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 // Look up a facet term, and return matching id
 232 // If facet type not found, returns 0
 233 // If facet type found, but no matching term, returns ""
 234 const char *session_lookup_id_facet(struct session *s, struct client *cl,
 235                                     const char *type,
 236                                     const char *term)
 237 {
 238     char *retval = 0;
 239     struct facet_id *t = s->facet_id_list;
 240     for (; t; t = t->next)
 241     {
 242         if (!strcmp(client_get_id(cl), t->client_id) &&  !strcmp(t->type, type) )
 243         {
 244             retval = "";
 245             if ( !strcmp(t->term, term))
 246             {
 247                 return t->id;
 248             }
 249         }
 250     }
 251     return retval;
 252 }
 253
 254 void add_facet(struct session *s, const char *type, const char *value, int count, struct client *cl)
 255 {
 256     WRBUF facet_wrbuf = wrbuf_alloc();
 257     WRBUF display_wrbuf = wrbuf_alloc();
 258     const char *id = 0;
 259     size_t id_len = 0;
 260
 261     /* inspect pz:facetmap:split:name ?? */
 262     if (!strncmp(type, "split:", 6))
 263     {
 264         const char *cp = strchr(value, ':');
 265         if (cp)
 266         {
 267             id = value;
 268             id_len = cp - value;
 269             value = cp + 1;
 270         }
 271         type += 6;
 272     }
 273     session_normalize_facet(s, type, value, display_wrbuf, facet_wrbuf);
 274     yaz_log(YLOG_DEBUG,"add_facet t='%s' v='%s' id='%s' d='%s' f='%s'",
 275             type, value, id, wrbuf_cstr(display_wrbuf), wrbuf_cstr(facet_wrbuf) );
 276     if (wrbuf_len(facet_wrbuf))
 277     {
 278         struct named_termlist **tp = &s->termlists;
 279         for (; (*tp); tp = &(*tp)->next)
 280             if (!strcmp((*tp)->name, type))
 281                 break;
 282         if (!*tp)
 283         {
 284             *tp = nmem_malloc(s->nmem, sizeof(**tp));
 285             (*tp)->name = nmem_strdup(s->nmem, type);
 286             (*tp)->termlist = termlist_create(s->nmem);
 287             (*tp)->next = 0;
 288         }
 289         termlist_insert((*tp)->termlist, wrbuf_cstr(display_wrbuf),
 290                         wrbuf_cstr(facet_wrbuf), id, id_len, count);
 291         if (id)
 292             session_add_id_facet(s, cl, type, id, id_len,
 293                                  wrbuf_cstr(display_wrbuf));
 294     }
 295     wrbuf_destroy(facet_wrbuf);
 296     wrbuf_destroy(display_wrbuf);
 297 }
 298
 299 static xmlDoc *record_to_xml(struct session *se,
 300                              struct session_database *sdb, const char *rec)
 301 {
 302     struct database *db = sdb->database;
 303     xmlDoc *rdoc = 0;
 304
 305     rdoc = xmlParseMemory(rec, strlen(rec));
 306
 307     if (!rdoc)
 308     {
 309         session_log(se, YLOG_WARN, "Non-wellformed XML");
 310         return 0;
 311     }
 312
 313     if (global_parameters.dump_records)
 314     {
 315         session_log(se, YLOG_LOG, "Un-normalized record from %s", db->id);
 316         log_xml_doc(rdoc);
 317     }
 318
 319     return rdoc;
 320 }
 321
 322 #define MAX_XSLT_ARGS 16
 323
 324 // Add static values from session database settings if applicable
 325 static void insert_settings_parameters(struct session_database *sdb,
 326                                        struct conf_service *service,
 327                                        char **parms,
 328                                        NMEM nmem)
 329 {
 330     int i;
 331     int nparms = 0;
 332     int offset = 0;
 333
 334     for (i = 0; i < service->num_metadata; i++)
 335     {
 336         struct conf_metadata *md = &service->metadata[i];
 337         int setting;
 338
 339         if (md->setting == Metadata_setting_parameter &&
 340             (setting = settings_lookup_offset(service, md->name)) >= 0)
 341         {
 342             const char *val = session_setting_oneval(sdb, setting);
 343             if (val && nparms < MAX_XSLT_ARGS)
 344             {
 345                 char *buf;
 346                 int len = strlen(val);
 347                 buf = nmem_malloc(nmem, len + 3);
 348                 buf[0] = '\'';
 349                 strcpy(buf + 1, val);
 350                 buf[len+1] = '\'';
 351                 buf[len+2] = '\0';
 352                 parms[offset++] = md->name;
 353                 parms[offset++] = buf;
 354                 nparms++;
 355             }
 356         }
 357     }
 358     parms[offset] = 0;
 359 }
 360
 361 // Add static values from session database settings if applicable
 362 static void insert_settings_values(struct session_database *sdb, xmlDoc *doc,
 363                                    xmlNode *root,
 364                                    struct conf_service *service)
 365 {
 366     int i;
 367
 368     for (i = 0; i < service->num_metadata; i++)
 369     {
 370         struct conf_metadata *md = &service->metadata[i];
 371         int offset;
 372
 373         if (md->setting == Metadata_setting_postproc &&
 374             (offset = settings_lookup_offset(service, md->name)) >= 0)
 375         {
 376             const char *val = session_setting_oneval(sdb, offset);
 377             if (val)
 378             {
 379                 xmlNode *n = xmlNewTextChild(root, 0, (xmlChar *) "metadata",
 380                                              (xmlChar *) val);
 381                 xmlSetProp(n, (xmlChar *) "type", (xmlChar *) md->name);
 382             }
 383         }
 384     }
 385 }
 386
 387 static xmlDoc *normalize_record(struct session *se,
 388                                 struct session_database *sdb,
 389                                 struct conf_service *service,
 390                                 const char *rec, NMEM nmem)
 391 {
 392     xmlDoc *rdoc = record_to_xml(se, sdb, rec);
 393
 394     if (rdoc)
 395     {
 396         char *parms[MAX_XSLT_ARGS*2+1];
 397
 398         insert_settings_parameters(sdb, service, parms, nmem);
 399
 400         if (normalize_record_transform(sdb->map, &rdoc, (const char **)parms))
 401         {
 402             session_log(se, YLOG_WARN, "Normalize failed");
 403         }
 404     }
 405     return rdoc;
 406 }
 407
 408 void session_settings_dump(struct session *se,
 409                            struct session_database *db,
 410                            WRBUF w)
 411 {
 412     if (db->settings)
 413     {
 414         int i, num = db->num_settings;
 415         for (i = 0; i < num; i++)
 416         {
 417             struct setting *s = db->settings[i];
 418             for (;s ; s = s->next)
 419             {
 420                 wrbuf_puts(w, "<set name=\"");
 421                 wrbuf_xmlputs(w, s->name);
 422                 wrbuf_puts(w, "\" value=\"");
 423                 wrbuf_xmlputs(w, s->value);
 424                 wrbuf_puts(w, "\"/>");
 425             }
 426             if (db->settings[i])
 427                 wrbuf_puts(w, "\n");
 428         }
 429     }
 430 }
 431
 432 // Retrieve first defined value for 'name' for given database.
 433 // Will be extended to take into account user associated with session
 434 const char *session_setting_oneval(struct session_database *db, int offset)
 435 {
 436     if (offset >= db->num_settings || !db->settings[offset])
 437         return "";
 438     return db->settings[offset]->value;
 439 }
 440
 441 // Prepare XSLT stylesheets for record normalization
 442 // Structures are allocated on the session_wide nmem to avoid having
 443 // to recompute this for every search. This would lead
 444 // to leaking if a single session was to repeatedly change the PZ_XSLT
 445 // setting. However, this is not a realistic use scenario.
 446 static int prepare_map(struct session *se, struct session_database *sdb)
 447 {
 448     if (sdb->settings && !sdb->map)
 449     {
 450         const char *s;
 451
 452         if (sdb->settings[PZ_XSLT] &&
 453             (s = session_setting_oneval(sdb, PZ_XSLT)))
 454         {
 455             char auto_stylesheet[256];
 456
 457             if (!strcmp(s, "auto"))
 458             {
 459                 const char *request_syntax = session_setting_oneval(
 460                     sdb, PZ_REQUESTSYNTAX);
 461                 if (request_syntax)
 462                 {
 463                     char *cp;
 464                     yaz_snprintf(auto_stylesheet, sizeof(auto_stylesheet),
 465                                  "%s.xsl", request_syntax);
 466                     for (cp = auto_stylesheet; *cp; cp++)
 467                     {
 468                         /* deliberately only consider ASCII */
 469                         if (*cp > 32 && *cp < 127)
 470                             *cp = tolower(*cp);
 471                     }
 472                     s = auto_stylesheet;
 473                 }
 474                 else
 475                 {
 476                     session_log(se, YLOG_WARN,
 477                                 "No pz:requestsyntax for auto stylesheet");
 478                 }
 479             }
 480             sdb->map = normalize_cache_get(se->normalize_cache,
 481                                            se->service, s);
 482             if (!sdb->map)
 483                 return -1;
 484         }
 485     }
 486     return 0;
 487 }
 488
 489 // called if watch should be removed because http_channel is to be destroyed
 490 static void session_watch_cancel(void *data, struct http_channel *c,
 491                                  void *data2)
 492 {
 493     struct session_watchentry *ent = data;
 494
 495     ent->fun = 0;
 496     ent->data = 0;
 497     ent->obs = 0;
 498 }
 499
 500 // set watch. Returns 0=OK, -1 if watch is already set
 501 int session_set_watch(struct session *s, int what,
 502                       session_watchfun fun, void *data,
 503                       struct http_channel *chan)
 504 {
 505     int ret;
 506     session_enter(s, "session_set_watch");
 507     if (s->watchlist[what].fun)
 508         ret = -1;
 509     else
 510     {
 511
 512         s->watchlist[what].fun = fun;
 513         s->watchlist[what].data = data;
 514         s->watchlist[what].obs = http_add_observer(chan, &s->watchlist[what],
 515                                                    session_watch_cancel);
 516         ret = 0;
 517     }
 518     session_leave(s, "session_set_watch");
 519     return ret;
 520 }
 521
 522 void session_alert_watch(struct session *s, int what)
 523 {
 524     assert(s);
 525     session_enter(s, "session_alert_watch");
 526     if (s->watchlist[what].fun)
 527     {
 528         /* our watch is no longer associated with http_channel */
 529         void *data;
 530         session_watchfun fun;
 531
 532         http_remove_observer(s->watchlist[what].obs);
 533         fun  = s->watchlist[what].fun;
 534         data = s->watchlist[what].data;
 535
 536         /* reset watch before fun is invoked - in case fun wants to set
 537            it again */
 538         s->watchlist[what].fun = 0;
 539         s->watchlist[what].data = 0;
 540         s->watchlist[what].obs = 0;
 541
 542         session_leave(s, "session_alert_watch");
 543         session_log(s, YLOG_DEBUG,
 544                     "Alert Watch: %d calling function: %p", what, fun);
 545         fun(data);
 546     }
 547     else
 548         session_leave(s,"session_alert_watch");
 549 }
 550
 551 //callback for grep_databases
 552 static void select_targets_callback(struct session *se,
 553                                     struct session_database *db)
 554 {
 555     struct client *cl;
 556     struct client_list *l;
 557
 558     for (l = se->clients_cached; l; l = l->next)
 559         if (client_get_database(l->client) == db)
 560             break;
 561
 562     if (l)
 563         cl = l->client;
 564     else
 565     {
 566         cl = client_create(db->database->id);
 567         client_set_database(cl, db);
 568
 569         l = xmalloc(sizeof(*l));
 570         l->client = cl;
 571         l->next = se->clients_cached;
 572         se->clients_cached = l;
 573     }
 574     /* set session always. If may be 0 if client is not active */
 575     client_set_session(cl, se);
 576
 577     l = xmalloc(sizeof(*l));
 578     l->client = cl;
 579     l->next = se->clients_active;
 580     se->clients_active = l;
 581 }
 582
 583 static void session_reset_active_clients(struct session *se,
 584                                          struct client_list *new_list)
 585 {
 586     struct client_list *l;
 587
 588     session_enter(se, "session_reset_active_clients");
 589     l = se->clients_active;
 590     se->clients_active = new_list;
 591     session_leave(se, "session_reset_active_clients");
 592
 593     while (l)
 594     {
 595         struct client_list *l_next = l->next;
 596
 597         client_lock(l->client);
 598         client_set_session(l->client, 0); /* mark client inactive */
 599         client_unlock(l->client);
 600
 601         xfree(l);
 602         l = l_next;
 603     }
 604 }
 605
 606 static void session_remove_cached_clients(struct session *se)
 607 {
 608     struct client_list *l;
 609
 610     session_reset_active_clients(se, 0);
 611
 612     session_enter(se, "session_remove_cached_clients");
 613     l = se->clients_cached;
 614     se->clients_cached = 0;
 615     session_leave(se, "session_remove_cached_clients");
 616
 617     while (l)
 618     {
 619         struct client_list *l_next = l->next;
 620         client_lock(l->client);
 621         client_set_session(l->client, 0);
 622         client_set_database(l->client, 0);
 623         client_unlock(l->client);
 624         client_destroy(l->client);
 625         xfree(l);
 626         l = l_next;
 627     }
 628 }
 629
 630 // Associates a set of clients with a session;
 631 // Note: Session-databases represent databases with per-session
 632 // setting overrides
 633 static int select_targets(struct session *se, const char *filter)
 634 {
 635     return session_grep_databases(se, filter, select_targets_callback);
 636 }
 637
 638 int session_active_clients(struct session *s)
 639 {
 640     struct client_list *l;
 641     int res = 0;
 642
 643     for (l = s->clients_active; l; l = l->next)
 644         if (client_is_active(l->client))
 645             res++;
 646
 647     return res;
 648 }
 649
 650 int session_is_preferred_clients_ready(struct session *s)
 651 {
 652     struct client_list *l;
 653     int res = 0;
 654
 655     for (l = s->clients_active; l; l = l->next)
 656         if (client_is_active_preferred(l->client))
 657             res++;
 658     session_log(s, YLOG_DEBUG, "Has %d active preferred clients.", res);
 659     return res == 0;
 660 }
 661
 662 static void session_clear_set(struct session *se, struct reclist_sortparms *sp)
 663 {
 664     reclist_destroy(se->reclist);
 665     if (nmem_total(se->nmem))
 666         session_log(se, YLOG_DEBUG, "NMEN operation usage %zd",
 667                     nmem_total(se->nmem));
 668     nmem_reset(se->nmem);
 669     se->total_records = se->total_merged = 0;
 670     se->termlists = 0;
 671     relevance_clear(se->relevance);
 672
 673     /* reset list of sorted results and clear to relevance search */
 674     se->sorted_results = nmem_malloc(se->nmem, sizeof(*se->sorted_results));
 675     se->sorted_results->name = nmem_strdup(se->nmem, sp->name);
 676     se->sorted_results->increasing = sp->increasing;
 677     se->sorted_results->type = sp->type;
 678     se->sorted_results->next = 0;
 679
 680     session_log(se, YLOG_DEBUG, "clear_set session_sort: field=%s increasing=%d type=%d configured",
 681             sp->name, sp->increasing, sp->type);
 682
 683     se->reclist = reclist_create(se->nmem);
 684 }
 685
 686 void session_sort(struct session *se, struct reclist_sortparms *sp,
 687                   const char *mergekey, const char *rank)
 688 {
 689     struct client_list *l;
 690     const char *field = sp->name;
 691     int increasing = sp->increasing;
 692     int type  = sp->type;
 693     int clients_research = 0;
 694
 695     session_enter(se, "session_sort");
 696     session_log(se, YLOG_DEBUG, "session_sort field=%s increasing=%d type=%d",
 697                 field, increasing, type);
 698
 699     if (rank && (!se->rank || strcmp(se->rank, rank)))
 700     {
 701         /* new rank must research/reingest anyway */
 702         assert(rank);
 703         xfree(se->rank);
 704         se->rank = *rank ? xstrdup(rank) : 0;
 705         clients_research = 1;
 706         session_log(se, YLOG_DEBUG, "session_sort: new rank = %s",
 707                     rank);
 708     }
 709     if (mergekey && (!se->mergekey || strcmp(se->mergekey, mergekey)))
 710     {
 711         /* new mergekey must research/reingest anyway */
 712         assert(mergekey);
 713         xfree(se->mergekey);
 714         se->mergekey = *mergekey ? xstrdup(mergekey) : 0;
 715         clients_research = 1;
 716         session_log(se, YLOG_DEBUG, "session_sort: new mergekey = %s",
 717                     mergekey);
 718     }
 719     if (clients_research == 0)
 720     {
 721         struct reclist_sortparms *sr;
 722         for (sr = se->sorted_results; sr; sr = sr->next)
 723             if (!reclist_sortparms_cmp(sr, sp))
 724                 break;
 725         if (sr)
 726         {
 727             session_log(se, YLOG_DEBUG, "session_sort: field=%s increasing=%d type=%d already fetched",
 728                         field, increasing, type);
 729             session_leave(se, "session_sort");
 730             return;
 731         }
 732     }
 733     session_log(se, YLOG_DEBUG, "session_sort: field=%s increasing=%d type=%d must fetch",
 734                 field, increasing, type);
 735
 736     // We need to reset reclist on every sort that changes the records, not just for position
 737     // So if just one client requires new searching, we need to clear set.
 738     // Ask each of the client if sorting requires re-search due to native sort
 739     // If it does it will require us to
 740     for (l = se->clients_active; l; l = l->next)
 741     {
 742         struct client *cl = l->client;
 743         // Assume no re-search is required.
 744         client_parse_init(cl, 1);
 745         clients_research += client_parse_sort(cl, sp);
 746     }
 747     if (!clients_research || se->clients_starting)
 748     {
 749         // A new sorting based on same record set
 750         struct reclist_sortparms *sr = nmem_malloc(se->nmem, sizeof(*sr));
 751         sr->name = nmem_strdup(se->nmem, field);
 752         sr->increasing = increasing;
 753         sr->type = type;
 754         sr->next = se->sorted_results;
 755         se->sorted_results = sr;
 756         session_log(se, YLOG_DEBUG, "session_sort: no research/ingesting done");
 757         session_leave(se, "session_sort");
 758     }
 759     else
 760     {
 761         se->clients_starting = 1;
 762         session_log(se, YLOG_DEBUG,
 763                     "session_sort: reset results due to %d clients researching",
 764                     clients_research);
 765         session_clear_set(se, sp);
 766         session_log(se, YLOG_DEBUG, "Re- search/ingesting for clients due to change in sort order");
 767
 768         session_leave(se, "session_sort");
 769         for (l = se->clients_active; l; l = l->next)
 770         {
 771             struct client *cl = l->client;
 772             if (client_get_state(cl) == Client_Connecting ||
 773                 client_get_state(cl) == Client_Idle ||
 774                 client_get_state(cl) == Client_Working) {
 775                 client_start_search(cl);
 776             }
 777             else
 778             {
 779                 session_log(se, YLOG_DEBUG,
 780                             "session_sort: %s: No re-start/ingest in show. "
 781                             "Wrong client state: %d",
 782                             client_get_id(cl), client_get_state(cl));
 783             }
 784         }
 785         session_enter(se, "session_sort");
 786         se->clients_starting = 0;
 787         session_leave(se, "session_sort");
 788     }
 789 }
 790
 791 void session_stop(struct session *se)
 792 {
 793     struct client_list *l;
 794     session_enter(se, "session_stop1");
 795     if (se->clients_starting)
 796     {
 797         session_leave(se, "session_stop1");
 798         return;
 799     }
 800     se->clients_starting = 1;
 801     session_leave(se, "session_stop1");
 802
 803     session_alert_watch(se, SESSION_WATCH_SHOW);
 804     session_alert_watch(se, SESSION_WATCH_BYTARGET);
 805     session_alert_watch(se, SESSION_WATCH_TERMLIST);
 806     session_alert_watch(se, SESSION_WATCH_SHOW_PREF);
 807
 808     for (l = se->clients_active; l; l = l->next)
 809     {
 810         struct client *cl = l->client;
 811         client_stop(cl);
 812     }
 813     session_enter(se, "session_stop2");
 814     se->clients_starting = 0;
 815     session_leave(se, "session_stop2");
 816 }
 817
 818 enum pazpar2_error_code session_search(struct session *se,
 819                                        const char *query,
 820                                        const char *startrecs,
 821                                        const char *maxrecs,
 822                                        const char *filter,
 823                                        const char *limit,
 824                                        const char **addinfo,
 825                                        const char **addinfo2,
 826                                        struct reclist_sortparms *sp,
 827                                        const char *mergekey,
 828                                        const char *rank)
 829 {
 830     int live_channels = 0;
 831     int no_working = 0;
 832     int no_failed_query = 0;
 833     int no_failed_limit = 0;
 834     struct client_list *l;
 835
 836     session_log(se, YLOG_DEBUG, "Search");
 837
 838     *addinfo = 0;
 839
 840     session_enter(se, "session_search0");
 841     if (se->clients_starting)
 842     {
 843         session_leave(se, "session_search0");
 844         return PAZPAR2_NO_ERROR;
 845     }
 846     se->clients_starting = 1;
 847     session_leave(se, "session_search0");
 848
 849     if (se->settings_modified) {
 850         session_remove_cached_clients(se);
 851     }
 852     else
 853         session_reset_active_clients(se, 0);
 854
 855     session_enter(se, "session_search");
 856     se->settings_modified = 0;
 857
 858     if (mergekey)
 859     {
 860         xfree(se->mergekey);
 861         se->mergekey = *mergekey ? xstrdup(mergekey) : 0;
 862     }
 863     if (rank)
 864     {
 865         xfree(se->rank);
 866         se->rank = *rank ? xstrdup(rank) : 0;
 867     }
 868
 869     session_clear_set(se, sp);
 870     relevance_destroy(&se->relevance);
 871
 872     live_channels = select_targets(se, filter);
 873     if (!live_channels)
 874     {
 875         session_leave(se, "session_search");
 876         se->clients_starting = 0;
 877         return PAZPAR2_NO_TARGETS;
 878     }
 879
 880     facet_limits_destroy(se->facet_limits);
 881     se->facet_limits = facet_limits_create(limit);
 882     if (!se->facet_limits)
 883     {
 884         *addinfo = "limit";
 885         session_leave(se, "session_search");
 886         se->clients_starting = 0;
 887         return PAZPAR2_MALFORMED_PARAMETER_VALUE;
 888     }
 889
 890     session_leave(se, "session_search");
 891
 892     session_alert_watch(se, SESSION_WATCH_SHOW);
 893     session_alert_watch(se, SESSION_WATCH_BYTARGET);
 894     session_alert_watch(se, SESSION_WATCH_TERMLIST);
 895     session_alert_watch(se, SESSION_WATCH_SHOW_PREF);
 896
 897     for (l = se->clients_active; l; l = l->next)
 898     {
 899         int parse_ret;
 900         struct client *cl = l->client;
 901         client_parse_init(cl, 1);
 902         if (prepare_map(se, client_get_database(cl)) < 0)
 903             continue;
 904
 905         parse_ret = client_parse_query(cl, query, se->facet_limits, addinfo2);
 906         if (parse_ret == -1)
 907             no_failed_query++;
 908         else if (parse_ret == -2)
 909             no_failed_limit++;
 910         else if (parse_ret < 0)
 911             no_working++; /* other error, such as bad CCL map */
 912         else
 913         {
 914             client_parse_range(cl, startrecs, maxrecs);
 915             client_parse_sort(cl, sp);
 916             client_start_search(cl);
 917             no_working++;
 918         }
 919     }
 920     session_enter(se, "session_search2");
 921     se->clients_starting = 0;
 922     session_leave(se, "session_search2");
 923     if (no_working == 0)
 924     {
 925         if (no_failed_query > 0)
 926         {
 927             *addinfo = "query";
 928             return PAZPAR2_MALFORMED_PARAMETER_VALUE;
 929         }
 930         else if (no_failed_limit > 0)
 931         {
 932             *addinfo = "limit";
 933             return PAZPAR2_MALFORMED_PARAMETER_VALUE;
 934         }
 935         else
 936             return PAZPAR2_NO_TARGETS;
 937     }
 938     return PAZPAR2_NO_ERROR;
 939 }
 940
 941 // Creates a new session_database object for a database
 942 static void session_init_databases_fun(void *context, struct database *db)
 943 {
 944     struct session *se = (struct session *) context;
 945     struct session_database *new = nmem_malloc(se->session_nmem, sizeof(*new));
 946     int i;
 947
 948     new->database = db;
 949
 950     new->map = 0;
 951     assert(db->settings);
 952     new->settings = nmem_malloc(se->session_nmem,
 953                                 sizeof(struct settings *) * db->num_settings);
 954     new->num_settings = db->num_settings;
 955     for (i = 0; i < db->num_settings; i++)
 956     {
 957         struct setting *setting = db->settings[i];
 958         new->settings[i] = setting;
 959     }
 960     new->next = se->databases;
 961     se->databases = new;
 962 }
 963
 964 // Doesn't free memory associated with sdb -- nmem takes care of that
 965 static void session_database_destroy(struct session_database *sdb)
 966 {
 967     sdb->map = 0;
 968 }
 969
 970 // Initialize session_database list -- this represents this session's view
 971 // of the database list -- subject to modification by the settings ws command
 972 void session_init_databases(struct session *se)
 973 {
 974     se->databases = 0;
 975     predef_grep_databases(se, se->service, session_init_databases_fun);
 976 }
 977
 978 // Probably session_init_databases_fun should be refactored instead of
 979 // called here.
 980 static struct session_database *load_session_database(struct session *se,
 981                                                       const char *id)
 982 {
 983     struct database *db = new_database_inherit_settings(id, se->session_nmem, se->service->settings);
 984     session_init_databases_fun((void*) se, db);
 985
 986     // New sdb is head of se->databases list
 987     return se->databases;
 988 }
 989
 990 // Find an existing session database. If not found, load it
 991 static struct session_database *find_session_database(struct session *se,
 992                                                       const char *id)
 993 {
 994     struct session_database *sdb;
 995
 996     for (sdb = se->databases; sdb; sdb = sdb->next)
 997         if (!strcmp(sdb->database->id, id))
 998             return sdb;
 999     return load_session_database(se, id);
1000 }
1001
1002 // Apply a session override to a database
1003 void session_apply_setting(struct session *se, const char *dbname,
1004                            const char *name, const char *value)
1005 {
1006     session_enter(se, "session_apply_setting");
1007     {
1008         struct session_database *sdb = find_session_database(se, dbname);
1009         struct conf_service *service = se->service;
1010         struct setting *s;
1011         int offset = settings_create_offset(service, name);
1012
1013         expand_settings_array(&sdb->settings, &sdb->num_settings, offset,
1014                               se->session_nmem);
1015         // Force later recompute of settings-driven data structures
1016         // (happens when a search starts and client connections are prepared)
1017         if (offset == PZ_XSLT)
1018             sdb->map = 0;
1019         se->settings_modified = 1;
1020         for (s = sdb->settings[offset]; s; s = s->next)
1021             if (!strcmp(s->name, name) &&
1022                 dbname && s->target && !strcmp(dbname, s->target))
1023                 break;
1024         if (!s)
1025         {
1026             s = nmem_malloc(se->session_nmem, sizeof(*s));
1027             s->precedence = 0;
1028             s->target = nmem_strdup(se->session_nmem, dbname);
1029             s->name = nmem_strdup(se->session_nmem, name);
1030             s->next = sdb->settings[offset];
1031             sdb->settings[offset] = s;
1032         }
1033         s->value = nmem_strdup(se->session_nmem, value);
1034     }
1035     session_leave(se, "session_apply_setting");
1036 }
1037
1038 void session_destroy(struct session *se)
1039 {
1040     struct session_database *sdb;
1041     session_log(se, YLOG_LOG, "destroy");
1042     session_use(-1);
1043     session_remove_cached_clients(se);
1044
1045     for (sdb = se->databases; sdb; sdb = sdb->next)
1046         session_database_destroy(sdb);
1047     normalize_cache_destroy(se->normalize_cache);
1048     relevance_destroy(&se->relevance);
1049     reclist_destroy(se->reclist);
1050     xfree(se->mergekey);
1051     xfree(se->rank);
1052     if (nmem_total(se->nmem))
1053         session_log(se, YLOG_DEBUG, "NMEN operation usage %zd", nmem_total(se->nmem));
1054     if (nmem_total(se->session_nmem))
1055         session_log(se, YLOG_DEBUG, "NMEN session usage %zd", nmem_total(se->session_nmem));
1056     facet_limits_destroy(se->facet_limits);
1057     nmem_destroy(se->nmem);
1058     service_destroy(se->service);
1059     yaz_mutex_destroy(&se->session_mutex);
1060 }
1061
1062 size_t session_get_memory_status(struct session *session) {
1063     size_t session_nmem;
1064     if (session == 0)
1065         return 0;
1066     session_enter(session, "session_get_memory_status");
1067     session_nmem = nmem_total(session->nmem);
1068     session_leave(session, "session_get_memory_status");
1069     return session_nmem;
1070 }
1071
1072
1073 struct session *new_session(NMEM nmem, struct conf_service *service,
1074                             unsigned session_id)
1075 {
1076     int i;
1077     struct session *session = nmem_malloc(nmem, sizeof(*session));
1078
1079     char tmp_str[50];
1080
1081     sprintf(tmp_str, "session#%u", session_id);
1082
1083     session->session_id = session_id;
1084     session_log(session, YLOG_DEBUG, "New");
1085     session->service = service;
1086     session->relevance = 0;
1087     session->total_records = 0;
1088     session->number_of_warnings_unknown_elements = 0;
1089     session->number_of_warnings_unknown_metadata = 0;
1090     session->termlists = 0;
1091     session->reclist = reclist_create(nmem);
1092     session->clients_active = 0;
1093     session->clients_cached = 0;
1094     session->settings_modified = 0;
1095     session->session_nmem = nmem;
1096     session->facet_id_list = 0;
1097     session->nmem = nmem_create();
1098     session->databases = 0;
1099     session->sorted_results = 0;
1100     session->facet_limits = 0;
1101     session->mergekey = 0;
1102     session->rank = 0;
1103     session->clients_starting = 0;
1104
1105     for (i = 0; i <= SESSION_WATCH_MAX; i++)
1106     {
1107         session->watchlist[i].data = 0;
1108         session->watchlist[i].fun = 0;
1109     }
1110     session->normalize_cache = normalize_cache_create();
1111     session->session_mutex = 0;
1112     pazpar2_mutex_create(&session->session_mutex, tmp_str);
1113     session_log(session, YLOG_LOG, "create");
1114
1115     session_use(1);
1116     return session;
1117 }
1118
1119 static struct hitsbytarget *hitsbytarget_nb(struct session *se,
1120                                             int *count, NMEM nmem)
1121 {
1122     struct hitsbytarget *res = 0;
1123     struct client_list *l;
1124     size_t sz = 0;
1125
1126     for (l = se->clients_active; l; l = l->next)
1127         sz++;
1128
1129     res = nmem_malloc(nmem, sizeof(*res) * sz);
1130     *count = 0;
1131     for (l = se->clients_active; l; l = l->next)
1132     {
1133         struct client *cl = l->client;
1134         WRBUF w = wrbuf_alloc();
1135         const char *name = session_setting_oneval(client_get_database(cl),
1136                                                   PZ_NAME);
1137         res[*count].id = client_get_id(cl);
1138         res[*count].name = *name ? name : "Unknown";
1139         res[*count].hits = client_get_hits(cl);
1140         res[*count].approximation = client_get_approximation(cl);
1141         res[*count].records = client_get_num_records(cl,
1142                                                      &res[*count].filtered,
1143                                                      0, 0);
1144         res[*count].diagnostic =
1145             client_get_diagnostic(cl, &res[*count].message,
1146                                   &res[*count].addinfo);
1147         res[*count].state = client_get_state_str(cl);
1148         res[*count].connected  = client_get_connection(cl) ? 1 : 0;
1149         session_settings_dump(se, client_get_database(cl), w);
1150         res[*count].settings_xml = nmem_strdup(nmem, wrbuf_cstr(w));
1151         wrbuf_rewind(w);
1152         res[*count].suggestions_xml =
1153             nmem_strdup(nmem, client_get_suggestions_xml(cl, w));
1154
1155         res[*count].query_data =
1156             client_get_query(cl, &res[*count].query_type, nmem);
1157         wrbuf_destroy(w);
1158         (*count)++;
1159     }
1160     return res;
1161 }
1162
1163 struct hitsbytarget *get_hitsbytarget(struct session *se, int *count, NMEM nmem)
1164 {
1165     struct hitsbytarget *p;
1166     session_enter(se, "get_hitsbytarget");
1167     p = hitsbytarget_nb(se, count, nmem);
1168     session_leave(se, "get_hitsbytarget");
1169     return p;
1170 }
1171
1172 // Compares two hitsbytarget nodes by hitcount
1173 static int cmp_ht(const void *p1, const void *p2)
1174 {
1175     const struct hitsbytarget *h1 = p1;
1176     const struct hitsbytarget *h2 = p2;
1177     return h2->hits - h1->hits;
1178 }
1179
1180 // Compares two hitsbytarget nodes by hitcount
1181 static int cmp_ht_approx(const void *p1, const void *p2)
1182 {
1183     const struct hitsbytarget *h1 = p1;
1184     const struct hitsbytarget *h2 = p2;
1185     return h2->approximation - h1->approximation;
1186 }
1187
1188 static int targets_termlist_nb(WRBUF wrbuf, struct session *se, int num,
1189                                NMEM nmem, int version)
1190 {
1191     struct hitsbytarget *ht;
1192     int count, i;
1193
1194     ht = hitsbytarget_nb(se, &count, nmem);
1195     if (version >= 2)
1196         qsort(ht, count, sizeof(struct hitsbytarget), cmp_ht_approx);
1197     else
1198         qsort(ht, count, sizeof(struct hitsbytarget), cmp_ht);
1199     for (i = 0; i < count && i < num && ht[i].hits > 0; i++)
1200     {
1201
1202         // do only print terms which have display names
1203
1204         wrbuf_puts(wrbuf, "<term>\n");
1205
1206         wrbuf_puts(wrbuf, "<id>");
1207         wrbuf_xmlputs(wrbuf, ht[i].id);
1208         wrbuf_puts(wrbuf, "</id>\n");
1209
1210         wrbuf_puts(wrbuf, "<name>");
1211         if (!ht[i].name || !ht[i].name[0])
1212             wrbuf_xmlputs(wrbuf, "NO TARGET NAME");
1213         else
1214             wrbuf_xmlputs(wrbuf, ht[i].name);
1215         wrbuf_puts(wrbuf, "</name>\n");
1216
1217         wrbuf_printf(wrbuf, "<frequency>" ODR_INT_PRINTF "</frequency>\n",
1218                      ht[i].hits);
1219
1220         if (version >= 2) {
1221             // Should not print if we know it isn't a approximation.
1222             wrbuf_printf(wrbuf, "<approximation>" ODR_INT_PRINTF "</approximation>\n", ht[i].approximation);
1223             wrbuf_printf(wrbuf, "<records>%d</records>\n", ht[i].records - ht[i].filtered);
1224             wrbuf_printf(wrbuf, "<filtered>%d</filtered>\n", ht[i].filtered);
1225         }
1226
1227         wrbuf_puts(wrbuf, "<state>");
1228         wrbuf_xmlputs(wrbuf, ht[i].state);
1229         wrbuf_puts(wrbuf, "</state>\n");
1230
1231         wrbuf_printf(wrbuf, "<diagnostic>%d</diagnostic>\n",
1232                      ht[i].diagnostic);
1233         wrbuf_puts(wrbuf, "</term>\n");
1234     }
1235     return count;
1236 }
1237
1238 void perform_termlist(struct http_channel *c, struct session *se,
1239                       const char *name, int num, int version)
1240 {
1241     int j;
1242     NMEM nmem_tmp = nmem_create();
1243     char **names;
1244     int num_names = 0;
1245
1246     if (!name)
1247         name = "*";
1248
1249     nmem_strsplit(nmem_tmp, ",", name, &names, &num_names);
1250
1251     session_enter(se, "perform_termlist");
1252
1253     for (j = 0; j < num_names; j++)
1254     {
1255         const char *tname;
1256         int must_generate_empty = 1; /* bug 5350 */
1257
1258         struct named_termlist *t = se->termlists;
1259         for (; t; t = t->next)
1260         {
1261             tname = t->name;
1262             if (!strcmp(names[j], tname) || !strcmp(names[j], "*"))
1263             {
1264                 struct termlist_score **p = 0;
1265                 int len;
1266
1267                 wrbuf_puts(c->wrbuf, "<list name=\"");
1268                 wrbuf_xmlputs(c->wrbuf, tname);
1269                 wrbuf_puts(c->wrbuf, "\">\n");
1270                 must_generate_empty = 0;
1271
1272                 p = termlist_highscore(t->termlist, &len, nmem_tmp);
1273                 if (p)
1274                 {
1275                     int i;
1276                     for (i = 0; i < len && i < num; i++)
1277                     {
1278                         // prevent sending empty term elements
1279                         if (!p[i]->display_term || !p[i]->display_term[0])
1280                             continue;
1281
1282                         wrbuf_puts(c->wrbuf, "<term>");
1283                         wrbuf_puts(c->wrbuf, "<name>");
1284                         wrbuf_xmlputs(c->wrbuf, p[i]->display_term);
1285                         wrbuf_puts(c->wrbuf, "</name>");
1286                         wrbuf_printf(c->wrbuf,
1287                                      "<frequency>%d</frequency>",
1288                                      p[i]->frequency);
1289                         wrbuf_puts(c->wrbuf, "</term>\n");
1290                     }
1291                 }
1292                 wrbuf_puts(c->wrbuf, "</list>\n");
1293             }
1294         }
1295         tname = "xtargets";
1296         if (!strcmp(names[j], tname) || !strcmp(names[j], "*"))
1297         {
1298             wrbuf_puts(c->wrbuf, "<list name=\"");
1299             wrbuf_xmlputs(c->wrbuf, tname);
1300             wrbuf_puts(c->wrbuf, "\">\n");
1301
1302             targets_termlist_nb(c->wrbuf, se, num, c->nmem, version);
1303             wrbuf_puts(c->wrbuf, "</list>\n");
1304             must_generate_empty = 0;
1305         }
1306         if (must_generate_empty)
1307         {
1308             wrbuf_puts(c->wrbuf, "<list name=\"");
1309             wrbuf_xmlputs(c->wrbuf, names[j]);
1310             wrbuf_puts(c->wrbuf, "\"/>\n");
1311         }
1312     }
1313     session_leave(se, "perform_termlist");
1314     nmem_destroy(nmem_tmp);
1315 }
1316
1317 #ifdef MISSING_HEADERS
1318 void report_nmem_stats(void)
1319 {
1320     size_t in_use, is_free;
1321
1322     nmem_get_memory_in_use(&in_use);
1323     nmem_get_memory_free(&is_free);
1324
1325     yaz_log(YLOG_LOG, "nmem stat: use=%ld free=%ld",
1326             (long) in_use, (long) is_free);
1327 }
1328 #endif
1329
1330 struct record_cluster *show_single_start(struct session *se, const char *id,
1331                                          struct record_cluster **prev_r,
1332                                          struct record_cluster **next_r)
1333 {
1334     struct record_cluster *r = 0;
1335
1336     session_enter(se, "show_single_start");
1337     *prev_r = 0;
1338     *next_r = 0;
1339     reclist_limit(se->reclist, se, 1);
1340
1341     reclist_enter(se->reclist);
1342     while ((r = reclist_read_record(se->reclist)))
1343     {
1344         if (!strcmp(r->recid, id))
1345         {
1346             *next_r = reclist_read_record(se->reclist);
1347             break;
1348         }
1349         *prev_r = r;
1350     }
1351     reclist_leave(se->reclist);
1352     if (!r)
1353         session_leave(se, "show_single_start");
1354     return r;
1355 }
1356
1357 void show_single_stop(struct session *se, struct record_cluster *rec)
1358 {
1359     session_leave(se, "show_single_stop");
1360 }
1361
1362
1363 int session_fetch_more(struct session *se)
1364 {
1365     struct client_list *l;
1366     int ret = 0;
1367
1368     for (l = se->clients_active; l; l = l->next)
1369     {
1370         struct client *cl = l->client;
1371         if (client_get_state(cl) == Client_Idle)
1372         {
1373             if (client_fetch_more(cl))
1374             {
1375                 session_log(se, YLOG_LOG, "%s: more to fetch",
1376                             client_get_id(cl));
1377                 ret = 1;
1378             }
1379             else
1380             {
1381                 int filtered;
1382                 int ingest_failures;
1383                 int record_failures;
1384                 int num = client_get_num_records(
1385                     cl, &filtered, &ingest_failures, &record_failures);
1386
1387                 session_log(se, YLOG_LOG, "%s: hits=" ODR_INT_PRINTF
1388                             " fetched=%d filtered=%d",
1389                             client_get_id(cl),
1390                             client_get_hits(cl),
1391                             num, filtered);
1392                 if (ingest_failures || record_failures)
1393                 {
1394                     session_log(se, YLOG_WARN, "%s:"
1395                                 " ingest failures=%d record failures=%d",
1396                                 client_get_id(cl),
1397                                 ingest_failures, record_failures);
1398                 }
1399             }
1400         }
1401         else
1402         {
1403             session_log(se, YLOG_LOG, "%s: no fetch due to state=%s",
1404                         client_get_id(cl), client_get_state_str(cl));
1405         }
1406
1407     }
1408     return ret;
1409 }
1410
1411 struct record_cluster **show_range_start(struct session *se,
1412                                          struct reclist_sortparms *sp,
1413                                          int start, int *num, int *total,
1414                                          Odr_int *sumhits, Odr_int *approx_hits,
1415                                          void (*show_records_ready)(void *data),
1416                                          struct http_channel *chan)
1417 {
1418     struct record_cluster **recs = 0;
1419     struct reclist_sortparms *spp;
1420     struct client_list *l;
1421     int i;
1422 #if USE_TIMING
1423     yaz_timing_t t = yaz_timing_create();
1424 #endif
1425     session_enter(se, "show_range_start");
1426     *sumhits = 0;
1427     *approx_hits = 0;
1428     *total = 0;
1429     reclist_limit(se->reclist, se, 0);
1430     if (se->relevance)
1431     {
1432         for (spp = sp; spp; spp = spp->next)
1433             if (spp->type == Metadata_type_relevance)
1434             {
1435                 relevance_prepare_read(se->relevance, se->reclist);
1436                 break;
1437             }
1438         for (l = se->clients_active; l; l = l->next) {
1439             *sumhits += client_get_hits(l->client);
1440             *approx_hits += client_get_approximation(l->client);
1441         }
1442     }
1443     reclist_sort(se->reclist, sp);
1444
1445     reclist_enter(se->reclist);
1446     *total = reclist_get_num_records(se->reclist);
1447
1448     for (l = se->clients_active; l; l = l->next)
1449         client_update_show_stat(l->client, 0);
1450
1451     for (i = 0; i < start; i++)
1452     {
1453         struct record_cluster *r = reclist_read_record(se->reclist);
1454         if (!r)
1455         {
1456             *num = 0;
1457             break;
1458         }
1459         else
1460         {
1461             struct record *rec = r->records;
1462             for (;rec; rec = rec->next)
1463                 client_update_show_stat(rec->client, 1);
1464         }
1465     }
1466     recs = nmem_malloc(se->nmem, (*num > 0 ? *num : 1) * sizeof(*recs));
1467     for (i = 0; i < *num; i++)
1468     {
1469         struct record_cluster *r = reclist_read_record(se->reclist);
1470         if (!r)
1471         {
1472             *num = i;
1473             break;
1474         }
1475         else
1476         {
1477             struct record *rec = r->records;
1478             for (;rec; rec = rec->next)
1479                 client_update_show_stat(rec->client, 1);
1480             recs[i] = r;
1481         }
1482     }
1483     reclist_leave(se->reclist);
1484 #if USE_TIMING
1485     yaz_timing_stop(t);
1486     session_log(se, YLOG_LOG, "show %6.5f %3.2f %3.2f",
1487             yaz_timing_get_real(t), yaz_timing_get_user(t),
1488             yaz_timing_get_sys(t));
1489     yaz_timing_destroy(&t);
1490 #endif
1491
1492     if (!session_fetch_more(se))
1493         session_log(se, YLOG_LOG, "can not fetch more");
1494     else
1495     {
1496         show_range_stop(se, recs);
1497         session_log(se, YLOG_LOG, "fetching more in progress");
1498         if (session_set_watch(se, SESSION_WATCH_SHOW,
1499                               show_records_ready, chan, chan))
1500         {
1501             session_log(se, YLOG_WARN, "Ignoring show block");
1502             session_enter(se, "show_range_start");
1503         }
1504         else
1505         {
1506             session_log(se, YLOG_LOG, "session watch OK");
1507             return 0;
1508         }
1509     }
1510     return recs;
1511 }
1512
1513 void show_range_stop(struct session *se, struct record_cluster **recs)
1514 {
1515     session_leave(se, "show_range_stop");
1516 }
1517
1518 void statistics(struct session *se, struct statistics *stat)
1519 {
1520     struct client_list *l;
1521     int count = 0;
1522
1523     memset(stat, 0, sizeof(*stat));
1524     stat->num_hits = 0;
1525     for (l = se->clients_active; l; l = l->next)
1526     {
1527         struct client *cl = l->client;
1528         if (!client_get_connection(cl))
1529             stat->num_no_connection++;
1530         stat->num_hits += client_get_hits(cl);
1531         switch (client_get_state(cl))
1532         {
1533         case Client_Connecting: stat->num_connecting++; break;
1534         case Client_Working: stat->num_working++; break;
1535         case Client_Idle: stat->num_idle++; break;
1536         case Client_Failed: stat->num_failed++; break;
1537         case Client_Error: stat->num_error++; break;
1538         default: break;
1539         }
1540         count++;
1541     }
1542     stat->num_records = se->total_records;
1543
1544     stat->num_clients = count;
1545 }
1546
1547 static struct record_metadata *record_metadata_init(
1548     NMEM nmem, const char *value, const char *norm,
1549     enum conf_metadata_type type,
1550     struct _xmlAttr *attr)
1551 {
1552     struct record_metadata *rec_md = record_metadata_create(nmem);
1553     struct record_metadata_attr **attrp = &rec_md->attributes;
1554
1555     for (; attr; attr = attr->next)
1556     {
1557         if (attr->children && attr->children->content)
1558         {
1559             if (strcmp((const char *) attr->name, "type")
1560                 && strcmp((const char *) attr->name, "empty"))
1561             {  /* skip the "type" + "empty" attribute..
1562                   The "Type" is already part of the element in output
1563                   (md-%s) and so repeating it here is redundant */
1564                 *attrp = nmem_malloc(nmem, sizeof(**attrp));
1565                 (*attrp)->name =
1566                     nmem_strdup(nmem, (const char *) attr->name);
1567                 (*attrp)->value =
1568                     nmem_strdup(nmem, (const char *) attr->children->content);
1569                 attrp = &(*attrp)->next;
1570             }
1571         }
1572     }
1573     *attrp = 0;
1574
1575     switch (type)
1576     {
1577     case Metadata_type_generic:
1578     case Metadata_type_skiparticle:
1579         if (norm)
1580         {
1581             rec_md->data.text.disp = nmem_strdup(nmem, value);
1582             rec_md->data.text.norm = nmem_strdup(nmem, norm);
1583         }
1584         else
1585         {
1586             if (strstr(value, "://")) /* looks like a URL */
1587                 rec_md->data.text.disp = nmem_strdup(nmem, value);
1588             else
1589                 rec_md->data.text.disp =
1590                     normalize7bit_generic(nmem_strdup(nmem, value), " ,/.:([");
1591             rec_md->data.text.norm = rec_md->data.text.disp;
1592         }
1593         rec_md->data.text.sort = 0;
1594         rec_md->data.text.snippet = 0;
1595         break;
1596     case Metadata_type_year:
1597     case Metadata_type_date:
1598     {
1599         int first, last;
1600         int longdate = 0;
1601
1602         if (type == Metadata_type_date)
1603             longdate = 1;
1604         if (extract7bit_dates((char *) value, &first, &last, longdate) < 0)
1605             return 0;
1606
1607         rec_md->data.number.min = first;
1608         rec_md->data.number.max = last;
1609     }
1610     break;
1611     case Metadata_type_float:
1612         rec_md->data.fnumber = atof(value);
1613         break;
1614     case Metadata_type_relevance:
1615     case Metadata_type_position:
1616     case Metadata_type_retrieval:
1617         return 0;
1618     }
1619     return rec_md;
1620 }
1621
1622 static void mergekey_norm_wr(pp2_charset_fact_t charsets,
1623                              WRBUF norm_wr, const char *value)
1624 {
1625     const char *norm_str;
1626     pp2_charset_token_t prt =
1627         pp2_charset_token_create(charsets, "mergekey");
1628
1629     pp2_charset_token_first(prt, value, 0);
1630     while ((norm_str = pp2_charset_token_next(prt)))
1631     {
1632         if (*norm_str)
1633         {
1634             if (wrbuf_len(norm_wr))
1635                 wrbuf_puts(norm_wr, " ");
1636             wrbuf_puts(norm_wr, norm_str);
1637         }
1638     }
1639     pp2_charset_token_destroy(prt);
1640 }
1641
1642 static int get_mergekey_from_doc(xmlDoc *doc, xmlNode *root, const char *name,
1643                                  struct conf_service *service, WRBUF norm_wr)
1644 {
1645     xmlNode *n;
1646     int no_found = 0;
1647     for (n = root->children; n; n = n->next)
1648     {
1649         if (n->type != XML_ELEMENT_NODE)
1650             continue;
1651         if (!strcmp((const char *) n->name, "metadata"))
1652         {
1653             const char *type = yaz_xml_get_prop(n, "type");
1654             if (type == NULL) {
1655                 yaz_log(YLOG_FATAL, "Missing type attribute on metadata element. Skipping!");
1656             }
1657             else if (!strcmp(name, (const char *) type))
1658             {
1659                 xmlChar *value = xmlNodeListGetString(doc, n->children, 1);
1660                 if (value && *value)
1661                 {
1662                     if (wrbuf_len(norm_wr) > 0)
1663                         wrbuf_puts(norm_wr, " ");
1664                     wrbuf_puts(norm_wr, name);
1665                     mergekey_norm_wr(service->charsets, norm_wr,
1666                                      (const char *) value);
1667                     no_found++;
1668                 }
1669                 if (value)
1670                     xmlFree(value);
1671             }
1672         }
1673     }
1674     return no_found;
1675 }
1676
1677 static const char *get_mergekey(xmlDoc *doc, xmlNode *root,
1678                                 struct client *cl, int record_no,
1679                                 struct conf_service *service, NMEM nmem,
1680                                 const char *session_mergekey)
1681 {
1682     char *mergekey_norm = 0;
1683     WRBUF norm_wr = wrbuf_alloc();
1684     const char *mergekey;
1685
1686     if (session_mergekey)
1687     {
1688         int i, num = 0;
1689         char **values = 0;
1690         nmem_strsplit_escape2(nmem, ",", session_mergekey, &values,
1691                               &num, 1, '\\', 1);
1692
1693         for (i = 0; i < num; i++)
1694             get_mergekey_from_doc(doc, root, values[i], service, norm_wr);
1695     }
1696     else if ((mergekey = yaz_xml_get_prop(root, "mergekey")))
1697     {
1698         mergekey_norm_wr(service->charsets, norm_wr, mergekey);
1699     }
1700     else
1701     {
1702         /* no mergekey defined in XSL. Look for mergekey metadata instead */
1703         int field_id;
1704         for (field_id = 0; field_id < service->num_metadata; field_id++)
1705         {
1706             struct conf_metadata *ser_md = &service->metadata[field_id];
1707             if (ser_md->mergekey != Metadata_mergekey_no)
1708             {
1709                 int r = get_mergekey_from_doc(doc, root, ser_md->name,
1710                                               service, norm_wr);
1711                 if (r == 0 && ser_md->mergekey == Metadata_mergekey_required)
1712                 {
1713                     /* no mergekey on this one and it is required..
1714                        Generate unique key instead */
1715                     wrbuf_rewind(norm_wr);
1716                     break;
1717                 }
1718             }
1719         }
1720     }
1721
1722     /* generate unique key if none is not generated already or is empty */
1723     if (wrbuf_len(norm_wr) == 0)
1724     {
1725         wrbuf_printf(norm_wr, "position: %s-%d",
1726                      client_get_id(cl), record_no);
1727     }
1728     else
1729     {
1730         const char *lead = "content: ";
1731         wrbuf_insert(norm_wr, 0, lead, strlen(lead));
1732     }
1733     if (wrbuf_len(norm_wr) > 0)
1734         mergekey_norm = nmem_strdup(nmem, wrbuf_cstr(norm_wr));
1735     wrbuf_destroy(norm_wr);
1736     return mergekey_norm;
1737 }
1738
1739 /** \brief see if metadata for pz:recordfilter exists
1740     \param root xml root element of normalized record
1741     \param sdb session database for client
1742     \retval 0 if there is no metadata for pz:recordfilter
1743     \retval 1 if there is metadata for pz:recordfilter
1744
1745     If there is no pz:recordfilter defined, this function returns 1
1746     as well.
1747 */
1748
1749 static int check_record_filter(xmlNode *root, struct session_database *sdb)
1750 {
1751     int match = 0;
1752     xmlNode *n;
1753     const char *s;
1754     s = session_setting_oneval(sdb, PZ_RECORDFILTER);
1755
1756     if (!s || !*s)
1757         return 1;
1758
1759     for (n = root->children; n; n = n->next)
1760     {
1761         if (n->type != XML_ELEMENT_NODE)
1762             continue;
1763         if (!strcmp((const char *) n->name, "metadata"))
1764         {
1765             const char *type = yaz_xml_get_prop(n, "type");
1766             if (type)
1767             {
1768                 size_t len;
1769                 int substring;
1770                 const char *eq;
1771
1772                 if ((eq = strchr(s, '=')))
1773                     substring = 0;
1774                 else if ((eq = strchr(s, '~')))
1775                     substring = 1;
1776                 if (eq)
1777                     len = eq - s;
1778                 else
1779                     len = strlen(s);
1780                 if (len == strlen((const char *)type) &&
1781                     !memcmp((const char *) type, s, len))
1782                 {
1783                     xmlChar *value = xmlNodeGetContent(n);
1784                     if (value && *value)
1785                     {
1786                         if (!eq ||
1787                             (substring && strstr((const char *) value, eq+1)) ||
1788                             (!substring && !strcmp((const char *) value, eq + 1)))
1789                             match = 1;
1790                     }
1791                     xmlFree(value);
1792                 }
1793             }
1794         }
1795     }
1796     return match;
1797 }
1798
1799 static int ingest_to_cluster(struct client *cl,
1800                              WRBUF wrbuf_disp,
1801                              WRBUF wrbuf_norm,
1802                              xmlDoc *xdoc,
1803                              xmlNode *root,
1804                              int record_no,
1805                              struct record_metadata_attr *mergekey);
1806
1807 static int ingest_sub_record(struct client *cl, xmlDoc *xdoc, xmlNode *root,
1808                              int record_no, NMEM nmem,
1809                              struct session_database *sdb,
1810                              struct record_metadata_attr *mergekeys)
1811 {
1812     int ret = 0;
1813     struct session *se = client_get_session(cl);
1814     WRBUF wrbuf_disp, wrbuf_norm;
1815
1816     if (!check_record_filter(root, sdb))
1817     {
1818         session_log(se, YLOG_LOG,
1819                     "Filtered out record no %d from %s",
1820                     record_no, sdb->database->id);
1821         return 0;
1822     }
1823     wrbuf_disp = wrbuf_alloc();
1824     wrbuf_norm = wrbuf_alloc();
1825     session_enter(se, "ingest_sub_record");
1826     if (client_get_session(cl) == se && se->relevance)
1827         ret = ingest_to_cluster(cl, wrbuf_disp, wrbuf_norm,
1828                                 xdoc, root, record_no, mergekeys);
1829     session_leave(se, "ingest_sub_record");
1830     wrbuf_destroy(wrbuf_norm);
1831     wrbuf_destroy(wrbuf_disp);
1832     return ret;
1833 }
1834
1835 /** \brief ingest XML record
1836     \param cl client holds the result set for record
1837     \param rec record buffer (0 terminated)
1838     \param record_no record position (1, 2, ..)
1839     \param nmem working NMEM
1840     \retval 0 OK
1841     \retval -1 failure
1842     \retval -2 Filtered
1843 */
1844 int ingest_record(struct client *cl, const char *rec,
1845                   int record_no, NMEM nmem)
1846 {
1847     struct session *se = client_get_session(cl);
1848     struct session_database *sdb = client_get_database(cl);
1849     struct conf_service *service = se->service;
1850     xmlDoc *xdoc = normalize_record(se, sdb, service, rec, nmem);
1851     int r = ingest_xml_record(cl, xdoc, record_no, nmem, 0);
1852     client_store_xdoc(cl, record_no, xdoc);
1853     return r;
1854 }
1855
1856 int ingest_xml_record(struct client *cl, xmlDoc *xdoc,
1857                       int record_no, NMEM nmem, int cached_copy)
1858 {
1859     struct session *se = client_get_session(cl);
1860     struct session_database *sdb = client_get_database(cl);
1861     struct conf_service *service = se->service;
1862     xmlNode *root;
1863     int r = 0;
1864     if (!xdoc)
1865         return -1;
1866
1867     if (global_parameters.dump_records)
1868     {
1869         session_log(se, YLOG_LOG, "Normalized record from %s",
1870                     sdb->database->id);
1871         log_xml_doc(xdoc);
1872     }
1873
1874     root = xmlDocGetRootElement(xdoc);
1875
1876     if (!strcmp((const char *) root->name, "cluster"))
1877     {
1878         int no_merge_keys = 0;
1879         int no_merge_dups = 0;
1880         xmlNode *sroot;
1881         struct record_metadata_attr *mk = 0;
1882
1883         for (sroot = root->children; sroot; sroot = sroot->next)
1884             if (sroot->type == XML_ELEMENT_NODE &&
1885                 !strcmp((const char *) sroot->name, "record"))
1886             {
1887                 struct record_metadata_attr **mkp;
1888                 const char *mergekey_norm =
1889                     get_mergekey(xdoc, sroot, cl, record_no, service, nmem,
1890                                  se->mergekey);
1891                 if (!mergekey_norm)
1892                 {
1893                     r = -1;
1894                     break;
1895                 }
1896                 for (mkp = &mk; *mkp; mkp = &(*mkp)->next)
1897                     if (!strcmp((*mkp)->value, mergekey_norm))
1898                         break;
1899                 if (!*mkp)
1900                 {
1901                     *mkp = (struct record_metadata_attr*)
1902                         nmem_malloc(nmem, sizeof(**mkp));
1903                     (*mkp)->name = 0;
1904                     (*mkp)->value = nmem_strdup(nmem, mergekey_norm);
1905                     (*mkp)->next = 0;
1906                     no_merge_keys++;
1907                 }
1908                 else
1909                     no_merge_dups++;
1910             }
1911         if (no_merge_keys > 1 || no_merge_dups > 0)
1912         {
1913             yaz_log(YLOG_LOG, "Got %d mergekeys, %d dups for position %d",
1914                     no_merge_keys, no_merge_dups, record_no);
1915         }
1916         for (sroot = root->children; !r && sroot; sroot = sroot->next)
1917             if (sroot->type == XML_ELEMENT_NODE &&
1918                 !strcmp((const char *) sroot->name, "record"))
1919             {
1920                 if (!cached_copy)
1921                     insert_settings_values(sdb, xdoc, root, service);
1922                 r = ingest_sub_record(cl, xdoc, sroot, record_no, nmem, sdb,
1923                                       mk);
1924             }
1925     }
1926     else if (!strcmp((const char *) root->name, "record"))
1927     {
1928         const char *mergekey_norm =
1929             get_mergekey(xdoc, root, cl, record_no, service, nmem,
1930                          se->mergekey);
1931         if (mergekey_norm)
1932         {
1933             struct record_metadata_attr *mk = (struct record_metadata_attr*)
1934                 nmem_malloc(nmem, sizeof(*mk));
1935             mk->name = 0;
1936             mk->value = nmem_strdup(nmem, mergekey_norm);
1937             mk->next = 0;
1938
1939             if (!cached_copy)
1940                 insert_settings_values(sdb, xdoc, root, service);
1941             r = ingest_sub_record(cl, xdoc, root, record_no, nmem, sdb, mk);
1942         }
1943     }
1944     else
1945     {
1946         session_log(se, YLOG_WARN, "Bad pz root element: %s",
1947                     (const char *) root->name);
1948         r = -1;
1949     }
1950     return r;
1951 }
1952
1953
1954 //    struct conf_metadata *ser_md = &service->metadata[md_field_id];
1955 //    struct record_metadata *rec_md = record->metadata[md_field_id];
1956 static int match_metadata_local(struct conf_service *service,
1957                                 struct conf_metadata *ser_md,
1958                                 struct record_metadata *rec_md0,
1959                                 char **values, int num_v)
1960 {
1961     int i;
1962     struct record_metadata *rec_md = rec_md0;
1963     WRBUF val_wr = 0;
1964     WRBUF text_wr = wrbuf_alloc();
1965     for (i = 0; i < num_v; )
1966     {
1967         if (rec_md)
1968         {
1969             if (ser_md->type == Metadata_type_year
1970                 || ser_md->type == Metadata_type_date)
1971             {
1972                 int y = atoi(values[i]);
1973                 if (y >= rec_md->data.number.min
1974                     && y <= rec_md->data.number.max)
1975                     break;
1976             }
1977             else
1978             {
1979                 if (!val_wr)
1980                 {
1981                     val_wr = wrbuf_alloc();
1982                     mergekey_norm_wr(service->charsets, val_wr, values[i]);
1983                 }
1984                 wrbuf_rewind(text_wr);
1985                 mergekey_norm_wr(service->charsets, text_wr,
1986                                  rec_md->data.text.disp);
1987                 if (!strcmp(wrbuf_cstr(val_wr), wrbuf_cstr(text_wr)))
1988                     break;
1989             }
1990             rec_md = rec_md->next;
1991         }
1992         else
1993         {
1994             rec_md = rec_md0;
1995             wrbuf_destroy(val_wr);
1996             val_wr = 0;
1997             i++;
1998         }
1999     }
2000     wrbuf_destroy(val_wr);
2001     wrbuf_destroy(text_wr);
2002     return i < num_v ? 1 : 0;
2003 }
2004
2005 int session_check_cluster_limit(struct session *se, struct record_cluster *rec)
2006 {
2007     int i;
2008     struct conf_service *service = se->service;
2009     int ret = 1;
2010     const char *name;
2011     const char *value;
2012     NMEM nmem_tmp = nmem_create();
2013
2014     for (i = 0; (name = facet_limits_get(se->facet_limits, i, &value)); i++)
2015     {
2016         int j;
2017         for (j = 0; j < service->num_metadata; j++)
2018         {
2019             struct conf_metadata *md = service->metadata + j;
2020             if (!strcmp(md->name, name) && md->limitcluster)
2021             {
2022                 char **values = 0;
2023                 int num = 0;
2024                 int md_field_id =
2025                     conf_service_metadata_field_id(service,
2026                                                    md->limitcluster);
2027
2028                 if (md_field_id < 0)
2029                 {
2030                     ret = 0;
2031                     break;
2032                 }
2033
2034                 nmem_strsplit_escape2(nmem_tmp, "|", value, &values,
2035                                       &num, 1, '\\', 1);
2036
2037                 if (!match_metadata_local(service,
2038                                           &service->metadata[md_field_id],
2039                                           rec->metadata[md_field_id],
2040                                           values, num))
2041                 {
2042                     ret = 0;
2043                     break;
2044                 }
2045             }
2046         }
2047     }
2048     nmem_destroy(nmem_tmp);
2049     return ret;
2050 }
2051
2052 // Skip record on non-zero
2053 static int check_limit_local(struct client *cl,
2054                              struct record *record,
2055                              int record_no)
2056 {
2057     int skip_record = 0;
2058     struct session *se = client_get_session(cl);
2059     struct conf_service *service = se->service;
2060     NMEM nmem_tmp = nmem_create();
2061     struct session_database *sdb = client_get_database(cl);
2062     int l = 0;
2063     while (!skip_record)
2064     {
2065         int md_field_id;
2066         char **values = 0;
2067         int num_v = 0;
2068         const char *name =
2069             client_get_facet_limit_local(cl, sdb, &l, nmem_tmp,
2070                                          &num_v, &values);
2071         if (!name)
2072             break;
2073
2074         if (!strcmp(name, "*"))
2075         {
2076             for (md_field_id = 0; md_field_id < service->num_metadata;
2077                  md_field_id++)
2078             {
2079                 if (match_metadata_local(
2080                         service,
2081                         &service->metadata[md_field_id],
2082                         record->metadata[md_field_id],
2083                         values, num_v))
2084                     break;
2085             }
2086             if (md_field_id == service->num_metadata)
2087                 skip_record = 1;
2088         }
2089         else
2090         {
2091             md_field_id = conf_service_metadata_field_id(service, name);
2092             if (md_field_id < 0)
2093             {
2094                 skip_record = 1;
2095                 break;
2096             }
2097             if (!match_metadata_local(
2098                     service,
2099                     &service->metadata[md_field_id],
2100                     record->metadata[md_field_id],
2101                     values, num_v))
2102             {
2103                 skip_record = 1;
2104             }
2105         }
2106     }
2107     nmem_destroy(nmem_tmp);
2108     return skip_record;
2109 }
2110
2111 static int ingest_to_cluster(struct client *cl,
2112                              WRBUF wrbuf_disp,
2113                              WRBUF wrbuf_norm,
2114                              xmlDoc *xdoc,
2115                              xmlNode *root,
2116                              int record_no,
2117                              struct record_metadata_attr *merge_keys)
2118 {
2119     xmlNode *n;
2120     struct session *se = client_get_session(cl);
2121     struct conf_service *service = se->service;
2122     int term_factor = 1;
2123     struct record_cluster *cluster;
2124     struct record_metadata **metadata0;
2125     struct session_database *sdb = client_get_database(cl);
2126     NMEM ingest_nmem = 0;
2127     char **rank_values = 0;
2128     int rank_num = 0;
2129     struct record *record = record_create(se->nmem,
2130                                           service->num_metadata,
2131                                           service->num_sortkeys, cl,
2132                                           record_no);
2133
2134     for (n = root->children; n; n = n->next)
2135     {
2136         if (n->type != XML_ELEMENT_NODE)
2137             continue;
2138         if (!strcmp((const char *) n->name, "metadata"))
2139         {
2140             struct conf_metadata *ser_md = 0;
2141             struct record_metadata **wheretoput = 0;
2142             struct record_metadata *rec_md = 0;
2143             int md_field_id = -1;
2144             xmlChar *value0;
2145             const char *type = yaz_xml_get_prop(n, "type");
2146
2147             if (!type)
2148                 continue;
2149
2150             md_field_id
2151                 = conf_service_metadata_field_id(service, (const char *) type);
2152             if (md_field_id < 0)
2153             {
2154                 if (se->number_of_warnings_unknown_metadata == 0)
2155                 {
2156                     session_log(se, YLOG_WARN,
2157                             "Ignoring unknown metadata element: %s", type);
2158                 }
2159                 se->number_of_warnings_unknown_metadata++;
2160                 continue;
2161             }
2162
2163             wrbuf_rewind(wrbuf_disp);
2164             value0 = xmlNodeListGetString(xdoc, n->children, 1);
2165             if (!value0 || !*value0)
2166             {
2167                 const char *empty = yaz_xml_get_prop(n, "empty");
2168                 if (!empty)
2169                     continue;
2170                 wrbuf_puts(wrbuf_disp, (const char *) empty);
2171             }
2172             else
2173             {
2174                 wrbuf_puts(wrbuf_disp, (const char *) value0);
2175             }
2176             if (value0)
2177                 xmlFree(value0);
2178             ser_md = &service->metadata[md_field_id];
2179
2180             // non-merged metadata
2181             rec_md = record_metadata_init(se->nmem, wrbuf_cstr(wrbuf_disp), 0,
2182                                           ser_md->type, n->properties);
2183             if (!rec_md)
2184             {
2185                 session_log(se, YLOG_WARN, "bad metadata data '%s' "
2186                             "for element '%s'", wrbuf_cstr(wrbuf_disp), type);
2187                 continue;
2188             }
2189
2190             if (ser_md->type == Metadata_type_generic)
2191             {
2192                 WRBUF w = wrbuf_alloc();
2193                 if (relevance_snippet(se->relevance,
2194                                       wrbuf_cstr(wrbuf_disp), ser_md->name, w))
2195                     rec_md->data.text.snippet = nmem_strdup(se->nmem,
2196                                                             wrbuf_cstr(w));
2197                 wrbuf_destroy(w);
2198             }
2199
2200
2201             wheretoput = &record->metadata[md_field_id];
2202             while (*wheretoput)
2203                 wheretoput = &(*wheretoput)->next;
2204             *wheretoput = rec_md;
2205         }
2206     }
2207
2208     if (check_limit_local(cl, record, record_no))
2209     {
2210         return -2;
2211     }
2212     cluster = reclist_insert(se->reclist, se->relevance, service, record,
2213                              merge_keys, &se->total_merged);
2214     if (!cluster)
2215     {
2216         return 0; // complete match with existing record
2217     }
2218
2219     {
2220         const char *use_term_factor_str =
2221             session_setting_oneval(sdb, PZ_TERMLIST_TERM_FACTOR);
2222         if (use_term_factor_str && use_term_factor_str[0] == '1')
2223         {
2224             int maxrecs = client_get_maxrecs(cl);
2225             int hits = (int) client_get_hits(cl);
2226             term_factor = MAX(hits, maxrecs) /  MAX(1, maxrecs);
2227             assert(term_factor >= 1);
2228             session_log(se, YLOG_DEBUG, "Using term factor: %d (%d / %d)",
2229                         term_factor, MAX(hits, maxrecs), MAX(1, maxrecs));
2230         }
2231     }
2232
2233     if (global_parameters.dump_records)
2234         session_log(se, YLOG_LOG, "Cluster id %s from %s (#%d)", cluster->recid,
2235                     sdb->database->id, record_no);
2236
2237     // original metadata, to check if first existence of a field
2238     metadata0 = xmalloc(sizeof(*metadata0) * service->num_metadata);
2239     memcpy(metadata0, cluster->metadata,
2240            sizeof(*metadata0) * service->num_metadata);
2241
2242     ingest_nmem = nmem_create();
2243     if (se->rank)
2244     {
2245         yaz_log(YLOG_LOG, "local in sort : %s", se->rank);
2246         nmem_strsplit_escape2(ingest_nmem, ",", se->rank, &rank_values,
2247                               &rank_num, 1, '\\', 1);
2248     }
2249
2250     // now parsing XML record and adding data to cluster or record metadata
2251     for (n = root->children; n; n = n->next)
2252     {
2253         if (n->type != XML_ELEMENT_NODE)
2254             continue;
2255         if (!strcmp((const char *) n->name, "metadata"))
2256         {
2257             struct conf_metadata *ser_md = 0;
2258             struct conf_sortkey *ser_sk = 0;
2259             struct record_metadata **wheretoput = 0;
2260             struct record_metadata *rec_md = 0;
2261             int md_field_id = -1;
2262             int sk_field_id = -1;
2263             const char *rank = 0;
2264             const char *xml_rank = 0;
2265             const char *type = 0;
2266             xmlChar *value0;
2267
2268             type = yaz_xml_get_prop(n, "type");
2269             if (!type)
2270                 continue;
2271
2272             md_field_id
2273                 = conf_service_metadata_field_id(service, (const char *) type);
2274             if (md_field_id < 0)
2275                 continue;
2276
2277             ser_md = &service->metadata[md_field_id];
2278
2279             if (ser_md->sortkey_offset >= 0)
2280             {
2281                 sk_field_id = ser_md->sortkey_offset;
2282                 ser_sk = &service->sortkeys[sk_field_id];
2283             }
2284
2285             wrbuf_rewind(wrbuf_disp);
2286             wrbuf_rewind(wrbuf_norm);
2287
2288             value0 = xmlNodeListGetString(xdoc, n->children, 1);
2289             if (!value0 || !*value0)
2290             {
2291                 if (value0)
2292                     xmlFree(value0);
2293                 continue;
2294             }
2295
2296             if (ser_md->icurule)
2297             {
2298                 run_icu(se, ser_md->icurule, (const char *) value0,
2299                         wrbuf_norm, wrbuf_disp);
2300                 yaz_log(YLOG_LOG, "run_icu input=%s norm=%s disp=%s",
2301                         (const char *) value0,
2302                         wrbuf_cstr(wrbuf_norm), wrbuf_cstr(wrbuf_disp));
2303                 rec_md = record_metadata_init(se->nmem, wrbuf_cstr(wrbuf_disp),
2304                                               wrbuf_cstr(wrbuf_norm),
2305                                               ser_md->type, 0);
2306             }
2307             else
2308             {
2309                 wrbuf_puts(wrbuf_disp, (const char *) value0);
2310                 rec_md = record_metadata_init(se->nmem, wrbuf_cstr(wrbuf_disp),
2311                                               0,
2312                                               ser_md->type, 0);
2313             }
2314
2315             xmlFree(value0);
2316
2317             // see if the field was not in cluster already (from beginning)
2318             if (!rec_md)
2319                 continue;
2320
2321             if (rank_num)
2322             {
2323                 int i;
2324                 for (i = 0; i < rank_num; i++)
2325                 {
2326                     const char *val = rank_values[i];
2327                     const char *cp = strchr(val, '=');
2328                     if (!cp)
2329                         continue;
2330                     if ((cp - val) == strlen((const char *) type)
2331                         && !memcmp(val, type, cp - val))
2332                     {
2333                         rank = cp + 1;
2334                         break;
2335                     }
2336                 }
2337             }
2338             else
2339             {
2340                 xml_rank = yaz_xml_get_prop(n, "rank");
2341                 rank = xml_rank ? (const char *) xml_rank : ser_md->rank;
2342             }
2343
2344             wheretoput = &cluster->metadata[md_field_id];
2345
2346             if (ser_md->merge == Metadata_merge_first)
2347             {
2348                 if (!metadata0[md_field_id])
2349                 {
2350                     while (*wheretoput)
2351                         wheretoput = &(*wheretoput)->next;
2352                     *wheretoput = rec_md;
2353                 }
2354             }
2355             else if (ser_md->merge == Metadata_merge_unique)
2356             {
2357                 while (*wheretoput)
2358                 {
2359                     if (!strcmp((const char *) (*wheretoput)->data.text.norm,
2360                                 rec_md->data.text.norm))
2361                         break;
2362                     wheretoput = &(*wheretoput)->next;
2363                 }
2364                 if (!*wheretoput)
2365                     *wheretoput = rec_md;
2366             }
2367             else if (ser_md->merge == Metadata_merge_longest)
2368             {
2369                 if (!*wheretoput
2370                     || strlen(rec_md->data.text.norm)
2371                     > strlen((*wheretoput)->data.text.norm))
2372                 {
2373                     *wheretoput = rec_md;
2374                     if (ser_sk)
2375                     {
2376                         pp2_charset_token_t prt;
2377                         const char *sort_str = 0;
2378                         int skip_article =
2379                             ser_sk->type == Metadata_type_skiparticle;
2380
2381                         if (!cluster->sortkeys[sk_field_id])
2382                             cluster->sortkeys[sk_field_id] =
2383                                 nmem_malloc(se->nmem,
2384                                             sizeof(union data_types));
2385
2386                         prt =
2387                             pp2_charset_token_create(service->charsets, "sort");
2388
2389                         pp2_charset_token_first(prt, rec_md->data.text.disp,
2390                                                 skip_article);
2391
2392                         pp2_charset_token_next(prt);
2393
2394                         sort_str = pp2_get_sort(prt);
2395
2396                         cluster->sortkeys[sk_field_id]->text.disp =
2397                             rec_md->data.text.disp;
2398                         if (!sort_str)
2399                         {
2400                             sort_str = rec_md->data.text.disp;
2401                             session_log(se, YLOG_WARN,
2402                                     "Could not make sortkey. Bug #1858");
2403                         }
2404                         cluster->sortkeys[sk_field_id]->text.sort =
2405                             nmem_strdup(se->nmem, sort_str);
2406                         pp2_charset_token_destroy(prt);
2407                     }
2408                 }
2409             }
2410             else if (ser_md->merge == Metadata_merge_all)
2411             {
2412                 while (*wheretoput)
2413                     wheretoput = &(*wheretoput)->next;
2414                 *wheretoput = rec_md;
2415             }
2416             else if (ser_md->merge == Metadata_merge_range)
2417             {
2418                 if (!*wheretoput)
2419                 {
2420                     *wheretoput = rec_md;
2421                     if (ser_sk)
2422                         cluster->sortkeys[sk_field_id]
2423                             = &rec_md->data;
2424                 }
2425                 else
2426                 {
2427                     int this_min = rec_md->data.number.min;
2428                     int this_max = rec_md->data.number.max;
2429                     if (this_min < (*wheretoput)->data.number.min)
2430                         (*wheretoput)->data.number.min = this_min;
2431                     if (this_max > (*wheretoput)->data.number.max)
2432                         (*wheretoput)->data.number.max = this_max;
2433                 }
2434             }
2435
2436             // ranking of _all_ fields enabled ...
2437             if (rank)
2438             {
2439                 relevance_countwords(se->relevance, cluster,
2440                                      wrbuf_cstr(wrbuf_disp),
2441                                      rank, ser_md->name);
2442             }
2443             // construct facets ... unless the client already has reported them
2444             if (ser_md->termlist && !client_has_facet(cl, (char *) type))
2445             {
2446                 if (ser_md->type == Metadata_type_year)
2447                 {
2448                     char year[64];
2449                     sprintf(year, "%d", rec_md->data.number.max);
2450
2451                     add_facet(se, (char *) type, year, term_factor, cl);
2452                     if (rec_md->data.number.max != rec_md->data.number.min)
2453                     {
2454                         sprintf(year, "%d", rec_md->data.number.min);
2455                         add_facet(se, (char *) type, year, term_factor, cl);
2456                     }
2457                 }
2458                 else
2459                     add_facet(se, type, wrbuf_cstr(wrbuf_disp), term_factor, cl);
2460             }
2461         }
2462         else
2463         {
2464             if (se->number_of_warnings_unknown_elements == 0)
2465                 session_log(se, YLOG_WARN,
2466                         "Unexpected element in internal record: %s", n->name);
2467             se->number_of_warnings_unknown_elements++;
2468         }
2469     }
2470     nmem_destroy(ingest_nmem);
2471     xfree(metadata0);
2472     relevance_donerecord(se->relevance, cluster);
2473     se->total_records++;
2474
2475     return 0;
2476 }
2477
2478 void session_log(struct session *s, int level, const char *fmt, ...)
2479 {
2480     char buf[1024];
2481     va_list ap;
2482     va_start(ap, fmt);
2483
2484     yaz_vsnprintf(buf, sizeof(buf)-30, fmt, ap);
2485     yaz_log(level, "Session %u: %s", s ? s->session_id : 0, buf);
2486
2487     va_end(ap);
2488 }
2489
2490 /*
2491  * Local variables:
2492  * c-basic-offset: 4
2493  * c-file-style: "Stroustrup"
2494  * indent-tabs-mode: nil
2495  * End:
2496  * vim: shiftwidth=4 tabstop=8 expandtab
2497  */
2498