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_get_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
 274     session_normalize_facet(s, type, value, display_wrbuf, facet_wrbuf);
 275     if (wrbuf_len(facet_wrbuf))
 276     {
 277         struct named_termlist **tp = &s->termlists;
 278         for (; (*tp); tp = &(*tp)->next)
 279             if (!strcmp((*tp)->name, type))
 280                 break;
 281         if (!*tp)
 282         {
 283             *tp = nmem_malloc(s->nmem, sizeof(**tp));
 284             (*tp)->name = nmem_strdup(s->nmem, type);
 285             (*tp)->termlist = termlist_create(s->nmem);
 286             (*tp)->next = 0;
 287         }
 288         termlist_insert((*tp)->termlist, wrbuf_cstr(display_wrbuf),
 289                         wrbuf_cstr(facet_wrbuf), id, id_len, count);
 290         if (id)
 291             session_add_id_facet(s, cl, type, id, id_len,
 292                                  wrbuf_cstr(display_wrbuf));
 293     }
 294     wrbuf_destroy(facet_wrbuf);
 295     wrbuf_destroy(display_wrbuf);
 296 }
 297
 298 static xmlDoc *record_to_xml(struct session *se,
 299                              struct session_database *sdb, const char *rec)
 300 {
 301     struct database *db = sdb->database;
 302     xmlDoc *rdoc = 0;
 303
 304     rdoc = xmlParseMemory(rec, strlen(rec));
 305
 306     if (!rdoc)
 307     {
 308         session_log(se, YLOG_WARN, "Non-wellformed XML");
 309         return 0;
 310     }
 311
 312     if (global_parameters.dump_records)
 313     {
 314         session_log(se, YLOG_LOG, "Un-normalized record from %s", db->id);
 315         log_xml_doc(rdoc);
 316     }
 317
 318     return rdoc;
 319 }
 320
 321 #define MAX_XSLT_ARGS 16
 322
 323 // Add static values from session database settings if applicable
 324 static void insert_settings_parameters(struct session_database *sdb,
 325                                        struct conf_service *service,
 326                                        char **parms,
 327                                        NMEM nmem)
 328 {
 329     int i;
 330     int nparms = 0;
 331     int offset = 0;
 332
 333     for (i = 0; i < service->num_metadata; i++)
 334     {
 335         struct conf_metadata *md = &service->metadata[i];
 336         int setting;
 337
 338         if (md->setting == Metadata_setting_parameter &&
 339             (setting = settings_lookup_offset(service, md->name)) >= 0)
 340         {
 341             const char *val = session_setting_oneval(sdb, setting);
 342             if (val && nparms < MAX_XSLT_ARGS)
 343             {
 344                 char *buf;
 345                 int len = strlen(val);
 346                 buf = nmem_malloc(nmem, len + 3);
 347                 buf[0] = '\'';
 348                 strcpy(buf + 1, val);
 349                 buf[len+1] = '\'';
 350                 buf[len+2] = '\0';
 351                 parms[offset++] = md->name;
 352                 parms[offset++] = buf;
 353                 nparms++;
 354             }
 355         }
 356     }
 357     parms[offset] = 0;
 358 }
 359
 360 // Add static values from session database settings if applicable
 361 static void insert_settings_values(struct session_database *sdb, xmlDoc *doc,
 362                                    xmlNode *root,
 363                                    struct conf_service *service)
 364 {
 365     int i;
 366
 367     for (i = 0; i < service->num_metadata; i++)
 368     {
 369         struct conf_metadata *md = &service->metadata[i];
 370         int offset;
 371
 372         if (md->setting == Metadata_setting_postproc &&
 373             (offset = settings_lookup_offset(service, md->name)) >= 0)
 374         {
 375             const char *val = session_setting_oneval(sdb, offset);
 376             if (val)
 377             {
 378                 xmlNode *n = xmlNewTextChild(root, 0, (xmlChar *) "metadata",
 379                                              (xmlChar *) val);
 380                 xmlSetProp(n, (xmlChar *) "type", (xmlChar *) md->name);
 381             }
 382         }
 383     }
 384 }
 385
 386 static xmlDoc *normalize_record(struct session *se,
 387                                 struct session_database *sdb,
 388                                 struct conf_service *service,
 389                                 const char *rec, NMEM nmem)
 390 {
 391     xmlDoc *rdoc = record_to_xml(se, sdb, rec);
 392
 393     if (rdoc)
 394     {
 395         char *parms[MAX_XSLT_ARGS*2+1];
 396
 397         insert_settings_parameters(sdb, service, parms, nmem);
 398
 399         if (normalize_record_transform(sdb->map, &rdoc, (const char **)parms))
 400         {
 401             session_log(se, YLOG_WARN, "Normalize failed");
 402         }
 403     }
 404     return rdoc;
 405 }
 406
 407 void session_settings_dump(struct session *se,
 408                            struct session_database *db,
 409                            WRBUF w)
 410 {
 411     if (db->settings)
 412     {
 413         int i, num = db->num_settings;
 414         for (i = 0; i < num; i++)
 415         {
 416             struct setting *s = db->settings[i];
 417             for (;s ; s = s->next)
 418             {
 419                 wrbuf_puts(w, "<set name=\"");
 420                 wrbuf_xmlputs(w, s->name);
 421                 wrbuf_puts(w, "\" value=\"");
 422                 wrbuf_xmlputs(w, s->value);
 423                 wrbuf_puts(w, "\"/>");
 424             }
 425             if (db->settings[i])
 426                 wrbuf_puts(w, "\n");
 427         }
 428     }
 429 }
 430
 431 // Retrieve first defined value for 'name' for given database.
 432 // Will be extended to take into account user associated with session
 433 const char *session_setting_oneval(struct session_database *db, int offset)
 434 {
 435     if (offset >= db->num_settings || !db->settings[offset])
 436         return "";
 437     return db->settings[offset]->value;
 438 }
 439
 440 // Prepare XSLT stylesheets for record normalization
 441 // Structures are allocated on the session_wide nmem to avoid having
 442 // to recompute this for every search. This would lead
 443 // to leaking if a single session was to repeatedly change the PZ_XSLT
 444 // setting. However, this is not a realistic use scenario.
 445 static int prepare_map(struct session *se, struct session_database *sdb)
 446 {
 447     if (sdb->settings && !sdb->map)
 448     {
 449         const char *s;
 450
 451         if (sdb->settings[PZ_XSLT] &&
 452             (s = session_setting_oneval(sdb, PZ_XSLT)))
 453         {
 454             char auto_stylesheet[256];
 455
 456             if (!strcmp(s, "auto"))
 457             {
 458                 const char *request_syntax = session_setting_oneval(
 459                     sdb, PZ_REQUESTSYNTAX);
 460                 if (request_syntax)
 461                 {
 462                     char *cp;
 463                     yaz_snprintf(auto_stylesheet, sizeof(auto_stylesheet),
 464                                  "%s.xsl", request_syntax);
 465                     for (cp = auto_stylesheet; *cp; cp++)
 466                     {
 467                         /* deliberately only consider ASCII */
 468                         if (*cp > 32 && *cp < 127)
 469                             *cp = tolower(*cp);
 470                     }
 471                     s = auto_stylesheet;
 472                 }
 473                 else
 474                 {
 475                     session_log(se, YLOG_WARN,
 476                                 "No pz:requestsyntax for auto stylesheet");
 477                 }
 478             }
 479             sdb->map = normalize_cache_get(se->normalize_cache,
 480                                            se->service, s);
 481             if (!sdb->map)
 482                 return -1;
 483         }
 484     }
 485     return 0;
 486 }
 487
 488 // called if watch should be removed because http_channel is to be destroyed
 489 static void session_watch_cancel(void *data, struct http_channel *c,
 490                                  void *data2)
 491 {
 492     struct session_watchentry *ent = data;
 493
 494     ent->fun = 0;
 495     ent->data = 0;
 496     ent->obs = 0;
 497 }
 498
 499 // set watch. Returns 0=OK, -1 if watch is already set
 500 int session_set_watch(struct session *s, int what,
 501                       session_watchfun fun, void *data,
 502                       struct http_channel *chan)
 503 {
 504     int ret;
 505     session_enter(s, "session_set_watch");
 506     if (s->watchlist[what].fun)
 507         ret = -1;
 508     else
 509     {
 510
 511         s->watchlist[what].fun = fun;
 512         s->watchlist[what].data = data;
 513         s->watchlist[what].obs = http_add_observer(chan, &s->watchlist[what],
 514                                                    session_watch_cancel);
 515         ret = 0;
 516     }
 517     session_leave(s, "session_set_watch");
 518     return ret;
 519 }
 520
 521 void session_alert_watch(struct session *s, int what)
 522 {
 523     assert(s);
 524     session_enter(s, "session_alert_watch");
 525     if (s->watchlist[what].fun)
 526     {
 527         /* our watch is no longer associated with http_channel */
 528         void *data;
 529         session_watchfun fun;
 530
 531         http_remove_observer(s->watchlist[what].obs);
 532         fun  = s->watchlist[what].fun;
 533         data = s->watchlist[what].data;
 534
 535         /* reset watch before fun is invoked - in case fun wants to set
 536            it again */
 537         s->watchlist[what].fun = 0;
 538         s->watchlist[what].data = 0;
 539         s->watchlist[what].obs = 0;
 540
 541         session_leave(s, "session_alert_watch");
 542         session_log(s, YLOG_DEBUG,
 543                     "Alert Watch: %d calling function: %p", what, fun);
 544         fun(data);
 545     }
 546     else
 547         session_leave(s,"session_alert_watch");
 548 }
 549
 550 //callback for grep_databases
 551 static void select_targets_callback(struct session *se,
 552                                     struct session_database *db)
 553 {
 554     struct client *cl;
 555     struct client_list *l;
 556
 557     for (l = se->clients_cached; l; l = l->next)
 558         if (client_get_database(l->client) == db)
 559             break;
 560
 561     if (l)
 562         cl = l->client;
 563     else
 564     {
 565         cl = client_create(db->database->id);
 566         client_set_database(cl, db);
 567
 568         l = xmalloc(sizeof(*l));
 569         l->client = cl;
 570         l->next = se->clients_cached;
 571         se->clients_cached = l;
 572     }
 573     client_set_session(cl, se);
 574
 575     l = xmalloc(sizeof(*l));
 576     l->client = cl;
 577     l->next = se->clients_active;
 578     se->clients_active = l;
 579 }
 580
 581 static void session_reset_active_clients(struct session *se,
 582                                          struct client_list *new_list)
 583 {
 584     struct client_list *l;
 585
 586     session_enter(se, "session_reset_active_clients");
 587     l = se->clients_active;
 588     se->clients_active = new_list;
 589     session_leave(se, "session_reset_active_clients");
 590
 591     while (l)
 592     {
 593         struct client_list *l_next = l->next;
 594
 595         client_lock(l->client);
 596         client_set_session(l->client, 0); /* mark client inactive */
 597         client_unlock(l->client);
 598
 599         xfree(l);
 600         l = l_next;
 601     }
 602 }
 603
 604 static void session_remove_cached_clients(struct session *se)
 605 {
 606     struct client_list *l;
 607
 608     session_reset_active_clients(se, 0);
 609
 610     session_enter(se, "session_remove_cached_clients");
 611     l = se->clients_cached;
 612     se->clients_cached = 0;
 613     session_leave(se, "session_remove_cached_clients");
 614
 615     while (l)
 616     {
 617         struct client_list *l_next = l->next;
 618         client_lock(l->client);
 619         client_set_session(l->client, 0);
 620         client_set_database(l->client, 0);
 621         client_mark_dead(l->client);
 622         client_unlock(l->client);
 623         client_destroy(l->client);
 624         xfree(l);
 625         l = l_next;
 626     }
 627 }
 628
 629 // Associates a set of clients with a session;
 630 // Note: Session-databases represent databases with per-session
 631 // setting overrides
 632 static int select_targets(struct session *se, const char *filter)
 633 {
 634     return session_grep_databases(se, filter, select_targets_callback);
 635 }
 636
 637 int session_active_clients(struct session *s)
 638 {
 639     struct client_list *l;
 640     int res = 0;
 641
 642     for (l = s->clients_active; l; l = l->next)
 643         if (client_is_active(l->client))
 644             res++;
 645
 646     return res;
 647 }
 648
 649 int session_is_preferred_clients_ready(struct session *s)
 650 {
 651     struct client_list *l;
 652     int res = 0;
 653
 654     for (l = s->clients_active; l; l = l->next)
 655         if (client_is_active_preferred(l->client))
 656             res++;
 657     session_log(s, YLOG_DEBUG, "Has %d active preferred clients.", res);
 658     return res == 0;
 659 }
 660
 661 static void session_clear_set(struct session *se, struct reclist_sortparms *sp)
 662 {
 663     reclist_destroy(se->reclist);
 664     if (nmem_total(se->nmem))
 665         session_log(se, YLOG_DEBUG, "NMEN operation usage %zd",
 666                     nmem_total(se->nmem));
 667     nmem_reset(se->nmem);
 668     se->total_records = se->total_merged = 0;
 669     se->termlists = 0;
 670     relevance_clear(se->relevance);
 671
 672     /* reset list of sorted results and clear to relevance search */
 673     se->sorted_results = nmem_malloc(se->nmem, sizeof(*se->sorted_results));
 674     se->sorted_results->name = nmem_strdup(se->nmem, sp->name);
 675     se->sorted_results->increasing = sp->increasing;
 676     se->sorted_results->type = sp->type;
 677     se->sorted_results->next = 0;
 678
 679     session_log(se, YLOG_DEBUG, "clear_set session_sort: field=%s increasing=%d type=%d configured",
 680             sp->name, sp->increasing, sp->type);
 681
 682     se->reclist = reclist_create(se->nmem);
 683 }
 684
 685 void session_sort(struct session *se, struct reclist_sortparms *sp,
 686                   const char *mergekey, const char *rank)
 687 {
 688     struct client_list *l;
 689     const char *field = sp->name;
 690     int increasing = sp->increasing;
 691     int type  = sp->type;
 692     int clients_research = 0;
 693
 694     session_enter(se, "session_sort");
 695     session_log(se, YLOG_DEBUG, "session_sort field=%s increasing=%d type=%d",
 696                 field, increasing, type);
 697
 698     if (rank && (!se->rank || strcmp(se->rank, rank)))
 699     {
 700         /* new rank must research/reingest anyway */
 701         assert(rank);
 702         xfree(se->rank);
 703         se->rank = *rank ? xstrdup(rank) : 0;
 704         clients_research = 1;
 705         session_log(se, YLOG_DEBUG, "session_sort: new rank = %s",
 706                     rank);
 707     }
 708     if (mergekey && (!se->mergekey || strcmp(se->mergekey, mergekey)))
 709     {
 710         /* new mergekey must research/reingest anyway */
 711         assert(mergekey);
 712         xfree(se->mergekey);
 713         se->mergekey = *mergekey ? xstrdup(mergekey) : 0;
 714         clients_research = 1;
 715         session_log(se, YLOG_DEBUG, "session_sort: new mergekey = %s",
 716                     mergekey);
 717     }
 718     if (clients_research == 0)
 719     {
 720         struct reclist_sortparms *sr;
 721         for (sr = se->sorted_results; sr; sr = sr->next)
 722             if (!reclist_sortparms_cmp(sr, sp))
 723                 break;
 724         if (sr)
 725         {
 726             session_log(se, YLOG_LOG, "session_sort: field=%s increasing=%d type=%d already fetched",
 727                         field, increasing, type);
 728             session_leave(se, "session_sort");
 729             return;
 730         }
 731     }
 732     session_log(se, YLOG_DEBUG, "session_sort: field=%s increasing=%d type=%d must fetch",
 733                 field, increasing, type);
 734
 735     // We need to reset reclist on every sort that changes the records, not just for position
 736     // So if just one client requires new searching, we need to clear set.
 737     // Ask each of the client if sorting requires re-search due to native sort
 738     // If it does it will require us to
 739     for (l = se->clients_active; l; l = l->next)
 740     {
 741         struct client *cl = l->client;
 742         // Assume no re-search is required.
 743         client_parse_init(cl, 1);
 744         clients_research += client_parse_sort(cl, sp, 0);
 745     }
 746     if (!clients_research || se->clients_starting)
 747     {
 748         // A new sorting based on same record set
 749         struct reclist_sortparms *sr = nmem_malloc(se->nmem, sizeof(*sr));
 750         sr->name = nmem_strdup(se->nmem, field);
 751         sr->increasing = increasing;
 752         sr->type = type;
 753         sr->next = se->sorted_results;
 754         se->sorted_results = sr;
 755         session_log(se, YLOG_DEBUG, "session_sort: no research/ingesting done");
 756         session_leave(se, "session_sort");
 757     }
 758     else
 759     {
 760         se->clients_starting = 1;
 761         session_log(se, YLOG_DEBUG,
 762                     "session_sort: reset results due to %d clients researching",
 763                     clients_research);
 764         session_clear_set(se, sp);
 765         session_log(se, YLOG_DEBUG, "Re- search/ingesting for clients due to change in sort order");
 766
 767         session_leave(se, "session_sort");
 768         for (l = se->clients_active; l; l = l->next)
 769         {
 770             struct client *cl = l->client;
 771             if (client_get_state(cl) == Client_Connecting ||
 772                 client_get_state(cl) == Client_Idle ||
 773                 client_get_state(cl) == Client_Working) {
 774                 client_start_search(cl);
 775             }
 776             else
 777             {
 778                 session_log(se, YLOG_DEBUG,
 779                             "session_sort: %s: No re-start/ingest in show. "
 780                             "Wrong client state: %d",
 781                             client_get_id(cl), client_get_state(cl));
 782             }
 783         }
 784         session_enter(se, "session_sort");
 785         se->clients_starting = 0;
 786         se->force_position = 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     int no_sortmap = 0;
 835     struct client_list *l;
 836
 837     session_log(se, YLOG_LOG, "search query %s", query);
 838     if (filter)
 839         session_log(se, YLOG_LOG, "search filter %s", filter);
 840     if (limit)
 841         session_log(se, YLOG_LOG, "search limit %s", limit);
 842
 843     *addinfo = 0;
 844
 845     session_enter(se, "session_search0");
 846     if (se->clients_starting)
 847     {
 848         session_leave(se, "session_search0");
 849         return PAZPAR2_NO_ERROR;
 850     }
 851     se->clients_starting = 1;
 852     se->force_position = 0;
 853     session_leave(se, "session_search0");
 854
 855     if (se->settings_modified) {
 856         session_remove_cached_clients(se);
 857     }
 858     else
 859         session_reset_active_clients(se, 0);
 860
 861     session_enter(se, "session_search");
 862     se->settings_modified = 0;
 863
 864     if (mergekey)
 865     {
 866         xfree(se->mergekey);
 867         se->mergekey = *mergekey ? xstrdup(mergekey) : 0;
 868     }
 869     if (rank)
 870     {
 871         xfree(se->rank);
 872         se->rank = *rank ? xstrdup(rank) : 0;
 873     }
 874
 875     session_clear_set(se, sp);
 876     relevance_destroy(&se->relevance);
 877
 878     live_channels = select_targets(se, filter);
 879     if (!live_channels)
 880     {
 881         session_leave(se, "session_search");
 882         se->clients_starting = 0;
 883         return PAZPAR2_NO_TARGETS;
 884     }
 885
 886     facet_limits_destroy(se->facet_limits);
 887     se->facet_limits = facet_limits_create(limit);
 888     if (!se->facet_limits)
 889     {
 890         *addinfo = "limit";
 891         session_leave(se, "session_search");
 892         se->clients_starting = 0;
 893         session_reset_active_clients(se, 0);
 894         return PAZPAR2_MALFORMED_PARAMETER_VALUE;
 895     }
 896
 897     session_leave(se, "session_search");
 898
 899     session_alert_watch(se, SESSION_WATCH_SHOW);
 900     session_alert_watch(se, SESSION_WATCH_BYTARGET);
 901     session_alert_watch(se, SESSION_WATCH_TERMLIST);
 902     session_alert_watch(se, SESSION_WATCH_SHOW_PREF);
 903
 904     for (l = se->clients_active; l; l = l->next)
 905     {
 906         int parse_ret;
 907         struct client *cl = l->client;
 908         client_parse_init(cl, 1);
 909         if (prepare_map(se, client_get_database(cl)) < 0)
 910             continue;
 911
 912         parse_ret = client_parse_query(cl, query, se->facet_limits, addinfo2);
 913         if (parse_ret == -1)
 914             no_failed_query++;
 915         else if (parse_ret == -2)
 916             no_failed_limit++;
 917         else if (parse_ret < 0)
 918             no_working++; /* other error, such as bad CCL map */
 919         else
 920         {
 921             client_parse_range(cl, startrecs, maxrecs);
 922             client_parse_sort(cl, sp, &no_sortmap);
 923             client_start_search(cl);
 924             no_working++;
 925         }
 926     }
 927     session_log(se, YLOG_LOG, "search "
 928                 "working %d sortmap %d failed-query %d failed-limit %d",
 929                 no_working, no_sortmap, no_failed_query, no_failed_limit);
 930     session_enter(se, "session_search2");
 931     if (no_working == 1 && no_sortmap == 1)
 932     {
 933         se->force_position = 1;
 934         yaz_log(YLOG_LOG, "force_position=1");
 935     }
 936     se->clients_starting = 0;
 937     session_leave(se, "session_search2");
 938     if (no_working == 0)
 939     {
 940         if (no_failed_query > 0)
 941         {
 942             *addinfo = "query";
 943             return PAZPAR2_MALFORMED_PARAMETER_VALUE;
 944         }
 945         else if (no_failed_limit > 0)
 946         {
 947             *addinfo = "limit";
 948             return PAZPAR2_MALFORMED_PARAMETER_VALUE;
 949         }
 950         else
 951             return PAZPAR2_NO_TARGETS;
 952     }
 953     return PAZPAR2_NO_ERROR;
 954 }
 955
 956 // Creates a new session_database object for a database
 957 static void session_init_databases_fun(void *context, struct database *db)
 958 {
 959     struct session *se = (struct session *) context;
 960     struct session_database *new = nmem_malloc(se->session_nmem, sizeof(*new));
 961     int i;
 962
 963     new->database = db;
 964
 965     new->map = 0;
 966     assert(db->settings);
 967     new->settings = nmem_malloc(se->session_nmem,
 968                                 sizeof(struct settings *) * db->num_settings);
 969     new->num_settings = db->num_settings;
 970     for (i = 0; i < db->num_settings; i++)
 971     {
 972         struct setting *setting = db->settings[i];
 973         new->settings[i] = setting;
 974     }
 975     new->next = se->databases;
 976     se->databases = new;
 977 }
 978
 979 // Doesn't free memory associated with sdb -- nmem takes care of that
 980 static void session_database_destroy(struct session_database *sdb)
 981 {
 982     sdb->map = 0;
 983 }
 984
 985 // Initialize session_database list -- this represents this session's view
 986 // of the database list -- subject to modification by the settings ws command
 987 void session_init_databases(struct session *se)
 988 {
 989     se->databases = 0;
 990     predef_grep_databases(se, se->service, session_init_databases_fun);
 991 }
 992
 993 // Probably session_init_databases_fun should be refactored instead of
 994 // called here.
 995 static struct session_database *load_session_database(struct session *se,
 996                                                       const char *id)
 997 {
 998     struct database *db = new_database_inherit_settings(id, se->session_nmem, se->service->settings);
 999     session_init_databases_fun((void*) se, db);
1000
1001     // New sdb is head of se->databases list
1002     return se->databases;
1003 }
1004
1005 // Find an existing session database. If not found, load it
1006 static struct session_database *find_session_database(struct session *se,
1007                                                       const char *id)
1008 {
1009     struct session_database *sdb;
1010
1011     for (sdb = se->databases; sdb; sdb = sdb->next)
1012         if (!strcmp(sdb->database->id, id))
1013             return sdb;
1014     return load_session_database(se, id);
1015 }
1016
1017 // Apply a session override to a database
1018 void session_apply_setting(struct session *se, const char *dbname,
1019                            const char *name, const char *value)
1020 {
1021     session_enter(se, "session_apply_setting");
1022     {
1023         struct session_database *sdb = find_session_database(se, dbname);
1024         struct conf_service *service = se->service;
1025         struct setting *s;
1026         int offset = settings_create_offset(service, name);
1027
1028         expand_settings_array(&sdb->settings, &sdb->num_settings, offset,
1029                               se->session_nmem);
1030         // Force later recompute of settings-driven data structures
1031         // (happens when a search starts and client connections are prepared)
1032         if (offset == PZ_XSLT)
1033             sdb->map = 0;
1034         se->settings_modified = 1;
1035         for (s = sdb->settings[offset]; s; s = s->next)
1036             if (!strcmp(s->name, name) &&
1037                 dbname && s->target && !strcmp(dbname, s->target))
1038                 break;
1039         if (!s)
1040         {
1041             s = nmem_malloc(se->session_nmem, sizeof(*s));
1042             s->precedence = 0;
1043             s->target = nmem_strdup(se->session_nmem, dbname);
1044             s->name = nmem_strdup(se->session_nmem, name);
1045             s->next = sdb->settings[offset];
1046             sdb->settings[offset] = s;
1047         }
1048         s->value = nmem_strdup(se->session_nmem, value);
1049     }
1050     session_leave(se, "session_apply_setting");
1051 }
1052
1053 void session_destroy(struct session *se)
1054 {
1055     struct session_database *sdb;
1056     struct facet_id *t;
1057     int sessions_total = session_use(-1);
1058     int no_facet_ids = 0;
1059
1060     for (t = se->facet_id_list; t; t = t->next)
1061         no_facet_ids++;
1062     session_log(se, YLOG_LOG, "destroy "
1063                 "session-total %d nmem-op %zd nmem-ses %zd facets-ids %d",
1064                 sessions_total,
1065                 nmem_total(se->nmem), nmem_total(se->session_nmem),
1066                 no_facet_ids);
1067     session_remove_cached_clients(se);
1068
1069     for (sdb = se->databases; sdb; sdb = sdb->next)
1070         session_database_destroy(sdb);
1071     normalize_cache_destroy(se->normalize_cache);
1072     relevance_destroy(&se->relevance);
1073     reclist_destroy(se->reclist);
1074     xfree(se->mergekey);
1075     xfree(se->rank);
1076     facet_limits_destroy(se->facet_limits);
1077     nmem_destroy(se->nmem);
1078     service_destroy(se->service);
1079     yaz_mutex_destroy(&se->session_mutex);
1080 }
1081
1082 size_t session_get_memory_status(struct session *session) {
1083     size_t session_nmem;
1084     if (session == 0)
1085         return 0;
1086     session_enter(session, "session_get_memory_status");
1087     session_nmem = nmem_total(session->nmem);
1088     session_leave(session, "session_get_memory_status");
1089     return session_nmem;
1090 }
1091
1092
1093 struct session *session_create(NMEM nmem, struct conf_service *service,
1094                                unsigned session_id)
1095 {
1096     int i;
1097     struct session *session = nmem_malloc(nmem, sizeof(*session));
1098
1099     char tmp_str[50];
1100
1101     sprintf(tmp_str, "session#%u", session_id);
1102
1103     session->session_id = session_id;
1104     session_log(session, YLOG_DEBUG, "New");
1105     session->service = service;
1106     session->relevance = 0;
1107     session->total_records = 0;
1108     session->number_of_warnings_unknown_elements = 0;
1109     session->number_of_warnings_unknown_metadata = 0;
1110     session->termlists = 0;
1111     session->reclist = reclist_create(nmem);
1112     session->clients_active = 0;
1113     session->clients_cached = 0;
1114     session->settings_modified = 0;
1115     session->session_nmem = nmem;
1116     session->facet_id_list = 0;
1117     session->nmem = nmem_create();
1118     session->databases = 0;
1119     session->sorted_results = 0;
1120     session->facet_limits = 0;
1121     session->mergekey = 0;
1122     session->rank = 0;
1123     session->clients_starting = 0;
1124     session->force_position = 0;
1125
1126     for (i = 0; i <= SESSION_WATCH_MAX; i++)
1127     {
1128         session->watchlist[i].data = 0;
1129         session->watchlist[i].fun = 0;
1130     }
1131     session->normalize_cache = normalize_cache_create();
1132     session->session_mutex = 0;
1133     pazpar2_mutex_create(&session->session_mutex, tmp_str);
1134
1135     i = session_use(1);
1136     session_log(session, YLOG_LOG, "create session-total %d", i);
1137     return session;
1138 }
1139
1140 static struct hitsbytarget *hitsbytarget_nb(struct session *se,
1141                                             int *count, NMEM nmem)
1142 {
1143     struct hitsbytarget *res = 0;
1144     struct client_list *l;
1145     size_t sz = 0;
1146
1147     for (l = se->clients_active; l; l = l->next)
1148         sz++;
1149
1150     res = nmem_malloc(nmem, sizeof(*res) * sz);
1151     *count = 0;
1152     for (l = se->clients_active; l; l = l->next)
1153     {
1154         struct client *cl = l->client;
1155         WRBUF w = wrbuf_alloc();
1156         const char *name = session_setting_oneval(client_get_database(cl),
1157                                                   PZ_NAME);
1158         res[*count].id = client_get_id(cl);
1159         res[*count].name = *name ? name : "Unknown";
1160         res[*count].hits = client_get_hits(cl);
1161         res[*count].approximation = client_get_approximation(cl);
1162         res[*count].records = client_get_num_records(cl,
1163                                                      &res[*count].filtered,
1164                                                      0, 0);
1165         res[*count].diagnostic =
1166             client_get_diagnostic(cl, &res[*count].message,
1167                                   &res[*count].addinfo);
1168         res[*count].state = client_get_state_str(cl);
1169         res[*count].connected  = client_get_connection(cl) ? 1 : 0;
1170         session_settings_dump(se, client_get_database(cl), w);
1171         res[*count].settings_xml = nmem_strdup(nmem, wrbuf_cstr(w));
1172         wrbuf_rewind(w);
1173         res[*count].suggestions_xml =
1174             nmem_strdup(nmem, client_get_suggestions_xml(cl, w));
1175
1176         res[*count].query_data =
1177             client_get_query(cl, &res[*count].query_type, nmem);
1178         wrbuf_destroy(w);
1179         (*count)++;
1180     }
1181     return res;
1182 }
1183
1184 struct hitsbytarget *get_hitsbytarget(struct session *se, int *count, NMEM nmem)
1185 {
1186     struct hitsbytarget *p;
1187     session_enter(se, "get_hitsbytarget");
1188     p = hitsbytarget_nb(se, count, nmem);
1189     session_leave(se, "get_hitsbytarget");
1190     return p;
1191 }
1192
1193 // Compares two hitsbytarget nodes by hitcount
1194 static int cmp_ht(const void *p1, const void *p2)
1195 {
1196     const struct hitsbytarget *h1 = p1;
1197     const struct hitsbytarget *h2 = p2;
1198     return h2->hits - h1->hits;
1199 }
1200
1201 // Compares two hitsbytarget nodes by hitcount
1202 static int cmp_ht_approx(const void *p1, const void *p2)
1203 {
1204     const struct hitsbytarget *h1 = p1;
1205     const struct hitsbytarget *h2 = p2;
1206     return h2->approximation - h1->approximation;
1207 }
1208
1209 static int targets_termlist_nb(WRBUF wrbuf, struct session *se, int num,
1210                                NMEM nmem, int version)
1211 {
1212     struct hitsbytarget *ht;
1213     int count, i;
1214
1215     ht = hitsbytarget_nb(se, &count, nmem);
1216     if (version >= 2)
1217         qsort(ht, count, sizeof(struct hitsbytarget), cmp_ht_approx);
1218     else
1219         qsort(ht, count, sizeof(struct hitsbytarget), cmp_ht);
1220     for (i = 0; i < count && i < num && ht[i].hits > 0; i++)
1221     {
1222
1223         // do only print terms which have display names
1224
1225         wrbuf_puts(wrbuf, "<term>\n");
1226
1227         wrbuf_puts(wrbuf, "<id>");
1228         wrbuf_xmlputs(wrbuf, ht[i].id);
1229         wrbuf_puts(wrbuf, "</id>\n");
1230
1231         wrbuf_puts(wrbuf, "<name>");
1232         if (!ht[i].name || !ht[i].name[0])
1233             wrbuf_xmlputs(wrbuf, "NO TARGET NAME");
1234         else
1235             wrbuf_xmlputs(wrbuf, ht[i].name);
1236         wrbuf_puts(wrbuf, "</name>\n");
1237
1238         wrbuf_printf(wrbuf, "<frequency>" ODR_INT_PRINTF "</frequency>\n",
1239                      ht[i].hits);
1240
1241         if (version >= 2) {
1242             // Should not print if we know it isn't a approximation.
1243             wrbuf_printf(wrbuf, "<approximation>" ODR_INT_PRINTF "</approximation>\n", ht[i].approximation);
1244             wrbuf_printf(wrbuf, "<records>%d</records>\n", ht[i].records - ht[i].filtered);
1245             wrbuf_printf(wrbuf, "<filtered>%d</filtered>\n", ht[i].filtered);
1246         }
1247
1248         wrbuf_puts(wrbuf, "<state>");
1249         wrbuf_xmlputs(wrbuf, ht[i].state);
1250         wrbuf_puts(wrbuf, "</state>\n");
1251
1252         wrbuf_printf(wrbuf, "<diagnostic>%d</diagnostic>\n",
1253                      ht[i].diagnostic);
1254         wrbuf_puts(wrbuf, "</term>\n");
1255     }
1256     return count;
1257 }
1258
1259 void perform_termlist(struct http_channel *c, struct session *se,
1260                       const char *name, int num, int version)
1261 {
1262     int j;
1263     NMEM nmem_tmp = nmem_create();
1264     char **names;
1265     int num_names = 0;
1266
1267     if (!name)
1268         name = "*";
1269
1270     nmem_strsplit(nmem_tmp, ",", name, &names, &num_names);
1271
1272     session_enter(se, "perform_termlist");
1273
1274     for (j = 0; j < num_names; j++)
1275     {
1276         const char *tname;
1277         int must_generate_empty = 1; /* bug 5350 */
1278
1279         struct named_termlist *t = se->termlists;
1280         for (; t; t = t->next)
1281         {
1282             tname = t->name;
1283             if (!strcmp(names[j], tname) || !strcmp(names[j], "*"))
1284             {
1285                 struct termlist_score **p = 0;
1286                 int len;
1287
1288                 wrbuf_puts(c->wrbuf, "<list name=\"");
1289                 wrbuf_xmlputs(c->wrbuf, tname);
1290                 wrbuf_puts(c->wrbuf, "\">\n");
1291                 must_generate_empty = 0;
1292
1293                 p = termlist_highscore(t->termlist, &len, nmem_tmp);
1294                 if (p)
1295                 {
1296                     int i;
1297                     for (i = 0; i < len && i < num; i++)
1298                     {
1299                         // prevent sending empty term elements
1300                         if (!p[i]->display_term || !p[i]->display_term[0])
1301                             continue;
1302
1303                         wrbuf_puts(c->wrbuf, "<term>");
1304                         wrbuf_puts(c->wrbuf, "<name>");
1305                         wrbuf_xmlputs(c->wrbuf, p[i]->display_term);
1306                         wrbuf_puts(c->wrbuf, "</name>");
1307                         wrbuf_printf(c->wrbuf,
1308                                      "<frequency>%d</frequency>",
1309                                      p[i]->frequency);
1310                         wrbuf_puts(c->wrbuf, "</term>\n");
1311                     }
1312                 }
1313                 wrbuf_puts(c->wrbuf, "</list>\n");
1314             }
1315         }
1316         tname = "xtargets";
1317         if (!strcmp(names[j], tname) || !strcmp(names[j], "*"))
1318         {
1319             wrbuf_puts(c->wrbuf, "<list name=\"");
1320             wrbuf_xmlputs(c->wrbuf, tname);
1321             wrbuf_puts(c->wrbuf, "\">\n");
1322
1323             targets_termlist_nb(c->wrbuf, se, num, c->nmem, version);
1324             wrbuf_puts(c->wrbuf, "</list>\n");
1325             must_generate_empty = 0;
1326         }
1327         if (must_generate_empty)
1328         {
1329             wrbuf_puts(c->wrbuf, "<list name=\"");
1330             wrbuf_xmlputs(c->wrbuf, names[j]);
1331             wrbuf_puts(c->wrbuf, "\"/>\n");
1332         }
1333     }
1334     session_leave(se, "perform_termlist");
1335     nmem_destroy(nmem_tmp);
1336 }
1337
1338 #ifdef MISSING_HEADERS
1339 void report_nmem_stats(void)
1340 {
1341     size_t in_use, is_free;
1342
1343     nmem_get_memory_in_use(&in_use);
1344     nmem_get_memory_free(&is_free);
1345
1346     yaz_log(YLOG_LOG, "nmem stat: use=%ld free=%ld",
1347             (long) in_use, (long) is_free);
1348 }
1349 #endif
1350
1351 struct record_cluster *show_single_start(struct session *se, const char *id,
1352                                          struct record_cluster **prev_r,
1353                                          struct record_cluster **next_r)
1354 {
1355     struct record_cluster *r = 0;
1356
1357     session_enter(se, "show_single_start");
1358     *prev_r = 0;
1359     *next_r = 0;
1360     reclist_limit(se->reclist, se, 1);
1361
1362     reclist_enter(se->reclist);
1363     while ((r = reclist_read_record(se->reclist)))
1364     {
1365         if (!strcmp(r->recid, id))
1366         {
1367             *next_r = reclist_read_record(se->reclist);
1368             break;
1369         }
1370         *prev_r = r;
1371     }
1372     reclist_leave(se->reclist);
1373     if (!r)
1374         session_leave(se, "show_single_start");
1375     return r;
1376 }
1377
1378 void show_single_stop(struct session *se, struct record_cluster *rec)
1379 {
1380     session_leave(se, "show_single_stop");
1381 }
1382
1383
1384 int session_fetch_more(struct session *se)
1385 {
1386     struct client_list *l;
1387     int ret = 0;
1388
1389     for (l = se->clients_active; l; l = l->next)
1390     {
1391         struct client *cl = l->client;
1392         if (client_get_state(cl) == Client_Idle)
1393         {
1394             if (client_fetch_more(cl))
1395             {
1396                 session_log(se, YLOG_LOG, "%s: more to fetch",
1397                             client_get_id(cl));
1398                 ret = 1;
1399             }
1400             else
1401             {
1402                 int filtered;
1403                 int ingest_failures;
1404                 int record_failures;
1405                 int num = client_get_num_records(
1406                     cl, &filtered, &ingest_failures, &record_failures);
1407
1408                 session_log(se, YLOG_LOG, "%s: hits=" ODR_INT_PRINTF
1409                             " fetched=%d filtered=%d",
1410                             client_get_id(cl),
1411                             client_get_hits(cl),
1412                             num, filtered);
1413                 if (ingest_failures || record_failures)
1414                 {
1415                     session_log(se, YLOG_WARN, "%s:"
1416                                 " ingest failures=%d record failures=%d",
1417                                 client_get_id(cl),
1418                                 ingest_failures, record_failures);
1419                 }
1420             }
1421         }
1422         else
1423         {
1424             session_log(se, YLOG_LOG, "%s: no fetch due to state=%s",
1425                         client_get_id(cl), client_get_state_str(cl));
1426         }
1427
1428     }
1429     return ret;
1430 }
1431
1432 struct record_cluster **show_range_start(struct session *se,
1433                                          struct reclist_sortparms *sp,
1434                                          int start, int *num, int *total,
1435                                          Odr_int *sumhits, Odr_int *approx_hits,
1436                                          void (*show_records_ready)(void *data),
1437                                          struct http_channel *chan)
1438 {
1439     struct record_cluster **recs = 0;
1440     struct reclist_sortparms *spp;
1441     struct client_list *l;
1442     int i;
1443     NMEM nmem_tmp = 0;
1444 #if USE_TIMING
1445     yaz_timing_t t = yaz_timing_create();
1446 #endif
1447     session_enter(se, "show_range_start");
1448     *sumhits = 0;
1449     *approx_hits = 0;
1450     *total = 0;
1451     reclist_limit(se->reclist, se, 0);
1452     if (se->relevance)
1453     {
1454         for (spp = sp; spp; spp = spp->next)
1455             if (spp->type == Metadata_type_relevance)
1456             {
1457                 relevance_prepare_read(se->relevance, se->reclist);
1458                 break;
1459             }
1460         for (l = se->clients_active; l; l = l->next) {
1461             *sumhits += client_get_hits(l->client);
1462             *approx_hits += client_get_approximation(l->client);
1463         }
1464     }
1465     if (se->force_position)
1466     {
1467         nmem_tmp = nmem_create();
1468         sp = reclist_parse_sortparms(nmem_tmp, "position:1", 0);
1469         assert(sp);
1470     }
1471     reclist_sort(se->reclist, sp);
1472     if (nmem_tmp)
1473         nmem_destroy(nmem_tmp);
1474
1475     reclist_enter(se->reclist);
1476     *total = reclist_get_num_records(se->reclist);
1477
1478     for (l = se->clients_active; l; l = l->next)
1479         client_update_show_stat(l->client, 0);
1480
1481     for (i = 0; i < start; i++)
1482     {
1483         struct record_cluster *r = reclist_read_record(se->reclist);
1484         if (!r)
1485         {
1486             *num = 0;
1487             break;
1488         }
1489         else
1490         {
1491             struct record *rec = r->records;
1492             for (;rec; rec = rec->next)
1493                 client_update_show_stat(rec->client, 1);
1494         }
1495     }
1496     recs = nmem_malloc(se->nmem, (*num > 0 ? *num : 1) * sizeof(*recs));
1497     for (i = 0; i < *num; i++)
1498     {
1499         struct record_cluster *r = reclist_read_record(se->reclist);
1500         if (!r)
1501         {
1502             *num = i;
1503             break;
1504         }
1505         else
1506         {
1507             struct record *rec = r->records;
1508             for (;rec; rec = rec->next)
1509                 client_update_show_stat(rec->client, 1);
1510             recs[i] = r;
1511         }
1512     }
1513     reclist_leave(se->reclist);
1514 #if USE_TIMING
1515     yaz_timing_stop(t);
1516     session_log(se, YLOG_LOG, "show %6.5f %3.2f %3.2f",
1517             yaz_timing_get_real(t), yaz_timing_get_user(t),
1518             yaz_timing_get_sys(t));
1519     yaz_timing_destroy(&t);
1520 #endif
1521
1522     if (!session_fetch_more(se))
1523         session_log(se, YLOG_LOG, "can not fetch more");
1524     else
1525     {
1526         show_range_stop(se, recs);
1527         session_log(se, YLOG_LOG, "fetching more in progress");
1528         if (session_set_watch(se, SESSION_WATCH_SHOW,
1529                               show_records_ready, chan, chan))
1530         {
1531             session_log(se, YLOG_WARN, "Ignoring show block");
1532             session_enter(se, "show_range_start");
1533         }
1534         else
1535         {
1536             session_log(se, YLOG_LOG, "session watch OK");
1537             return 0;
1538         }
1539     }
1540     return recs;
1541 }
1542
1543 void show_range_stop(struct session *se, struct record_cluster **recs)
1544 {
1545     session_leave(se, "show_range_stop");
1546 }
1547
1548 void statistics(struct session *se, struct statistics *stat)
1549 {
1550     struct client_list *l;
1551     int count = 0;
1552
1553     memset(stat, 0, sizeof(*stat));
1554     stat->num_hits = 0;
1555     for (l = se->clients_active; l; l = l->next)
1556     {
1557         struct client *cl = l->client;
1558         if (!client_get_connection(cl))
1559             stat->num_no_connection++;
1560         stat->num_hits += client_get_hits(cl);
1561         switch (client_get_state(cl))
1562         {
1563         case Client_Connecting: stat->num_connecting++; break;
1564         case Client_Working: stat->num_working++; break;
1565         case Client_Idle: stat->num_idle++; break;
1566         case Client_Failed: stat->num_failed++; break;
1567         case Client_Error: stat->num_error++; break;
1568         default: break;
1569         }
1570         count++;
1571     }
1572     stat->num_records = se->total_records;
1573
1574     stat->num_clients = count;
1575 }
1576
1577 static struct record_metadata *record_metadata_init(
1578     NMEM nmem, const char *value, const char *norm,
1579     enum conf_metadata_type type,
1580     struct _xmlAttr *attr)
1581 {
1582     struct record_metadata *rec_md = record_metadata_create(nmem);
1583     struct record_metadata_attr **attrp = &rec_md->attributes;
1584
1585     for (; attr; attr = attr->next)
1586     {
1587         if (attr->children && attr->children->content)
1588         {
1589             if (strcmp((const char *) attr->name, "type")
1590                 && strcmp((const char *) attr->name, "empty"))
1591             {  /* skip the "type" + "empty" attribute..
1592                   The "Type" is already part of the element in output
1593                   (md-%s) and so repeating it here is redundant */
1594                 *attrp = nmem_malloc(nmem, sizeof(**attrp));
1595                 (*attrp)->name =
1596                     nmem_strdup(nmem, (const char *) attr->name);
1597                 (*attrp)->value =
1598                     nmem_strdup(nmem, (const char *) attr->children->content);
1599                 attrp = &(*attrp)->next;
1600             }
1601         }
1602     }
1603     *attrp = 0;
1604
1605     switch (type)
1606     {
1607     case Metadata_type_generic:
1608     case Metadata_type_skiparticle:
1609         if (norm)
1610         {
1611             rec_md->data.text.disp = nmem_strdup(nmem, value);
1612             rec_md->data.text.norm = nmem_strdup(nmem, norm);
1613         }
1614         else
1615         {
1616             if (strstr(value, "://")) /* looks like a URL */
1617                 rec_md->data.text.disp = nmem_strdup(nmem, value);
1618             else
1619                 rec_md->data.text.disp =
1620                     normalize7bit_generic(nmem_strdup(nmem, value), " ,/.:([");
1621             rec_md->data.text.norm = rec_md->data.text.disp;
1622         }
1623         rec_md->data.text.sort = 0;
1624         rec_md->data.text.snippet = 0;
1625         break;
1626     case Metadata_type_year:
1627     case Metadata_type_date:
1628     {
1629         int first, last;
1630         int longdate = 0;
1631
1632         if (type == Metadata_type_date)
1633             longdate = 1;
1634         if (extract7bit_dates((char *) value, &first, &last, longdate) < 0)
1635             return 0;
1636
1637         rec_md->data.number.min = first;
1638         rec_md->data.number.max = last;
1639     }
1640     break;
1641     case Metadata_type_float:
1642         rec_md->data.fnumber = atof(value);
1643         break;
1644     case Metadata_type_relevance:
1645     case Metadata_type_position:
1646     case Metadata_type_retrieval:
1647         return 0;
1648     }
1649     return rec_md;
1650 }
1651
1652 static void mergekey_norm_wr(pp2_charset_fact_t charsets,
1653                              WRBUF norm_wr, const char *value)
1654 {
1655     const char *norm_str;
1656     pp2_charset_token_t prt =
1657         pp2_charset_token_create(charsets, "mergekey");
1658
1659     pp2_charset_token_first(prt, value, 0);
1660     while ((norm_str = pp2_charset_token_next(prt)))
1661     {
1662         if (*norm_str)
1663         {
1664             if (wrbuf_len(norm_wr))
1665                 wrbuf_puts(norm_wr, " ");
1666             wrbuf_puts(norm_wr, norm_str);
1667         }
1668     }
1669     pp2_charset_token_destroy(prt);
1670 }
1671
1672 static int get_mergekey_from_doc(xmlDoc *doc, xmlNode *root, const char *name,
1673                                  struct conf_service *service, WRBUF norm_wr)
1674 {
1675     xmlNode *n;
1676     int no_found = 0;
1677     for (n = root->children; n; n = n->next)
1678     {
1679         if (n->type != XML_ELEMENT_NODE)
1680             continue;
1681         if (!strcmp((const char *) n->name, "metadata"))
1682         {
1683             const char *type = yaz_xml_get_prop(n, "type");
1684             if (type == NULL) {
1685                 yaz_log(YLOG_FATAL, "Missing type attribute on metadata element. Skipping!");
1686             }
1687             else if (!strcmp(name, (const char *) type))
1688             {
1689                 xmlChar *value = xmlNodeListGetString(doc, n->children, 1);
1690                 if (value && *value)
1691                 {
1692                     if (wrbuf_len(norm_wr) > 0)
1693                         wrbuf_puts(norm_wr, " ");
1694                     wrbuf_puts(norm_wr, name);
1695                     mergekey_norm_wr(service->charsets, norm_wr,
1696                                      (const char *) value);
1697                     no_found++;
1698                 }
1699                 if (value)
1700                     xmlFree(value);
1701             }
1702         }
1703     }
1704     return no_found;
1705 }
1706
1707 static const char *get_mergekey(xmlDoc *doc, xmlNode *root,
1708                                 struct client *cl, int record_no,
1709                                 struct conf_service *service, NMEM nmem,
1710                                 const char *session_mergekey)
1711 {
1712     char *mergekey_norm = 0;
1713     WRBUF norm_wr = wrbuf_alloc();
1714     const char *mergekey;
1715
1716     if (session_mergekey)
1717     {
1718         int i, num = 0;
1719         char **values = 0;
1720         nmem_strsplit_escape2(nmem, ",", session_mergekey, &values,
1721                               &num, 1, '\\', 1);
1722
1723         for (i = 0; i < num; i++)
1724             get_mergekey_from_doc(doc, root, values[i], service, norm_wr);
1725     }
1726     else if ((mergekey = yaz_xml_get_prop(root, "mergekey")))
1727     {
1728         mergekey_norm_wr(service->charsets, norm_wr, mergekey);
1729     }
1730     else
1731     {
1732         /* no mergekey defined in XSL. Look for mergekey metadata instead */
1733         int field_id;
1734         for (field_id = 0; field_id < service->num_metadata; field_id++)
1735         {
1736             struct conf_metadata *ser_md = &service->metadata[field_id];
1737             if (ser_md->mergekey != Metadata_mergekey_no)
1738             {
1739                 int r = get_mergekey_from_doc(doc, root, ser_md->name,
1740                                               service, norm_wr);
1741                 if (r == 0 && ser_md->mergekey == Metadata_mergekey_required)
1742                 {
1743                     /* no mergekey on this one and it is required..
1744                        Generate unique key instead */
1745                     wrbuf_rewind(norm_wr);
1746                     break;
1747                 }
1748             }
1749         }
1750     }
1751
1752     /* generate unique key if none is not generated already or is empty */
1753     if (wrbuf_len(norm_wr) == 0)
1754     {
1755         wrbuf_printf(norm_wr, "position: %s-%06d",
1756                      client_get_id(cl), record_no);
1757     }
1758     else
1759     {
1760         const char *lead = "content: ";
1761         wrbuf_insert(norm_wr, 0, lead, strlen(lead));
1762     }
1763     if (wrbuf_len(norm_wr) > 0)
1764         mergekey_norm = nmem_strdup(nmem, wrbuf_cstr(norm_wr));
1765     wrbuf_destroy(norm_wr);
1766     return mergekey_norm;
1767 }
1768
1769 /** \brief see if metadata for pz:recordfilter exists
1770     \param root xml root element of normalized record
1771     \param sdb session database for client
1772     \retval 0 if there is no metadata for pz:recordfilter
1773     \retval 1 if there is metadata for pz:recordfilter
1774
1775     If there is no pz:recordfilter defined, this function returns 1
1776     as well.
1777 */
1778
1779 static int check_record_filter(xmlNode *root, struct session_database *sdb)
1780 {
1781     int match = 0;
1782     xmlNode *n;
1783     const char *s;
1784     s = session_setting_oneval(sdb, PZ_RECORDFILTER);
1785
1786     if (!s || !*s)
1787         return 1;
1788
1789     for (n = root->children; n; n = n->next)
1790     {
1791         if (n->type != XML_ELEMENT_NODE)
1792             continue;
1793         if (!strcmp((const char *) n->name, "metadata"))
1794         {
1795             const char *type = yaz_xml_get_prop(n, "type");
1796             if (type)
1797             {
1798                 size_t len;
1799                 int substring;
1800                 const char *eq;
1801
1802                 if ((eq = strchr(s, '=')))
1803                     substring = 0;
1804                 else if ((eq = strchr(s, '~')))
1805                     substring = 1;
1806                 if (eq)
1807                     len = eq - s;
1808                 else
1809                     len = strlen(s);
1810                 if (len == strlen((const char *)type) &&
1811                     !memcmp((const char *) type, s, len))
1812                 {
1813                     xmlChar *value = xmlNodeGetContent(n);
1814                     if (value && *value)
1815                     {
1816                         if (!eq ||
1817                             (substring && strstr((const char *) value, eq+1)) ||
1818                             (!substring && !strcmp((const char *) value, eq + 1)))
1819                             match = 1;
1820                     }
1821                     xmlFree(value);
1822                 }
1823             }
1824         }
1825     }
1826     return match;
1827 }
1828
1829 static int ingest_to_cluster(struct client *cl,
1830                              WRBUF wrbuf_disp,
1831                              WRBUF wrbuf_norm,
1832                              xmlDoc *xdoc,
1833                              xmlNode *root,
1834                              int record_no,
1835                              struct record_metadata_attr *mergekey);
1836
1837 static int ingest_sub_record(struct client *cl, xmlDoc *xdoc, xmlNode *root,
1838                              int record_no, NMEM nmem,
1839                              struct session_database *sdb,
1840                              struct record_metadata_attr *mergekeys)
1841 {
1842     int ret = 0;
1843     struct session *se = client_get_session(cl);
1844     WRBUF wrbuf_disp, wrbuf_norm;
1845
1846     if (!check_record_filter(root, sdb))
1847     {
1848         session_log(se, YLOG_LOG,
1849                     "Filtered out record no %d from %s",
1850                     record_no, sdb->database->id);
1851         return 0;
1852     }
1853     wrbuf_disp = wrbuf_alloc();
1854     wrbuf_norm = wrbuf_alloc();
1855     session_enter(se, "ingest_sub_record");
1856     if (client_get_session(cl) == se && se->relevance)
1857         ret = ingest_to_cluster(cl, wrbuf_disp, wrbuf_norm,
1858                                 xdoc, root, record_no, mergekeys);
1859     session_leave(se, "ingest_sub_record");
1860     wrbuf_destroy(wrbuf_norm);
1861     wrbuf_destroy(wrbuf_disp);
1862     return ret;
1863 }
1864
1865 /** \brief ingest XML record
1866     \param cl client holds the result set for record
1867     \param rec record buffer (0 terminated)
1868     \param record_no record position (1, 2, ..)
1869     \param nmem working NMEM
1870     \retval 0 OK
1871     \retval -1 failure
1872     \retval -2 Filtered
1873 */
1874 int ingest_record(struct client *cl, const char *rec,
1875                   int record_no, NMEM nmem)
1876 {
1877     struct session *se = client_get_session(cl);
1878     struct session_database *sdb = client_get_database(cl);
1879     struct conf_service *service = se->service;
1880     xmlDoc *xdoc = normalize_record(se, sdb, service, rec, nmem);
1881     int r = ingest_xml_record(cl, xdoc, record_no, nmem, 0);
1882     client_store_xdoc(cl, record_no, xdoc);
1883     return r;
1884 }
1885
1886 int ingest_xml_record(struct client *cl, xmlDoc *xdoc,
1887                       int record_no, NMEM nmem, int cached_copy)
1888 {
1889     struct session *se = client_get_session(cl);
1890     struct session_database *sdb = client_get_database(cl);
1891     struct conf_service *service = se->service;
1892     xmlNode *root;
1893     int r = 0;
1894     if (!xdoc)
1895         return -1;
1896
1897     if (global_parameters.dump_records)
1898     {
1899         session_log(se, YLOG_LOG, "Normalized record from %s",
1900                     sdb->database->id);
1901         log_xml_doc(xdoc);
1902     }
1903
1904     root = xmlDocGetRootElement(xdoc);
1905
1906     if (!strcmp((const char *) root->name, "cluster"))
1907     {
1908         int no_merge_keys = 0;
1909         int no_merge_dups = 0;
1910         xmlNode *sroot;
1911         struct record_metadata_attr *mk = 0;
1912
1913         for (sroot = root->children; sroot; sroot = sroot->next)
1914             if (sroot->type == XML_ELEMENT_NODE &&
1915                 !strcmp((const char *) sroot->name, "record"))
1916             {
1917                 struct record_metadata_attr **mkp;
1918                 const char *mergekey_norm =
1919                     get_mergekey(xdoc, sroot, cl, record_no, service, nmem,
1920                                  se->mergekey);
1921                 if (!mergekey_norm)
1922                 {
1923                     r = -1;
1924                     break;
1925                 }
1926                 for (mkp = &mk; *mkp; mkp = &(*mkp)->next)
1927                     if (!strcmp((*mkp)->value, mergekey_norm))
1928                         break;
1929                 if (!*mkp)
1930                 {
1931                     *mkp = (struct record_metadata_attr*)
1932                         nmem_malloc(nmem, sizeof(**mkp));
1933                     (*mkp)->name = 0;
1934                     (*mkp)->value = nmem_strdup(nmem, mergekey_norm);
1935                     (*mkp)->next = 0;
1936                     no_merge_keys++;
1937                 }
1938                 else
1939                     no_merge_dups++;
1940             }
1941         if (no_merge_keys > 1 || no_merge_dups > 0)
1942         {
1943             yaz_log(YLOG_LOG, "Got %d mergekeys, %d dups for position %d",
1944                     no_merge_keys, no_merge_dups, record_no);
1945         }
1946         for (sroot = root->children; !r && sroot; sroot = sroot->next)
1947             if (sroot->type == XML_ELEMENT_NODE &&
1948                 !strcmp((const char *) sroot->name, "record"))
1949             {
1950                 if (!cached_copy)
1951                     insert_settings_values(sdb, xdoc, root, service);
1952                 r = ingest_sub_record(cl, xdoc, sroot, record_no, nmem, sdb,
1953                                       mk);
1954             }
1955     }
1956     else if (!strcmp((const char *) root->name, "record"))
1957     {
1958         const char *mergekey_norm =
1959             get_mergekey(xdoc, root, cl, record_no, service, nmem,
1960                          se->mergekey);
1961         if (mergekey_norm)
1962         {
1963             struct record_metadata_attr *mk = (struct record_metadata_attr*)
1964                 nmem_malloc(nmem, sizeof(*mk));
1965             mk->name = 0;
1966             mk->value = nmem_strdup(nmem, mergekey_norm);
1967             mk->next = 0;
1968
1969             if (!cached_copy)
1970                 insert_settings_values(sdb, xdoc, root, service);
1971             r = ingest_sub_record(cl, xdoc, root, record_no, nmem, sdb, mk);
1972         }
1973     }
1974     else
1975     {
1976         session_log(se, YLOG_WARN, "Bad pz root element: %s",
1977                     (const char *) root->name);
1978         r = -1;
1979     }
1980     return r;
1981 }
1982
1983
1984 //    struct conf_metadata *ser_md = &service->metadata[md_field_id];
1985 //    struct record_metadata *rec_md = record->metadata[md_field_id];
1986 static int match_metadata_local(struct conf_service *service,
1987                                 struct conf_metadata *ser_md,
1988                                 struct record_metadata *rec_md0,
1989                                 char **values, int num_v)
1990 {
1991     int i;
1992     struct record_metadata *rec_md = rec_md0;
1993     WRBUF val_wr = 0;
1994     WRBUF text_wr = wrbuf_alloc();
1995     for (i = 0; i < num_v; )
1996     {
1997         if (rec_md)
1998         {
1999             if (ser_md->type == Metadata_type_year
2000                 || ser_md->type == Metadata_type_date)
2001             {
2002                 int y = atoi(values[i]);
2003                 if (y >= rec_md->data.number.min
2004                     && y <= rec_md->data.number.max)
2005                     break;
2006             }
2007             else
2008             {
2009                 if (!val_wr)
2010                 {
2011                     val_wr = wrbuf_alloc();
2012                     mergekey_norm_wr(service->charsets, val_wr, values[i]);
2013                 }
2014                 wrbuf_rewind(text_wr);
2015                 mergekey_norm_wr(service->charsets, text_wr,
2016                                  rec_md->data.text.disp);
2017                 if (!strcmp(wrbuf_cstr(val_wr), wrbuf_cstr(text_wr)))
2018                     break;
2019             }
2020             rec_md = rec_md->next;
2021         }
2022         else
2023         {
2024             rec_md = rec_md0;
2025             wrbuf_destroy(val_wr);
2026             val_wr = 0;
2027             i++;
2028         }
2029     }
2030     wrbuf_destroy(val_wr);
2031     wrbuf_destroy(text_wr);
2032     return i < num_v ? 1 : 0;
2033 }
2034
2035 int session_check_cluster_limit(struct session *se, struct record_cluster *rec)
2036 {
2037     int i;
2038     struct conf_service *service = se->service;
2039     int ret = 1;
2040     const char *name;
2041     const char *value;
2042     NMEM nmem_tmp = nmem_create();
2043
2044     for (i = 0; (name = facet_limits_get(se->facet_limits, i, &value)); i++)
2045     {
2046         int j;
2047         for (j = 0; j < service->num_metadata; j++)
2048         {
2049             struct conf_metadata *md = service->metadata + j;
2050             if (!strcmp(md->name, name) && md->limitcluster)
2051             {
2052                 char **values = 0;
2053                 int num = 0;
2054                 int md_field_id =
2055                     conf_service_metadata_field_id(service,
2056                                                    md->limitcluster);
2057
2058                 if (md_field_id < 0)
2059                 {
2060                     ret = 0;
2061                     break;
2062                 }
2063
2064                 nmem_strsplit_escape2(nmem_tmp, "|", value, &values,
2065                                       &num, 1, '\\', 1);
2066
2067                 if (!match_metadata_local(service,
2068                                           &service->metadata[md_field_id],
2069                                           rec->metadata[md_field_id],
2070                                           values, num))
2071                 {
2072                     ret = 0;
2073                     break;
2074                 }
2075             }
2076         }
2077     }
2078     nmem_destroy(nmem_tmp);
2079     return ret;
2080 }
2081
2082 // Skip record on non-zero
2083 static int check_limit_local(struct client *cl,
2084                              struct record *record,
2085                              int record_no)
2086 {
2087     int skip_record = 0;
2088     struct session *se = client_get_session(cl);
2089     struct conf_service *service = se->service;
2090     NMEM nmem_tmp = nmem_create();
2091     struct session_database *sdb = client_get_database(cl);
2092     int l = 0;
2093     while (!skip_record)
2094     {
2095         int md_field_id;
2096         char **values = 0;
2097         int num_v = 0;
2098         const char *name =
2099             client_get_facet_limit_local(cl, sdb, &l, nmem_tmp,
2100                                          &num_v, &values);
2101         if (!name)
2102             break;
2103
2104         if (!strcmp(name, "*"))
2105         {
2106             for (md_field_id = 0; md_field_id < service->num_metadata;
2107                  md_field_id++)
2108             {
2109                 if (match_metadata_local(
2110                         service,
2111                         &service->metadata[md_field_id],
2112                         record->metadata[md_field_id],
2113                         values, num_v))
2114                     break;
2115             }
2116             if (md_field_id == service->num_metadata)
2117                 skip_record = 1;
2118         }
2119         else
2120         {
2121             md_field_id = conf_service_metadata_field_id(service, name);
2122             if (md_field_id < 0)
2123             {
2124                 skip_record = 1;
2125                 break;
2126             }
2127             if (!match_metadata_local(
2128                     service,
2129                     &service->metadata[md_field_id],
2130                     record->metadata[md_field_id],
2131                     values, num_v))
2132             {
2133                 skip_record = 1;
2134             }
2135         }
2136     }
2137     nmem_destroy(nmem_tmp);
2138     return skip_record;
2139 }
2140
2141 static int ingest_to_cluster(struct client *cl,
2142                              WRBUF wrbuf_disp,
2143                              WRBUF wrbuf_norm,
2144                              xmlDoc *xdoc,
2145                              xmlNode *root,
2146                              int record_no,
2147                              struct record_metadata_attr *merge_keys)
2148 {
2149     xmlNode *n;
2150     struct session *se = client_get_session(cl);
2151     struct conf_service *service = se->service;
2152     int term_factor = 1;
2153     struct record_cluster *cluster;
2154     struct record_metadata **metadata0;
2155     struct session_database *sdb = client_get_database(cl);
2156     NMEM ingest_nmem = 0;
2157     char **rank_values = 0;
2158     int rank_num = 0;
2159     struct record *record = record_create(se->nmem,
2160                                           service->num_metadata,
2161                                           service->num_sortkeys, cl,
2162                                           record_no);
2163
2164     for (n = root->children; n; n = n->next)
2165     {
2166         if (n->type != XML_ELEMENT_NODE)
2167             continue;
2168         if (!strcmp((const char *) n->name, "metadata"))
2169         {
2170             struct conf_metadata *ser_md = 0;
2171             struct record_metadata **wheretoput = 0;
2172             struct record_metadata *rec_md = 0;
2173             int md_field_id = -1;
2174             xmlChar *value0;
2175             const char *type = yaz_xml_get_prop(n, "type");
2176
2177             if (!type)
2178                 continue;
2179
2180             md_field_id
2181                 = conf_service_metadata_field_id(service, (const char *) type);
2182             if (md_field_id < 0)
2183             {
2184                 if (se->number_of_warnings_unknown_metadata == 0)
2185                 {
2186                     session_log(se, YLOG_WARN,
2187                             "Ignoring unknown metadata element: %s", type);
2188                 }
2189                 se->number_of_warnings_unknown_metadata++;
2190                 continue;
2191             }
2192
2193             wrbuf_rewind(wrbuf_disp);
2194             value0 = xmlNodeListGetString(xdoc, n->children, 1);
2195             if (!value0 || !*value0)
2196             {
2197                 const char *empty = yaz_xml_get_prop(n, "empty");
2198                 if (value0)
2199                     xmlFree(value0);
2200                 if (!empty)
2201                     continue;
2202                 wrbuf_puts(wrbuf_disp, (const char *) empty);
2203             }
2204             else
2205             {
2206                 wrbuf_puts(wrbuf_disp, (const char *) value0);
2207                 xmlFree(value0);
2208             }
2209             ser_md = &service->metadata[md_field_id];
2210
2211             // non-merged metadata
2212             rec_md = record_metadata_init(se->nmem, wrbuf_cstr(wrbuf_disp), 0,
2213                                           ser_md->type, n->properties);
2214             if (!rec_md)
2215             {
2216                 session_log(se, YLOG_WARN, "bad metadata data '%s' "
2217                             "for element '%s'", wrbuf_cstr(wrbuf_disp), type);
2218                 continue;
2219             }
2220
2221             if (ser_md->type == Metadata_type_generic)
2222             {
2223                 WRBUF w = wrbuf_alloc();
2224                 if (relevance_snippet(se->relevance,
2225                                       wrbuf_cstr(wrbuf_disp), ser_md->name, w))
2226                     rec_md->data.text.snippet = nmem_strdup(se->nmem,
2227                                                             wrbuf_cstr(w));
2228                 wrbuf_destroy(w);
2229             }
2230
2231
2232             wheretoput = &record->metadata[md_field_id];
2233             while (*wheretoput)
2234                 wheretoput = &(*wheretoput)->next;
2235             *wheretoput = rec_md;
2236         }
2237     }
2238
2239     if (check_limit_local(cl, record, record_no))
2240     {
2241         return -2;
2242     }
2243     cluster = reclist_insert(se->reclist, se->relevance, service, record,
2244                              merge_keys, &se->total_merged);
2245     if (!cluster)
2246     {
2247         return 0; // complete match with existing record
2248     }
2249
2250     {
2251         const char *use_term_factor_str =
2252             session_setting_oneval(sdb, PZ_TERMLIST_TERM_FACTOR);
2253         if (use_term_factor_str && use_term_factor_str[0] == '1')
2254         {
2255             int maxrecs = client_get_maxrecs(cl);
2256             int hits = (int) client_get_hits(cl);
2257             term_factor = MAX(hits, maxrecs) /  MAX(1, maxrecs);
2258             assert(term_factor >= 1);
2259             session_log(se, YLOG_DEBUG, "Using term factor: %d (%d / %d)",
2260                         term_factor, MAX(hits, maxrecs), MAX(1, maxrecs));
2261         }
2262     }
2263
2264     if (global_parameters.dump_records)
2265         session_log(se, YLOG_LOG, "Cluster id %s from %s (#%d)", cluster->recid,
2266                     sdb->database->id, record_no);
2267
2268     // original metadata, to check if first existence of a field
2269     metadata0 = xmalloc(sizeof(*metadata0) * service->num_metadata);
2270     memcpy(metadata0, cluster->metadata,
2271            sizeof(*metadata0) * service->num_metadata);
2272
2273     ingest_nmem = nmem_create();
2274     if (se->rank)
2275     {
2276         yaz_log(YLOG_LOG, "local in sort : %s", se->rank);
2277         nmem_strsplit_escape2(ingest_nmem, ",", se->rank, &rank_values,
2278                               &rank_num, 1, '\\', 1);
2279     }
2280
2281     // now parsing XML record and adding data to cluster or record metadata
2282     for (n = root->children; n; n = n->next)
2283     {
2284         if (n->type != XML_ELEMENT_NODE)
2285             continue;
2286         if (!strcmp((const char *) n->name, "metadata"))
2287         {
2288             struct conf_metadata *ser_md = 0;
2289             struct conf_sortkey *ser_sk = 0;
2290             struct record_metadata **wheretoput = 0;
2291             struct record_metadata *rec_md = 0;
2292             int md_field_id = -1;
2293             int sk_field_id = -1;
2294             const char *rank = 0;
2295             const char *xml_rank = 0;
2296             const char *type = 0;
2297             xmlChar *value0;
2298
2299             type = yaz_xml_get_prop(n, "type");
2300             if (!type)
2301                 continue;
2302
2303             md_field_id
2304                 = conf_service_metadata_field_id(service, (const char *) type);
2305             if (md_field_id < 0)
2306                 continue;
2307
2308             ser_md = &service->metadata[md_field_id];
2309
2310             if (ser_md->sortkey_offset >= 0)
2311             {
2312                 sk_field_id = ser_md->sortkey_offset;
2313                 ser_sk = &service->sortkeys[sk_field_id];
2314             }
2315
2316             wrbuf_rewind(wrbuf_disp);
2317             wrbuf_rewind(wrbuf_norm);
2318
2319             value0 = xmlNodeListGetString(xdoc, n->children, 1);
2320             if (!value0 || !*value0)
2321             {
2322                 if (value0)
2323                     xmlFree(value0);
2324                 continue;
2325             }
2326
2327             if (ser_md->icurule)
2328             {
2329                 run_icu(se, ser_md->icurule, (const char *) value0,
2330                         wrbuf_norm, wrbuf_disp);
2331                 yaz_log(YLOG_LOG, "run_icu input=%s norm=%s disp=%s",
2332                         (const char *) value0,
2333                         wrbuf_cstr(wrbuf_norm), wrbuf_cstr(wrbuf_disp));
2334                 rec_md = record_metadata_init(se->nmem, wrbuf_cstr(wrbuf_disp),
2335                                               wrbuf_cstr(wrbuf_norm),
2336                                               ser_md->type, 0);
2337             }
2338             else
2339             {
2340                 wrbuf_puts(wrbuf_disp, (const char *) value0);
2341                 rec_md = record_metadata_init(se->nmem, wrbuf_cstr(wrbuf_disp),
2342                                               0,
2343                                               ser_md->type, 0);
2344             }
2345
2346             xmlFree(value0);
2347
2348             // see if the field was not in cluster already (from beginning)
2349             if (!rec_md)
2350                 continue;
2351
2352             if (rank_num)
2353             {
2354                 int i;
2355                 for (i = 0; i < rank_num; i++)
2356                 {
2357                     const char *val = rank_values[i];
2358                     const char *cp = strchr(val, '=');
2359                     if (!cp)
2360                         continue;
2361                     if ((cp - val) == strlen((const char *) type)
2362                         && !memcmp(val, type, cp - val))
2363                     {
2364                         rank = cp + 1;
2365                         break;
2366                     }
2367                 }
2368             }
2369             else
2370             {
2371                 xml_rank = yaz_xml_get_prop(n, "rank");
2372                 rank = xml_rank ? (const char *) xml_rank : ser_md->rank;
2373             }
2374
2375             wheretoput = &cluster->metadata[md_field_id];
2376
2377             if (ser_md->merge == Metadata_merge_first)
2378             {
2379                 if (!metadata0[md_field_id])
2380                 {
2381                     while (*wheretoput)
2382                         wheretoput = &(*wheretoput)->next;
2383                     *wheretoput = rec_md;
2384                 }
2385             }
2386             else if (ser_md->merge == Metadata_merge_unique)
2387             {
2388                 while (*wheretoput)
2389                 {
2390                     if (!strcmp((const char *) (*wheretoput)->data.text.norm,
2391                                 rec_md->data.text.norm))
2392                         break;
2393                     wheretoput = &(*wheretoput)->next;
2394                 }
2395                 if (!*wheretoput)
2396                     *wheretoput = rec_md;
2397             }
2398             else if (ser_md->merge == Metadata_merge_longest)
2399             {
2400                 if (!*wheretoput
2401                     || strlen(rec_md->data.text.norm)
2402                     > strlen((*wheretoput)->data.text.norm))
2403                 {
2404                     *wheretoput = rec_md;
2405                     if (ser_sk)
2406                     {
2407                         pp2_charset_token_t prt;
2408                         const char *sort_str = 0;
2409                         int skip_article =
2410                             ser_sk->type == Metadata_type_skiparticle;
2411
2412                         if (!cluster->sortkeys[sk_field_id])
2413                             cluster->sortkeys[sk_field_id] =
2414                                 nmem_malloc(se->nmem,
2415                                             sizeof(union data_types));
2416
2417                         prt =
2418                             pp2_charset_token_create(service->charsets, "sort");
2419
2420                         pp2_charset_token_first(prt, rec_md->data.text.disp,
2421                                                 skip_article);
2422
2423                         pp2_charset_token_next(prt);
2424
2425                         sort_str = pp2_get_sort(prt);
2426
2427                         cluster->sortkeys[sk_field_id]->text.disp =
2428                             rec_md->data.text.disp;
2429                         if (!sort_str)
2430                         {
2431                             sort_str = rec_md->data.text.disp;
2432                             session_log(se, YLOG_WARN,
2433                                     "Could not make sortkey. Bug #1858");
2434                         }
2435                         cluster->sortkeys[sk_field_id]->text.sort =
2436                             nmem_strdup(se->nmem, sort_str);
2437                         pp2_charset_token_destroy(prt);
2438                     }
2439                 }
2440             }
2441             else if (ser_md->merge == Metadata_merge_all)
2442             {
2443                 while (*wheretoput)
2444                     wheretoput = &(*wheretoput)->next;
2445                 *wheretoput = rec_md;
2446             }
2447             else if (ser_md->merge == Metadata_merge_range)
2448             {
2449                 if (!*wheretoput)
2450                 {
2451                     *wheretoput = rec_md;
2452                     if (ser_sk)
2453                         cluster->sortkeys[sk_field_id]
2454                             = &rec_md->data;
2455                 }
2456                 else
2457                 {
2458                     int this_min = rec_md->data.number.min;
2459                     int this_max = rec_md->data.number.max;
2460                     if (this_min < (*wheretoput)->data.number.min)
2461                         (*wheretoput)->data.number.min = this_min;
2462                     if (this_max > (*wheretoput)->data.number.max)
2463                         (*wheretoput)->data.number.max = this_max;
2464                 }
2465             }
2466
2467             // ranking of _all_ fields enabled ...
2468             if (rank)
2469             {
2470                 relevance_countwords(se->relevance, cluster,
2471                                      wrbuf_cstr(wrbuf_disp),
2472                                      rank, ser_md->name);
2473             }
2474             // construct facets ... unless the client already has reported them
2475             if (ser_md->termlist && !client_has_facet(cl, (char *) type))
2476             {
2477                 if (ser_md->type == Metadata_type_year)
2478                 {
2479                     char year[64];
2480                     sprintf(year, "%d", rec_md->data.number.max);
2481
2482                     add_facet(se, (char *) type, year, term_factor, cl);
2483                     if (rec_md->data.number.max != rec_md->data.number.min)
2484                     {
2485                         sprintf(year, "%d", rec_md->data.number.min);
2486                         add_facet(se, (char *) type, year, term_factor, cl);
2487                     }
2488                 }
2489                 else
2490                     add_facet(se, type, wrbuf_cstr(wrbuf_disp), term_factor, cl);
2491             }
2492         }
2493         else
2494         {
2495             if (se->number_of_warnings_unknown_elements == 0)
2496                 session_log(se, YLOG_WARN,
2497                         "Unexpected element in internal record: %s", n->name);
2498             se->number_of_warnings_unknown_elements++;
2499         }
2500     }
2501     nmem_destroy(ingest_nmem);
2502     xfree(metadata0);
2503     relevance_donerecord(se->relevance, cluster);
2504     se->total_records++;
2505
2506     return 0;
2507 }
2508
2509 void session_log(struct session *s, int level, const char *fmt, ...)
2510 {
2511     char buf[1024];
2512     va_list ap;
2513     va_start(ap, fmt);
2514
2515     yaz_vsnprintf(buf, sizeof(buf)-30, fmt, ap);
2516     yaz_log(level, "Session %u: %s", s ? s->session_id : 0, buf);
2517
2518     va_end(ap);
2519 }
2520
2521 /*
2522  * Local variables:
2523  * c-basic-offset: 4
2524  * c-file-style: "Stroustrup"
2525  * indent-tabs-mode: nil
2526  * End:
2527  * vim: shiftwidth=4 tabstop=8 expandtab
2528  */
2529