445                         add_facet(se, (char *) type, year, term_factor, cl);
2446                     }
2447                 }
2448                 else
2449                     add_facet(se, type, wrbuf_cstr(wrbuf_disp), term_factor, cl);
2450             }
2451         }
2452         else
2453         {
2454             if (se->number_of_warnings_unknown_elements == 0)
2455                 session_log(se, YLOG_WARN,
2456                         "Unexpected element in internal record: %s", n->name);
2457             se->number_of_warnings_unknown_elements++;
2458         }
2459     }
2460     nmem_destroy(ingest_nmem);
2461     xfree(metadata0);
2462     relevance_donerecord(se->relevance, cluster);
2463     se->total_records++;
2464
2465     return 0;
2466 }
2467
2468 void session_log(struct session *s, int level, const char *fmt, ...)
2469 {
2470     char buf[1024];
2471     va_list ap;
2472     va_start(ap, fmt);
2473
2474     yaz_vsnprintf(buf, sizeof(buf)-30, fmt, ap);
2475     yaz_log(level, "Session %u: %s", s ? s->session_id : 0, buf);
2476
2477     va_end(ap);
2478 }
2479
2480 /*
2481  * Local variables:
2482  * c-basic-offset: 4
2483  * c-file-style: "Stroustrup"
2484  * indent-tabs-mode: nil
2485  * End:
2486  * vim: shiftwidth=4 tabstop=8 expandtab
2487  */
2488