src/session.c

   1 /* This file is part of Pazpar2.
   2    Copyright (C) Index Data
   3
   4 Pazpar2 is free software; you can redistribute it and/or modify it under
   5 the terms of the GNU General Public License as published by the Free
   6 Software Foundation; either version 2, or (at your option) any later
   7 version.
   8
   9 Pazpar2 is distributed in the hope that it will be useful, but WITHOUT ANY
  10 WARRANTY; without even the implied warranty of MERCHANTABILITY or
  11 FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License
  12 for more details.
  13
  14 You should have received a copy of the GNU General Public License
  15 along with this program; if not, write to the Free Software
  16 Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
  17
  18 */
  19
  20 /** \file session.c
  21     \brief high-level logic; mostly user sessions and settings
  22 */
  23
  24 #if HAVE_CONFIG_H
  25 #include <config.h>
  26 #endif
  27
  28 #include <time.h>
  29 #include <stdlib.h>
  30 #include <stdio.h>
  31 #include <string.h>
  32 #if HAVE_SYS_TIME_H
  33 #include <sys/time.h>
  34 #endif
  35 #if HAVE_UNISTD_H
  36 #include <unistd.h>
  37 #endif
  38 #ifdef WIN32
  39 #include <windows.h>
  40 #endif
  41 #include <signal.h>
  42 #include <ctype.h>
  43 #include <assert.h>
  44 #include <math.h>
  45
  46 #include <yaz/marcdisp.h>
  47 #include <yaz/comstack.h>
  48 #include <yaz/tcpip.h>
  49 #include <yaz/proto.h>
  50 #include <yaz/readconf.h>
  51 #include <yaz/pquery.h>
  52 #include <yaz/otherinfo.h>
  53 #include <yaz/yaz-util.h>
  54 #include <yaz/nmem.h>
  55 #include <yaz/query-charset.h>
  56 #include <yaz/querytowrbuf.h>
  57 #include <yaz/oid_db.h>
  58 #include <yaz/snprintf.h>
  59 #include <yaz/xml_get.h>
  60
  61 #define USE_TIMING 0
  62 #if USE_TIMING
  63 #include <yaz/timing.h>
  64 #endif
  65
  66 #include "ppmutex.h"
  67 #include "parameters.h"
  68 #include "session.h"
  69 #include "eventl.h"
  70 #include "http.h"
  71 #include "termlists.h"
  72 #include "reclists.h"
  73 #include "relevance.h"
  74 #include "database.h"
  75 #include "client.h"
  76 #include "settings.h"
  77 #include "normalize7bit.h"
  78
  79 #include <libxml/tree.h>
  80
  81 #define MAX_CHUNK 15
  82
  83 #define MAX(a,b) ((a)>(b)?(a):(b))
  84
  85 // Note: Some things in this structure will eventually move to configuration
  86 struct parameters global_parameters =
  87 {
  88     0,   // dump_records
  89     0,   // debug_mode
  90     0,   // predictable sessions
  91 };
  92
  93 struct client_list {
  94     struct client *client;
  95     struct client_list *next;
  96 };
  97
  98 /* session counting (1) , disable client counting (0) */
  99 static YAZ_MUTEX g_session_mutex = 0;
 100 static int no_sessions = 0;
 101
 102 static int session_use(int delta)
 103 {
 104     int sessions;
 105     if (!g_session_mutex)
 106         yaz_mutex_create(&g_session_mutex);
 107     yaz_mutex_enter(g_session_mutex);
 108     no_sessions += delta;
 109     sessions = no_sessions;
 110     yaz_mutex_leave(g_session_mutex);
 111     yaz_log(YLOG_DEBUG, "%s sessions=%d", delta == 0 ? "" :
 112             (delta > 0 ? "INC" : "DEC"), no_sessions);
 113     return sessions;
 114 }
 115
 116 int sessions_count(void)
 117 {
 118     return session_use(0);
 119 }
 120
 121 static void log_xml_doc(xmlDoc *doc)
 122 {
 123     FILE *lf = yaz_log_file();
 124     xmlChar *result = 0;
 125     int len = 0;
 126 #if LIBXML_VERSION >= 20600
 127     xmlDocDumpFormatMemory(doc, &result, &len, 1);
 128 #else
 129     xmlDocDumpMemory(doc, &result, &len);
 130 #endif
 131     if (lf && len)
 132     {
 133         (void) fwrite(result, 1, len, lf);
 134         fprintf(lf, "\n");
 135     }
 136     xmlFree(result);
 137 }
 138
 139 static void session_enter(struct session *s, const char *caller)
 140 {
 141     if (caller)
 142         session_log(s, YLOG_DEBUG, "Session lock by %s", caller);
 143     yaz_mutex_enter(s->session_mutex);
 144 }
 145
 146 static void session_leave(struct session *s, const char *caller)
 147 {
 148     yaz_mutex_leave(s->session_mutex);
 149     if (caller)
 150         session_log(s, YLOG_DEBUG, "Session unlock by %s", caller);
 151 }
 152
 153 static int run_icu(struct session *s, const char *icu_chain_id,
 154                    const char *value,
 155                    WRBUF norm_wr, WRBUF disp_wr)
 156 {
 157     const char *facet_component;
 158     struct conf_service *service = s->service;
 159     pp2_charset_token_t prt =
 160         pp2_charset_token_create(service->charsets, icu_chain_id);
 161     if (!prt)
 162     {
 163         session_log(s, YLOG_FATAL,
 164                     "Unknown ICU chain '%s'", icu_chain_id);
 165         return 0;
 166     }
 167     pp2_charset_token_first(prt, value, 0);
 168     while ((facet_component = pp2_charset_token_next(prt)))
 169     {
 170         const char *display_component;
 171         if (*facet_component)
 172         {
 173             if (wrbuf_len(norm_wr))
 174                 wrbuf_puts(norm_wr, " ");
 175             wrbuf_puts(norm_wr, facet_component);
 176         }
 177         display_component = pp2_get_display(prt);
 178         if (display_component)
 179         {
 180             if (wrbuf_len(disp_wr))
 181                 wrbuf_puts(disp_wr, " ");
 182             wrbuf_puts(disp_wr, display_component);
 183         }
 184     }
 185     pp2_charset_token_destroy(prt);
 186     return 1;
 187 }
 188
 189 static void session_normalize_facet(struct session *s,
 190                                     const char *type, const char *value,
 191                                     WRBUF display_wrbuf, WRBUF facet_wrbuf)
 192 {
 193     struct conf_service *service = s->service;
 194     int i;
 195     const char *icu_chain_id = 0;
 196
 197     for (i = 0; i < service->num_metadata; i++)
 198         if (!strcmp((service->metadata + i)->name, type))
 199             icu_chain_id = (service->metadata + i)->facetrule;
 200     if (!icu_chain_id)
 201         icu_chain_id = "facet";
 202
 203     run_icu(s, icu_chain_id, value, facet_wrbuf, display_wrbuf);
 204 }
 205
 206 struct facet_id {
 207     char *client_id;
 208     char *type;
 209     char *id;
 210     char *term;
 211     struct facet_id *next;
 212 };
 213
 214 static void session_add_id_facet(struct session *s, struct client *cl,
 215                                  const char *type,
 216                                  const char *id,
 217                                  size_t id_len,
 218                                  const char *term)
 219 {
 220     struct facet_id *t = nmem_malloc(s->session_nmem, sizeof(*t));
 221
 222     t->client_id = nmem_strdup(s->session_nmem, client_get_id(cl));
 223     t->type = nmem_strdup(s->session_nmem, type);
 224     t->id = nmem_strdupn(s->session_nmem, id, id_len);
 225     t->term = nmem_strdup(s->session_nmem, term);
 226     t->next = s->facet_id_list;
 227     s->facet_id_list = t;
 228 }
 229
 230
 231 const char *session_lookup_id_facet(struct session *s, struct client *cl,
 232                                     const char *type,
 233                                     const char *term)
 234 {
 235     struct facet_id *t = s->facet_id_list;
 236     for (; t; t = t->next)
 237         if (!strcmp(client_get_id(cl), t->client_id) &&
 238             !strcmp(t->type, type) && !strcmp(t->term, term))
 239         {
 240             return t->id;
 241         }
 242     return 0;
 243 }
 244
 245 void add_facet(struct session *s, const char *type, const char *value, int count, struct client *cl)
 246 {
 247     WRBUF facet_wrbuf = wrbuf_alloc();
 248     WRBUF display_wrbuf = wrbuf_alloc();
 249     const char *id = 0;
 250     size_t id_len = 0;
 251
 252     /* inspect pz:facetmap:split:name ?? */
 253     if (!strncmp(type, "split:", 6))
 254     {
 255         const char *cp = strchr(value, ':');
 256         if (cp)
 257         {
 258             id = value;
 259             id_len = cp - value;
 260             value = cp + 1;
 261         }
 262         type += 6;
 263     }
 264
 265     session_normalize_facet(s, type, value, display_wrbuf, facet_wrbuf);
 266     if (wrbuf_len(facet_wrbuf))
 267     {
 268         struct named_termlist **tp = &s->termlists;
 269         for (; (*tp); tp = &(*tp)->next)
 270             if (!strcmp((*tp)->name, type))
 271                 break;
 272         if (!*tp)
 273         {
 274             *tp = nmem_malloc(s->nmem, sizeof(**tp));
 275             (*tp)->name = nmem_strdup(s->nmem, type);
 276             (*tp)->termlist = termlist_create(s->nmem);
 277             (*tp)->next = 0;
 278         }
 279         termlist_insert((*tp)->termlist, wrbuf_cstr(display_wrbuf),
 280                         wrbuf_cstr(facet_wrbuf), id, id_len, count);
 281         if (id)
 282             session_add_id_facet(s, cl, type, id, id_len,
 283                                  wrbuf_cstr(display_wrbuf));
 284     }
 285     wrbuf_destroy(facet_wrbuf);
 286     wrbuf_destroy(display_wrbuf);
 287 }
 288
 289 static xmlDoc *record_to_xml(struct session *se,
 290                              struct session_database *sdb, const char *rec)
 291 {
 292     struct database *db = sdb->database;
 293     xmlDoc *rdoc = 0;
 294
 295     rdoc = xmlParseMemory(rec, strlen(rec));
 296
 297     if (!rdoc)
 298     {
 299         session_log(se, YLOG_WARN, "Non-wellformed XML");
 300         return 0;
 301     }
 302
 303     if (global_parameters.dump_records)
 304     {
 305         session_log(se, YLOG_LOG, "Un-normalized record from %s", db->id);
 306         log_xml_doc(rdoc);
 307     }
 308
 309     return rdoc;
 310 }
 311
 312 #define MAX_XSLT_ARGS 16
 313
 314 // Add static values from session database settings if applicable
 315 static void insert_settings_parameters(struct session_database *sdb,
 316                                        struct conf_service *service,
 317                                        char **parms,
 318                                        NMEM nmem)
 319 {
 320     int i;
 321     int nparms = 0;
 322     int offset = 0;
 323
 324     for (i = 0; i < service->num_metadata; i++)
 325     {
 326         struct conf_metadata *md = &service->metadata[i];
 327         int setting;
 328
 329         if (md->setting == Metadata_setting_parameter &&
 330             (setting = settings_lookup_offset(service, md->name)) >= 0)
 331         {
 332             const char *val = session_setting_oneval(sdb, setting);
 333             if (val && nparms < MAX_XSLT_ARGS)
 334             {
 335                 char *buf;
 336                 int len = strlen(val);
 337                 buf = nmem_malloc(nmem, len + 3);
 338                 buf[0] = '\'';
 339                 strcpy(buf + 1, val);
 340                 buf[len+1] = '\'';
 341                 buf[len+2] = '\0';
 342                 parms[offset++] = md->name;
 343                 parms[offset++] = buf;
 344                 nparms++;
 345             }
 346         }
 347     }
 348     parms[offset] = 0;
 349 }
 350
 351 // Add static values from session database settings if applicable
 352 static void insert_settings_values(struct session_database *sdb, xmlDoc *doc,
 353                                    xmlNode *root,
 354                                    struct conf_service *service)
 355 {
 356     int i;
 357
 358     for (i = 0; i < service->num_metadata; i++)
 359     {
 360         struct conf_metadata *md = &service->metadata[i];
 361         int offset;
 362
 363         if (md->setting == Metadata_setting_postproc &&
 364             (offset = settings_lookup_offset(service, md->name)) >= 0)
 365         {
 366             const char *val = session_setting_oneval(sdb, offset);
 367             if (val)
 368             {
 369                 xmlNode *n = xmlNewTextChild(root, 0, (xmlChar *) "metadata",
 370                                              (xmlChar *) val);
 371                 xmlSetProp(n, (xmlChar *) "type", (xmlChar *) md->name);
 372             }
 373         }
 374     }
 375 }
 376
 377 static xmlDoc *normalize_record(struct session *se,
 378                                 struct session_database *sdb,
 379                                 struct conf_service *service,
 380                                 const char *rec, NMEM nmem)
 381 {
 382     xmlDoc *rdoc = record_to_xml(se, sdb, rec);
 383
 384     if (rdoc)
 385     {
 386         char *parms[MAX_XSLT_ARGS*2+1];
 387
 388         insert_settings_parameters(sdb, service, parms, nmem);
 389
 390         if (normalize_record_transform(sdb->map, &rdoc, (const char **)parms))
 391         {
 392             session_log(se, YLOG_WARN, "Normalize failed");
 393         }
 394     }
 395     return rdoc;
 396 }
 397
 398 void session_settings_dump(struct session *se,
 399                            struct session_database *db,
 400                            WRBUF w)
 401 {
 402     if (db->settings)
 403     {
 404         int i, num = db->num_settings;
 405         for (i = 0; i < num; i++)
 406         {
 407             struct setting *s = db->settings[i];
 408             for (;s ; s = s->next)
 409             {
 410                 wrbuf_puts(w, "<set name=\"");
 411                 wrbuf_xmlputs(w, s->name);
 412                 wrbuf_puts(w, "\" value=\"");
 413                 wrbuf_xmlputs(w, s->value);
 414                 wrbuf_puts(w, "\"/>");
 415             }
 416             if (db->settings[i])
 417                 wrbuf_puts(w, "\n");
 418         }
 419     }
 420 }
 421
 422 // Retrieve first defined value for 'name' for given database.
 423 // Will be extended to take into account user associated with session
 424 const char *session_setting_oneval(struct session_database *db, int offset)
 425 {
 426     if (offset >= db->num_settings || !db->settings[offset])
 427         return "";
 428     return db->settings[offset]->value;
 429 }
 430
 431 // Prepare XSLT stylesheets for record normalization
 432 // Structures are allocated on the session_wide nmem to avoid having
 433 // to recompute this for every search. This would lead
 434 // to leaking if a single session was to repeatedly change the PZ_XSLT
 435 // setting. However, this is not a realistic use scenario.
 436 static int prepare_map(struct session *se, struct session_database *sdb)
 437 {
 438     if (sdb->settings && !sdb->map)
 439     {
 440         const char *s;
 441
 442         if (sdb->settings[PZ_XSLT] &&
 443             (s = session_setting_oneval(sdb, PZ_XSLT)))
 444         {
 445             char auto_stylesheet[256];
 446
 447             if (!strcmp(s, "auto"))
 448             {
 449                 const char *request_syntax = session_setting_oneval(
 450                     sdb, PZ_REQUESTSYNTAX);
 451                 if (request_syntax)
 452                 {
 453                     char *cp;
 454                     yaz_snprintf(auto_stylesheet, sizeof(auto_stylesheet),
 455                                  "%s.xsl", request_syntax);
 456                     for (cp = auto_stylesheet; *cp; cp++)
 457                     {
 458                         /* deliberately only consider ASCII */
 459                         if (*cp > 32 && *cp < 127)
 460                             *cp = tolower(*cp);
 461                     }
 462                     s = auto_stylesheet;
 463                 }
 464                 else
 465                 {
 466                     session_log(se, YLOG_WARN,
 467                                 "No pz:requestsyntax for auto stylesheet");
 468                 }
 469             }
 470             sdb->map = normalize_cache_get(se->normalize_cache,
 471                                            se->service, s);
 472             if (!sdb->map)
 473                 return -1;
 474         }
 475     }
 476     return 0;
 477 }
 478
 479 // called if watch should be removed because http_channel is to be destroyed
 480 static void session_watch_cancel(void *data, struct http_channel *c,
 481                                  void *data2)
 482 {
 483     struct session_watchentry *ent = data;
 484
 485     ent->fun = 0;
 486     ent->data = 0;
 487     ent->obs = 0;
 488 }
 489
 490 // set watch. Returns 0=OK, -1 if watch is already set
 491 int session_set_watch(struct session *s, int what,
 492                       session_watchfun fun, void *data,
 493                       struct http_channel *chan)
 494 {
 495     int ret;
 496     session_enter(s, "session_set_watch");
 497     if (s->watchlist[what].fun)
 498         ret = -1;
 499     else
 500     {
 501
 502         s->watchlist[what].fun = fun;
 503         s->watchlist[what].data = data;
 504         s->watchlist[what].obs = http_add_observer(chan, &s->watchlist[what],
 505                                                    session_watch_cancel);
 506         ret = 0;
 507     }
 508     session_leave(s, "session_set_watch");
 509     return ret;
 510 }
 511
 512 void session_alert_watch(struct session *s, int what)
 513 {
 514     assert(s);
 515     session_enter(s, "session_alert_watch");
 516     if (s->watchlist[what].fun)
 517     {
 518         /* our watch is no longer associated with http_channel */
 519         void *data;
 520         session_watchfun fun;
 521
 522         http_remove_observer(s->watchlist[what].obs);
 523         fun  = s->watchlist[what].fun;
 524         data = s->watchlist[what].data;
 525
 526         /* reset watch before fun is invoked - in case fun wants to set
 527            it again */
 528         s->watchlist[what].fun = 0;
 529         s->watchlist[what].data = 0;
 530         s->watchlist[what].obs = 0;
 531
 532         session_leave(s, "session_alert_watch");
 533         session_log(s, YLOG_DEBUG,
 534                     "Alert Watch: %d calling function: %p", what, fun);
 535         fun(data);
 536     }
 537     else
 538         session_leave(s,"session_alert_watch");
 539 }
 540
 541 //callback for grep_databases
 542 static void select_targets_callback(struct session *se,
 543                                     struct session_database *db)
 544 {
 545     struct client *cl;
 546     struct client_list *l;
 547
 548     for (l = se->clients_cached; l; l = l->next)
 549         if (client_get_database(l->client) == db)
 550             break;
 551
 552     if (l)
 553         cl = l->client;
 554     else
 555     {
 556         cl = client_create(db->database->id);
 557         client_set_database(cl, db);
 558
 559         l = xmalloc(sizeof(*l));
 560         l->client = cl;
 561         l->next = se->clients_cached;
 562         se->clients_cached = l;
 563     }
 564     /* set session always. If may be 0 if client is not active */
 565     client_set_session(cl, se);
 566
 567     l = xmalloc(sizeof(*l));
 568     l->client = cl;
 569     l->next = se->clients_active;
 570     se->clients_active = l;
 571 }
 572
 573 static void session_reset_active_clients(struct session *se,
 574                                          struct client_list *new_list)
 575 {
 576     struct client_list *l;
 577
 578     session_enter(se, "session_reset_active_clients");
 579     l = se->clients_active;
 580     se->clients_active = new_list;
 581     session_leave(se, "session_reset_active_clients");
 582
 583     while (l)
 584     {
 585         struct client_list *l_next = l->next;
 586
 587         client_lock(l->client);
 588         client_set_session(l->client, 0); /* mark client inactive */
 589         client_unlock(l->client);
 590
 591         xfree(l);
 592         l = l_next;
 593     }
 594 }
 595
 596 static void session_remove_cached_clients(struct session *se)
 597 {
 598     struct client_list *l;
 599
 600     session_reset_active_clients(se, 0);
 601
 602     session_enter(se, "session_remove_cached_clients");
 603     l = se->clients_cached;
 604     se->clients_cached = 0;
 605     session_leave(se, "session_remove_cached_clients");
 606
 607     while (l)
 608     {
 609         struct client_list *l_next = l->next;
 610         client_lock(l->client);
 611         client_set_session(l->client, 0);
 612         client_set_database(l->client, 0);
 613         client_unlock(l->client);
 614         client_destroy(l->client);
 615         xfree(l);
 616         l = l_next;
 617     }
 618 }
 619
 620 // Associates a set of clients with a session;
 621 // Note: Session-databases represent databases with per-session
 622 // setting overrides
 623 static int select_targets(struct session *se, const char *filter)
 624 {
 625     return session_grep_databases(se, filter, select_targets_callback);
 626 }
 627
 628 int session_active_clients(struct session *s)
 629 {
 630     struct client_list *l;
 631     int res = 0;
 632
 633     for (l = s->clients_active; l; l = l->next)
 634         if (client_is_active(l->client))
 635             res++;
 636
 637     return res;
 638 }
 639
 640 int session_is_preferred_clients_ready(struct session *s)
 641 {
 642     struct client_list *l;
 643     int res = 0;
 644
 645     for (l = s->clients_active; l; l = l->next)
 646         if (client_is_active_preferred(l->client))
 647             res++;
 648     session_log(s, YLOG_DEBUG, "Has %d active preferred clients.", res);
 649     return res == 0;
 650 }
 651
 652 static void session_clear_set(struct session *se, struct reclist_sortparms *sp)
 653 {
 654     reclist_destroy(se->reclist);
 655     if (nmem_total(se->nmem))
 656         session_log(se, YLOG_DEBUG, "NMEN operation usage %zd",
 657                     nmem_total(se->nmem));
 658     nmem_reset(se->nmem);
 659     se->total_records = se->total_merged = 0;
 660     se->termlists = 0;
 661     relevance_clear(se->relevance);
 662
 663     /* reset list of sorted results and clear to relevance search */
 664     se->sorted_results = nmem_malloc(se->nmem, sizeof(*se->sorted_results));
 665     se->sorted_results->name = nmem_strdup(se->nmem, sp->name);
 666     se->sorted_results->increasing = sp->increasing;
 667     se->sorted_results->type = sp->type;
 668     se->sorted_results->next = 0;
 669
 670     session_log(se, YLOG_DEBUG, "clear_set session_sort: field=%s increasing=%d type=%d configured",
 671             sp->name, sp->increasing, sp->type);
 672
 673     se->reclist = reclist_create(se->nmem);
 674 }
 675
 676 void session_sort(struct session *se, struct reclist_sortparms *sp,
 677                   const char *mergekey, const char *rank)
 678 {
 679     struct client_list *l;
 680     const char *field = sp->name;
 681     int increasing = sp->increasing;
 682     int type  = sp->type;
 683     int clients_research = 0;
 684
 685     session_enter(se, "session_sort");
 686     session_log(se, YLOG_DEBUG, "session_sort field=%s increasing=%d type=%d",
 687                 field, increasing, type);
 688
 689     if (rank && (!se->rank || strcmp(se->rank, rank)))
 690     {
 691         /* new rank must research/reingest anyway */
 692         assert(rank);
 693         xfree(se->rank);
 694         se->rank = *rank ? xstrdup(rank) : 0;
 695         clients_research = 1;
 696         session_log(se, YLOG_DEBUG, "session_sort: new rank = %s",
 697                     rank);
 698     }
 699     if (mergekey && (!se->mergekey || strcmp(se->mergekey, mergekey)))
 700     {
 701         /* new mergekey must research/reingest anyway */
 702         assert(mergekey);
 703         xfree(se->mergekey);
 704         se->mergekey = *mergekey ? xstrdup(mergekey) : 0;
 705         clients_research = 1;
 706         session_log(se, YLOG_DEBUG, "session_sort: new mergekey = %s",
 707                     mergekey);
 708     }
 709     if (clients_research == 0)
 710     {
 711         struct reclist_sortparms *sr;
 712         for (sr = se->sorted_results; sr; sr = sr->next)
 713             if (!reclist_sortparms_cmp(sr, sp))
 714                 break;
 715         if (sr)
 716         {
 717             session_log(se, YLOG_LOG, "session_sort: field=%s increasing=%d type=%d already fetched",
 718                         field, increasing, type);
 719             session_leave(se, "session_sort");
 720             return;
 721         }
 722     }
 723     session_log(se, YLOG_DEBUG, "session_sort: field=%s increasing=%d type=%d must fetch",
 724                 field, increasing, type);
 725
 726     // We need to reset reclist on every sort that changes the records, not just for position
 727     // So if just one client requires new searching, we need to clear set.
 728     // Ask each of the client if sorting requires re-search due to native sort
 729     // If it does it will require us to
 730     for (l = se->clients_active; l; l = l->next)
 731     {
 732         struct client *cl = l->client;
 733         // Assume no re-search is required.
 734         client_parse_init(cl, 1);
 735         clients_research += client_parse_sort(cl, sp, 0);
 736     }
 737     if (!clients_research || se->clients_starting)
 738     {
 739         // A new sorting based on same record set
 740         struct reclist_sortparms *sr = nmem_malloc(se->nmem, sizeof(*sr));
 741         sr->name = nmem_strdup(se->nmem, field);
 742         sr->increasing = increasing;
 743         sr->type = type;
 744         sr->next = se->sorted_results;
 745         se->sorted_results = sr;
 746         session_log(se, YLOG_DEBUG, "session_sort: no research/ingesting done");
 747         session_leave(se, "session_sort");
 748     }
 749     else
 750     {
 751         se->clients_starting = 1;
 752         session_log(se, YLOG_DEBUG,
 753                     "session_sort: reset results due to %d clients researching",
 754                     clients_research);
 755         session_clear_set(se, sp);
 756         session_log(se, YLOG_DEBUG, "Re- search/ingesting for clients due to change in sort order");
 757
 758         session_leave(se, "session_sort");
 759         for (l = se->clients_active; l; l = l->next)
 760         {
 761             struct client *cl = l->client;
 762             if (client_get_state(cl) == Client_Connecting ||
 763                 client_get_state(cl) == Client_Idle ||
 764                 client_get_state(cl) == Client_Working) {
 765                 client_start_search(cl);
 766             }
 767             else
 768             {
 769                 session_log(se, YLOG_DEBUG,
 770                             "session_sort: %s: No re-start/ingest in show. "
 771                             "Wrong client state: %d",
 772                             client_get_id(cl), client_get_state(cl));
 773             }
 774         }
 775         session_enter(se, "session_sort");
 776         se->clients_starting = 0;
 777         se->force_position = 0;
 778         session_leave(se, "session_sort");
 779     }
 780 }
 781
 782 void session_stop(struct session *se)
 783 {
 784     struct client_list *l;
 785     session_enter(se, "session_stop1");
 786     if (se->clients_starting)
 787     {
 788         session_leave(se, "session_stop1");
 789         return;
 790     }
 791     se->clients_starting = 1;
 792     session_leave(se, "session_stop1");
 793
 794     session_alert_watch(se, SESSION_WATCH_SHOW);
 795     session_alert_watch(se, SESSION_WATCH_BYTARGET);
 796     session_alert_watch(se, SESSION_WATCH_TERMLIST);
 797     session_alert_watch(se, SESSION_WATCH_SHOW_PREF);
 798
 799     for (l = se->clients_active; l; l = l->next)
 800     {
 801         struct client *cl = l->client;
 802         client_stop(cl);
 803     }
 804     session_enter(se, "session_stop2");
 805     se->clients_starting = 0;
 806     session_leave(se, "session_stop2");
 807 }
 808
 809 enum pazpar2_error_code session_search(struct session *se,
 810                                        const char *query,
 811                                        const char *startrecs,
 812                                        const char *maxrecs,
 813                                        const char *filter,
 814                                        const char *limit,
 815                                        const char **addinfo,
 816                                        const char **addinfo2,
 817                                        struct reclist_sortparms *sp,
 818                                        const char *mergekey,
 819                                        const char *rank)
 820 {
 821     int live_channels = 0;
 822     int no_working = 0;
 823     int no_failed_query = 0;
 824     int no_failed_limit = 0;
 825     int no_sortmap = 0;
 826     struct client_list *l;
 827
 828     session_log(se, YLOG_DEBUG, "Search");
 829
 830     *addinfo = 0;
 831
 832     session_enter(se, "session_search0");
 833     if (se->clients_starting)
 834     {
 835         session_leave(se, "session_search0");
 836         return PAZPAR2_NO_ERROR;
 837     }
 838     se->clients_starting = 1;
 839     se->force_position = 0;
 840     session_leave(se, "session_search0");
 841
 842     if (se->settings_modified) {
 843         session_remove_cached_clients(se);
 844     }
 845     else
 846         session_reset_active_clients(se, 0);
 847
 848     session_enter(se, "session_search");
 849     se->settings_modified = 0;
 850
 851     if (mergekey)
 852     {
 853         xfree(se->mergekey);
 854         se->mergekey = *mergekey ? xstrdup(mergekey) : 0;
 855     }
 856     if (rank)
 857     {
 858         xfree(se->rank);
 859         se->rank = *rank ? xstrdup(rank) : 0;
 860     }
 861
 862     session_clear_set(se, sp);
 863     relevance_destroy(&se->relevance);
 864
 865     live_channels = select_targets(se, filter);
 866     if (!live_channels)
 867     {
 868         session_leave(se, "session_search");
 869         se->clients_starting = 0;
 870         return PAZPAR2_NO_TARGETS;
 871     }
 872
 873     facet_limits_destroy(se->facet_limits);
 874     se->facet_limits = facet_limits_create(limit);
 875     if (!se->facet_limits)
 876     {
 877         *addinfo = "limit";
 878         session_leave(se, "session_search");
 879         se->clients_starting = 0;
 880         return PAZPAR2_MALFORMED_PARAMETER_VALUE;
 881     }
 882
 883     session_leave(se, "session_search");
 884
 885     session_alert_watch(se, SESSION_WATCH_SHOW);
 886     session_alert_watch(se, SESSION_WATCH_BYTARGET);
 887     session_alert_watch(se, SESSION_WATCH_TERMLIST);
 888     session_alert_watch(se, SESSION_WATCH_SHOW_PREF);
 889
 890     for (l = se->clients_active; l; l = l->next)
 891     {
 892         int parse_ret;
 893         struct client *cl = l->client;
 894         client_parse_init(cl, 1);
 895         if (prepare_map(se, client_get_database(cl)) < 0)
 896             continue;
 897
 898         parse_ret = client_parse_query(cl, query, se->facet_limits, addinfo2);
 899         if (parse_ret == -1)
 900             no_failed_query++;
 901         else if (parse_ret == -2)
 902             no_failed_limit++;
 903         else if (parse_ret < 0)
 904             no_working++; /* other error, such as bad CCL map */
 905         else
 906         {
 907             client_parse_range(cl, startrecs, maxrecs);
 908             client_parse_sort(cl, sp, &no_sortmap);
 909             client_start_search(cl);
 910             no_working++;
 911         }
 912     }
 913     yaz_log(YLOG_LOG, "session_search: no_working=%d no_sortmap=%d",
 914             no_working, no_sortmap);
 915     session_enter(se, "session_search2");
 916     if (no_working == 1 && no_sortmap == 1)
 917     {
 918         se->force_position = 1;
 919         yaz_log(YLOG_LOG, "force_position=1");
 920     }
 921     se->clients_starting = 0;
 922     session_leave(se, "session_search2");
 923     if (no_working == 0)
 924     {
 925         if (no_failed_query > 0)
 926         {
 927             *addinfo = "query";
 928             return PAZPAR2_MALFORMED_PARAMETER_VALUE;
 929         }
 930         else if (no_failed_limit > 0)
 931         {
 932             *addinfo = "limit";
 933             return PAZPAR2_MALFORMED_PARAMETER_VALUE;
 934         }
 935         else
 936             return PAZPAR2_NO_TARGETS;
 937     }
 938     return PAZPAR2_NO_ERROR;
 939 }
 940
 941 // Creates a new session_database object for a database
 942 static void session_init_databases_fun(void *context, struct database *db)
 943 {
 944     struct session *se = (struct session *) context;
 945     struct session_database *new = nmem_malloc(se->session_nmem, sizeof(*new));
 946     int i;
 947
 948     new->database = db;
 949
 950     new->map = 0;
 951     assert(db->settings);
 952     new->settings = nmem_malloc(se->session_nmem,
 953                                 sizeof(struct settings *) * db->num_settings);
 954     new->num_settings = db->num_settings;
 955     for (i = 0; i < db->num_settings; i++)
 956     {
 957         struct setting *setting = db->settings[i];
 958         new->settings[i] = setting;
 959     }
 960     new->next = se->databases;
 961     se->databases = new;
 962 }
 963
 964 // Doesn't free memory associated with sdb -- nmem takes care of that
 965 static void session_database_destroy(struct session_database *sdb)
 966 {
 967     sdb->map = 0;
 968 }
 969
 970 // Initialize session_database list -- this represents this session's view
 971 // of the database list -- subject to modification by the settings ws command
 972 void session_init_databases(struct session *se)
 973 {
 974     se->databases = 0;
 975     predef_grep_databases(se, se->service, session_init_databases_fun);
 976 }
 977
 978 // Probably session_init_databases_fun should be refactored instead of
 979 // called here.
 980 static struct session_database *load_session_database(struct session *se,
 981                                                       const char *id)
 982 {
 983     struct database *db = new_database_inherit_settings(id, se->session_nmem, se->service->settings);
 984     session_init_databases_fun((void*) se, db);
 985
 986     // New sdb is head of se->databases list
 987     return se->databases;
 988 }
 989
 990 // Find an existing session database. If not found, load it
 991 static struct session_database *find_session_database(struct session *se,
 992                                                       const char *id)
 993 {
 994     struct session_database *sdb;
 995
 996     for (sdb = se->databases; sdb; sdb = sdb->next)
 997         if (!strcmp(sdb->database->id, id))
 998             return sdb;
 999     return load_session_database(se, id);
1000 }
1001
1002 // Apply a session override to a database
1003 void session_apply_setting(struct session *se, const char *dbname,
1004                            const char *name, const char *value)
1005 {
1006     session_enter(se, "session_apply_setting");
1007     {
1008         struct session_database *sdb = find_session_database(se, dbname);
1009         struct conf_service *service = se->service;
1010         struct setting *s;
1011         int offset = settings_create_offset(service, name);
1012
1013         expand_settings_array(&sdb->settings, &sdb->num_settings, offset,
1014                               se->session_nmem);
1015         // Force later recompute of settings-driven data structures
1016         // (happens when a search starts and client connections are prepared)
1017         if (offset == PZ_XSLT)
1018             sdb->map = 0;
1019         se->settings_modified = 1;
1020         for (s = sdb->settings[offset]; s; s = s->next)
1021             if (!strcmp(s->name, name) &&
1022                 dbname && s->target && !strcmp(dbname, s->target))
1023                 break;
1024         if (!s)
1025         {
1026             s = nmem_malloc(se->session_nmem, sizeof(*s));
1027             s->precedence = 0;
1028             s->target = nmem_strdup(se->session_nmem, dbname);
1029             s->name = nmem_strdup(se->session_nmem, name);
1030             s->next = sdb->settings[offset];
1031             sdb->settings[offset] = s;
1032         }
1033         s->value = nmem_strdup(se->session_nmem, value);
1034     }
1035     session_leave(se, "session_apply_setting");
1036 }
1037
1038 void session_destroy(struct session *se)
1039 {
1040     struct session_database *sdb;
1041     session_log(se, YLOG_LOG, "destroy");
1042     session_use(-1);
1043     session_remove_cached_clients(se);
1044
1045     for (sdb = se->databases; sdb; sdb = sdb->next)
1046         session_database_destroy(sdb);
1047     normalize_cache_destroy(se->normalize_cache);
1048     relevance_destroy(&se->relevance);
1049     reclist_destroy(se->reclist);
1050     xfree(se->mergekey);
1051     xfree(se->rank);
1052     if (nmem_total(se->nmem))
1053         session_log(se, YLOG_DEBUG, "NMEN operation usage %zd", nmem_total(se->nmem));
1054     if (nmem_total(se->session_nmem))
1055         session_log(se, YLOG_DEBUG, "NMEN session usage %zd", nmem_total(se->session_nmem));
1056     facet_limits_destroy(se->facet_limits);
1057     nmem_destroy(se->nmem);
1058     service_destroy(se->service);
1059     yaz_mutex_destroy(&se->session_mutex);
1060 }
1061
1062 size_t session_get_memory_status(struct session *session) {
1063     size_t session_nmem;
1064     if (session == 0)
1065         return 0;
1066     session_enter(session, "session_get_memory_status");
1067     session_nmem = nmem_total(session->nmem);
1068     session_leave(session, "session_get_memory_status");
1069     return session_nmem;
1070 }
1071
1072
1073 struct session *new_session(NMEM nmem, struct conf_service *service,
1074                             unsigned session_id)
1075 {
1076     int i;
1077     struct session *session = nmem_malloc(nmem, sizeof(*session));
1078
1079     char tmp_str[50];
1080
1081     sprintf(tmp_str, "session#%u", session_id);
1082
1083     session->session_id = session_id;
1084     session_log(session, YLOG_DEBUG, "New");
1085     session->service = service;
1086     session->relevance = 0;
1087     session->total_records = 0;
1088     session->number_of_warnings_unknown_elements = 0;
1089     session->number_of_warnings_unknown_metadata = 0;
1090     session->termlists = 0;
1091     session->reclist = reclist_create(nmem);
1092     session->clients_active = 0;
1093     session->clients_cached = 0;
1094     session->settings_modified = 0;
1095     session->session_nmem = nmem;
1096     session->facet_id_list = 0;
1097     session->nmem = nmem_create();
1098     session->databases = 0;
1099     session->sorted_results = 0;
1100     session->facet_limits = 0;
1101     session->mergekey = 0;
1102     session->rank = 0;
1103     session->clients_starting = 0;
1104     session->force_position = 0;
1105
1106     for (i = 0; i <= SESSION_WATCH_MAX; i++)
1107     {
1108         session->watchlist[i].data = 0;
1109         session->watchlist[i].fun = 0;
1110     }
1111     session->normalize_cache = normalize_cache_create();
1112     session->session_mutex = 0;
1113     pazpar2_mutex_create(&session->session_mutex, tmp_str);
1114     session_log(session, YLOG_LOG, "create");
1115
1116     session_use(1);
1117     return session;
1118 }
1119
1120 static struct hitsbytarget *hitsbytarget_nb(struct session *se,
1121                                             int *count, NMEM nmem)
1122 {
1123     struct hitsbytarget *res = 0;
1124     struct client_list *l;
1125     size_t sz = 0;
1126
1127     for (l = se->clients_active; l; l = l->next)
1128         sz++;
1129
1130     res = nmem_malloc(nmem, sizeof(*res) * sz);
1131     *count = 0;
1132     for (l = se->clients_active; l; l = l->next)
1133     {
1134         struct client *cl = l->client;
1135         WRBUF w = wrbuf_alloc();
1136         const char *name = session_setting_oneval(client_get_database(cl),
1137                                                   PZ_NAME);
1138         res[*count].id = client_get_id(cl);
1139         res[*count].name = *name ? name : "Unknown";
1140         res[*count].hits = client_get_hits(cl);
1141         res[*count].approximation = client_get_approximation(cl);
1142         res[*count].records = client_get_num_records(cl,
1143                                                      &res[*count].filtered,
1144                                                      0, 0);
1145         res[*count].diagnostic =
1146             client_get_diagnostic(cl, &res[*count].message,
1147                                   &res[*count].addinfo);
1148         res[*count].state = client_get_state_str(cl);
1149         res[*count].connected  = client_get_connection(cl) ? 1 : 0;
1150         session_settings_dump(se, client_get_database(cl), w);
1151         res[*count].settings_xml = nmem_strdup(nmem, wrbuf_cstr(w));
1152         wrbuf_rewind(w);
1153         res[*count].suggestions_xml =
1154             nmem_strdup(nmem, client_get_suggestions_xml(cl, w));
1155
1156         res[*count].query_data =
1157             client_get_query(cl, &res[*count].query_type, nmem);
1158         wrbuf_destroy(w);
1159         (*count)++;
1160     }
1161     return res;
1162 }
1163
1164 struct hitsbytarget *get_hitsbytarget(struct session *se, int *count, NMEM nmem)
1165 {
1166     struct hitsbytarget *p;
1167     session_enter(se, "get_hitsbytarget");
1168     p = hitsbytarget_nb(se, count, nmem);
1169     session_leave(se, "get_hitsbytarget");
1170     return p;
1171 }
1172
1173 // Compares two hitsbytarget nodes by hitcount
1174 static int cmp_ht(const void *p1, const void *p2)
1175 {
1176     const struct hitsbytarget *h1 = p1;
1177     const struct hitsbytarget *h2 = p2;
1178     return h2->hits - h1->hits;
1179 }
1180
1181 // Compares two hitsbytarget nodes by hitcount
1182 static int cmp_ht_approx(const void *p1, const void *p2)
1183 {
1184     const struct hitsbytarget *h1 = p1;
1185     const struct hitsbytarget *h2 = p2;
1186     return h2->approximation - h1->approximation;
1187 }
1188
1189 static int targets_termlist_nb(WRBUF wrbuf, struct session *se, int num,
1190                                NMEM nmem, int version)
1191 {
1192     struct hitsbytarget *ht;
1193     int count, i;
1194
1195     ht = hitsbytarget_nb(se, &count, nmem);
1196     if (version >= 2)
1197         qsort(ht, count, sizeof(struct hitsbytarget), cmp_ht_approx);
1198     else
1199         qsort(ht, count, sizeof(struct hitsbytarget), cmp_ht);
1200     for (i = 0; i < count && i < num && ht[i].hits > 0; i++)
1201     {
1202
1203         // do only print terms which have display names
1204
1205         wrbuf_puts(wrbuf, "<term>\n");
1206
1207         wrbuf_puts(wrbuf, "<id>");
1208         wrbuf_xmlputs(wrbuf, ht[i].id);
1209         wrbuf_puts(wrbuf, "</id>\n");
1210
1211         wrbuf_puts(wrbuf, "<name>");
1212         if (!ht[i].name || !ht[i].name[0])
1213             wrbuf_xmlputs(wrbuf, "NO TARGET NAME");
1214         else
1215             wrbuf_xmlputs(wrbuf, ht[i].name);
1216         wrbuf_puts(wrbuf, "</name>\n");
1217
1218         wrbuf_printf(wrbuf, "<frequency>" ODR_INT_PRINTF "</frequency>\n",
1219                      ht[i].hits);
1220
1221         if (version >= 2) {
1222             // Should not print if we know it isn't a approximation.
1223             wrbuf_printf(wrbuf, "<approximation>" ODR_INT_PRINTF "</approximation>\n", ht[i].approximation);
1224             wrbuf_printf(wrbuf, "<records>%d</records>\n", ht[i].records - ht[i].filtered);
1225             wrbuf_printf(wrbuf, "<filtered>%d</filtered>\n", ht[i].filtered);
1226         }
1227
1228         wrbuf_puts(wrbuf, "<state>");
1229         wrbuf_xmlputs(wrbuf, ht[i].state);
1230         wrbuf_puts(wrbuf, "</state>\n");
1231
1232         wrbuf_printf(wrbuf, "<diagnostic>%d</diagnostic>\n",
1233                      ht[i].diagnostic);
1234         wrbuf_puts(wrbuf, "</term>\n");
1235     }
1236     return count;
1237 }
1238
1239 void perform_termlist(struct http_channel *c, struct session *se,
1240                       const char *name, int num, int version)
1241 {
1242     int j;
1243     NMEM nmem_tmp = nmem_create();
1244     char **names;
1245     int num_names = 0;
1246
1247     if (!name)
1248         name = "*";
1249
1250     nmem_strsplit(nmem_tmp, ",", name, &names, &num_names);
1251
1252     session_enter(se, "perform_termlist");
1253
1254     for (j = 0; j < num_names; j++)
1255     {
1256         const char *tname;
1257         int must_generate_empty = 1; /* bug 5350 */
1258
1259         struct named_termlist *t = se->termlists;
1260         for (; t; t = t->next)
1261         {
1262             tname = t->name;
1263             if (!strcmp(names[j], tname) || !strcmp(names[j], "*"))
1264             {
1265                 struct termlist_score **p = 0;
1266                 int len;
1267
1268                 wrbuf_puts(c->wrbuf, "<list name=\"");
1269                 wrbuf_xmlputs(c->wrbuf, tname);
1270                 wrbuf_puts(c->wrbuf, "\">\n");
1271                 must_generate_empty = 0;
1272
1273                 p = termlist_highscore(t->termlist, &len, nmem_tmp);
1274                 if (p)
1275                 {
1276                     int i;
1277                     for (i = 0; i < len && i < num; i++)
1278                     {
1279                         // prevent sending empty term elements
1280                         if (!p[i]->display_term || !p[i]->display_term[0])
1281                             continue;
1282
1283                         wrbuf_puts(c->wrbuf, "<term>");
1284                         wrbuf_puts(c->wrbuf, "<name>");
1285                         wrbuf_xmlputs(c->wrbuf, p[i]->display_term);
1286                         wrbuf_puts(c->wrbuf, "</name>");
1287                         wrbuf_printf(c->wrbuf,
1288                                      "<frequency>%d</frequency>",
1289                                      p[i]->frequency);
1290                         wrbuf_puts(c->wrbuf, "</term>\n");
1291                     }
1292                 }
1293                 wrbuf_puts(c->wrbuf, "</list>\n");
1294             }
1295         }
1296         tname = "xtargets";
1297         if (!strcmp(names[j], tname) || !strcmp(names[j], "*"))
1298         {
1299             wrbuf_puts(c->wrbuf, "<list name=\"");
1300             wrbuf_xmlputs(c->wrbuf, tname);
1301             wrbuf_puts(c->wrbuf, "\">\n");
1302
1303             targets_termlist_nb(c->wrbuf, se, num, c->nmem, version);
1304             wrbuf_puts(c->wrbuf, "</list>\n");
1305             must_generate_empty = 0;
1306         }
1307         if (must_generate_empty)
1308         {
1309             wrbuf_puts(c->wrbuf, "<list name=\"");
1310             wrbuf_xmlputs(c->wrbuf, names[j]);
1311             wrbuf_puts(c->wrbuf, "\"/>\n");
1312         }
1313     }
1314     session_leave(se, "perform_termlist");
1315     nmem_destroy(nmem_tmp);
1316 }
1317
1318 #ifdef MISSING_HEADERS
1319 void report_nmem_stats(void)
1320 {
1321     size_t in_use, is_free;
1322
1323     nmem_get_memory_in_use(&in_use);
1324     nmem_get_memory_free(&is_free);
1325
1326     yaz_log(YLOG_LOG, "nmem stat: use=%ld free=%ld",
1327             (long) in_use, (long) is_free);
1328 }
1329 #endif
1330
1331 struct record_cluster *show_single_start(struct session *se, const char *id,
1332                                          struct record_cluster **prev_r,
1333                                          struct record_cluster **next_r)
1334 {
1335     struct record_cluster *r = 0;
1336
1337     session_enter(se, "show_single_start");
1338     *prev_r = 0;
1339     *next_r = 0;
1340     reclist_limit(se->reclist, se, 1);
1341
1342     reclist_enter(se->reclist);
1343     while ((r = reclist_read_record(se->reclist)))
1344     {
1345         if (!strcmp(r->recid, id))
1346         {
1347             *next_r = reclist_read_record(se->reclist);
1348             break;
1349         }
1350         *prev_r = r;
1351     }
1352     reclist_leave(se->reclist);
1353     if (!r)
1354         session_leave(se, "show_single_start");
1355     return r;
1356 }
1357
1358 void show_single_stop(struct session *se, struct record_cluster *rec)
1359 {
1360     session_leave(se, "show_single_stop");
1361 }
1362
1363
1364 int session_fetch_more(struct session *se)
1365 {
1366     struct client_list *l;
1367     int ret = 0;
1368
1369     for (l = se->clients_active; l; l = l->next)
1370     {
1371         struct client *cl = l->client;
1372         if (client_get_state(cl) == Client_Idle)
1373         {
1374             if (client_fetch_more(cl))
1375             {
1376                 session_log(se, YLOG_LOG, "%s: more to fetch",
1377                             client_get_id(cl));
1378                 ret = 1;
1379             }
1380             else
1381             {
1382                 int filtered;
1383                 int ingest_failures;
1384                 int record_failures;
1385                 int num = client_get_num_records(
1386                     cl, &filtered, &ingest_failures, &record_failures);
1387
1388                 session_log(se, YLOG_LOG, "%s: hits=" ODR_INT_PRINTF
1389                             " fetched=%d filtered=%d",
1390                             client_get_id(cl),
1391                             client_get_hits(cl),
1392                             num, filtered);
1393                 if (ingest_failures || record_failures)
1394                 {
1395                     session_log(se, YLOG_WARN, "%s:"
1396                                 " ingest failures=%d record failures=%d",
1397                                 client_get_id(cl),
1398                                 ingest_failures, record_failures);
1399                 }
1400             }
1401         }
1402         else
1403         {
1404             session_log(se, YLOG_LOG, "%s: no fetch due to state=%s",
1405                         client_get_id(cl), client_get_state_str(cl));
1406         }
1407
1408     }
1409     return ret;
1410 }
1411
1412 struct record_cluster **show_range_start(struct session *se,
1413                                          struct reclist_sortparms *sp,
1414                                          int start, int *num, int *total,
1415                                          Odr_int *sumhits, Odr_int *approx_hits,
1416                                          void (*show_records_ready)(void *data),
1417                                          struct http_channel *chan)
1418 {
1419     struct record_cluster **recs = 0;
1420     struct reclist_sortparms *spp;
1421     struct client_list *l;
1422     int i;
1423     NMEM nmem_tmp = 0;
1424 #if USE_TIMING
1425     yaz_timing_t t = yaz_timing_create();
1426 #endif
1427     session_enter(se, "show_range_start");
1428     *sumhits = 0;
1429     *approx_hits = 0;
1430     *total = 0;
1431     reclist_limit(se->reclist, se, 0);
1432     if (se->relevance)
1433     {
1434         for (spp = sp; spp; spp = spp->next)
1435             if (spp->type == Metadata_type_relevance)
1436             {
1437                 relevance_prepare_read(se->relevance, se->reclist);
1438                 break;
1439             }
1440         for (l = se->clients_active; l; l = l->next) {
1441             *sumhits += client_get_hits(l->client);
1442             *approx_hits += client_get_approximation(l->client);
1443         }
1444     }
1445     if (se->force_position)
1446     {
1447         nmem_tmp = nmem_create();
1448         sp = reclist_parse_sortparms(nmem_tmp, "position:1", 0);
1449         assert(sp);
1450     }
1451     reclist_sort(se->reclist, sp);
1452     if (nmem_tmp)
1453         nmem_destroy(nmem_tmp);
1454
1455     reclist_enter(se->reclist);
1456     *total = reclist_get_num_records(se->reclist);
1457
1458     for (l = se->clients_active; l; l = l->next)
1459         client_update_show_stat(l->client, 0);
1460
1461     for (i = 0; i < start; i++)
1462     {
1463         struct record_cluster *r = reclist_read_record(se->reclist);
1464         if (!r)
1465         {
1466             *num = 0;
1467             break;
1468         }
1469         else
1470         {
1471             struct record *rec = r->records;
1472             for (;rec; rec = rec->next)
1473                 client_update_show_stat(rec->client, 1);
1474         }
1475     }
1476     recs = nmem_malloc(se->nmem, (*num > 0 ? *num : 1) * sizeof(*recs));
1477     for (i = 0; i < *num; i++)
1478     {
1479         struct record_cluster *r = reclist_read_record(se->reclist);
1480         if (!r)
1481         {
1482             *num = i;
1483             break;
1484         }
1485         else
1486         {
1487             struct record *rec = r->records;
1488             for (;rec; rec = rec->next)
1489                 client_update_show_stat(rec->client, 1);
1490             recs[i] = r;
1491         }
1492     }
1493     reclist_leave(se->reclist);
1494 #if USE_TIMING
1495     yaz_timing_stop(t);
1496     session_log(se, YLOG_LOG, "show %6.5f %3.2f %3.2f",
1497             yaz_timing_get_real(t), yaz_timing_get_user(t),
1498             yaz_timing_get_sys(t));
1499     yaz_timing_destroy(&t);
1500 #endif
1501
1502     if (!session_fetch_more(se))
1503         session_log(se, YLOG_LOG, "can not fetch more");
1504     else
1505     {
1506         show_range_stop(se, recs);
1507         session_log(se, YLOG_LOG, "fetching more in progress");
1508         if (session_set_watch(se, SESSION_WATCH_SHOW,
1509                               show_records_ready, chan, chan))
1510         {
1511             session_log(se, YLOG_WARN, "Ignoring show block");
1512             session_enter(se, "show_range_start");
1513         }
1514         else
1515         {
1516             session_log(se, YLOG_LOG, "session watch OK");
1517             return 0;
1518         }
1519     }
1520     return recs;
1521 }
1522
1523 void show_range_stop(struct session *se, struct record_cluster **recs)
1524 {
1525     session_leave(se, "show_range_stop");
1526 }
1527
1528 void statistics(struct session *se, struct statistics *stat)
1529 {
1530     struct client_list *l;
1531     int count = 0;
1532
1533     memset(stat, 0, sizeof(*stat));
1534     stat->num_hits = 0;
1535     for (l = se->clients_active; l; l = l->next)
1536     {
1537         struct client *cl = l->client;
1538         if (!client_get_connection(cl))
1539             stat->num_no_connection++;
1540         stat->num_hits += client_get_hits(cl);
1541         switch (client_get_state(cl))
1542         {
1543         case Client_Connecting: stat->num_connecting++; break;
1544         case Client_Working: stat->num_working++; break;
1545         case Client_Idle: stat->num_idle++; break;
1546         case Client_Failed: stat->num_failed++; break;
1547         case Client_Error: stat->num_error++; break;
1548         default: break;
1549         }
1550         count++;
1551     }
1552     stat->num_records = se->total_records;
1553
1554     stat->num_clients = count;
1555 }
1556
1557 static struct record_metadata *record_metadata_init(
1558     NMEM nmem, const char *value, const char *norm,
1559     enum conf_metadata_type type,
1560     struct _xmlAttr *attr)
1561 {
1562     struct record_metadata *rec_md = record_metadata_create(nmem);
1563     struct record_metadata_attr **attrp = &rec_md->attributes;
1564
1565     for (; attr; attr = attr->next)
1566     {
1567         if (attr->children && attr->children->content)
1568         {
1569             if (strcmp((const char *) attr->name, "type")
1570                 && strcmp((const char *) attr->name, "empty"))
1571             {  /* skip the "type" + "empty" attribute..
1572                   The "Type" is already part of the element in output
1573                   (md-%s) and so repeating it here is redundant */
1574                 *attrp = nmem_malloc(nmem, sizeof(**attrp));
1575                 (*attrp)->name =
1576                     nmem_strdup(nmem, (const char *) attr->name);
1577                 (*attrp)->value =
1578                     nmem_strdup(nmem, (const char *) attr->children->content);
1579                 attrp = &(*attrp)->next;
1580             }
1581         }
1582     }
1583     *attrp = 0;
1584
1585     switch (type)
1586     {
1587     case Metadata_type_generic:
1588     case Metadata_type_skiparticle:
1589         if (norm)
1590         {
1591             rec_md->data.text.disp = nmem_strdup(nmem, value);
1592             rec_md->data.text.norm = nmem_strdup(nmem, norm);
1593         }
1594         else
1595         {
1596             if (strstr(value, "://")) /* looks like a URL */
1597                 rec_md->data.text.disp = nmem_strdup(nmem, value);
1598             else
1599                 rec_md->data.text.disp =
1600                     normalize7bit_generic(nmem_strdup(nmem, value), " ,/.:([");
1601             rec_md->data.text.norm = rec_md->data.text.disp;
1602         }
1603         rec_md->data.text.sort = 0;
1604         rec_md->data.text.snippet = 0;
1605         break;
1606     case Metadata_type_year:
1607     case Metadata_type_date:
1608     {
1609         int first, last;
1610         int longdate = 0;
1611
1612         if (type == Metadata_type_date)
1613             longdate = 1;
1614         if (extract7bit_dates((char *) value, &first, &last, longdate) < 0)
1615             return 0;
1616
1617         rec_md->data.number.min = first;
1618         rec_md->data.number.max = last;
1619     }
1620     break;
1621     case Metadata_type_float:
1622         rec_md->data.fnumber = atof(value);
1623         break;
1624     case Metadata_type_relevance:
1625     case Metadata_type_position:
1626     case Metadata_type_retrieval:
1627         return 0;
1628     }
1629     return rec_md;
1630 }
1631
1632 static void mergekey_norm_wr(pp2_charset_fact_t charsets,
1633                              WRBUF norm_wr, const char *value)
1634 {
1635     const char *norm_str;
1636     pp2_charset_token_t prt =
1637         pp2_charset_token_create(charsets, "mergekey");
1638
1639     pp2_charset_token_first(prt, value, 0);
1640     while ((norm_str = pp2_charset_token_next(prt)))
1641     {
1642         if (*norm_str)
1643         {
1644             if (wrbuf_len(norm_wr))
1645                 wrbuf_puts(norm_wr, " ");
1646             wrbuf_puts(norm_wr, norm_str);
1647         }
1648     }
1649     pp2_charset_token_destroy(prt);
1650 }
1651
1652 static int get_mergekey_from_doc(xmlDoc *doc, xmlNode *root, const char *name,
1653                                  struct conf_service *service, WRBUF norm_wr)
1654 {
1655     xmlNode *n;
1656     int no_found = 0;
1657     for (n = root->children; n; n = n->next)
1658     {
1659         if (n->type != XML_ELEMENT_NODE)
1660             continue;
1661         if (!strcmp((const char *) n->name, "metadata"))
1662         {
1663             const char *type = yaz_xml_get_prop(n, "type");
1664             if (type == NULL) {
1665                 yaz_log(YLOG_FATAL, "Missing type attribute on metadata element. Skipping!");
1666             }
1667             else if (!strcmp(name, (const char *) type))
1668             {
1669                 xmlChar *value = xmlNodeListGetString(doc, n->children, 1);
1670                 if (value && *value)
1671                 {
1672                     if (wrbuf_len(norm_wr) > 0)
1673                         wrbuf_puts(norm_wr, " ");
1674                     wrbuf_puts(norm_wr, name);
1675                     mergekey_norm_wr(service->charsets, norm_wr,
1676                                      (const char *) value);
1677                     no_found++;
1678                 }
1679                 if (value)
1680                     xmlFree(value);
1681             }
1682         }
1683     }
1684     return no_found;
1685 }
1686
1687 static const char *get_mergekey(xmlDoc *doc, xmlNode *root,
1688                                 struct client *cl, int record_no,
1689                                 struct conf_service *service, NMEM nmem,
1690                                 const char *session_mergekey)
1691 {
1692     char *mergekey_norm = 0;
1693     WRBUF norm_wr = wrbuf_alloc();
1694     const char *mergekey;
1695
1696     if (session_mergekey)
1697     {
1698         int i, num = 0;
1699         char **values = 0;
1700         nmem_strsplit_escape2(nmem, ",", session_mergekey, &values,
1701                               &num, 1, '\\', 1);
1702
1703         for (i = 0; i < num; i++)
1704             get_mergekey_from_doc(doc, root, values[i], service, norm_wr);
1705     }
1706     else if ((mergekey = yaz_xml_get_prop(root, "mergekey")))
1707     {
1708         mergekey_norm_wr(service->charsets, norm_wr, mergekey);
1709     }
1710     else
1711     {
1712         /* no mergekey defined in XSL. Look for mergekey metadata instead */
1713         int field_id;
1714         for (field_id = 0; field_id < service->num_metadata; field_id++)
1715         {
1716             struct conf_metadata *ser_md = &service->metadata[field_id];
1717             if (ser_md->mergekey != Metadata_mergekey_no)
1718             {
1719                 int r = get_mergekey_from_doc(doc, root, ser_md->name,
1720                                               service, norm_wr);
1721                 if (r == 0 && ser_md->mergekey == Metadata_mergekey_required)
1722                 {
1723                     /* no mergekey on this one and it is required..
1724                        Generate unique key instead */
1725                     wrbuf_rewind(norm_wr);
1726                     break;
1727                 }
1728             }
1729         }
1730     }
1731
1732     /* generate unique key if none is not generated already or is empty */
1733     if (wrbuf_len(norm_wr) == 0)
1734     {
1735         wrbuf_printf(norm_wr, "position: %s-%06d",
1736                      client_get_id(cl), record_no);
1737     }
1738     else
1739     {
1740         const char *lead = "content: ";
1741         wrbuf_insert(norm_wr, 0, lead, strlen(lead));
1742     }
1743     if (wrbuf_len(norm_wr) > 0)
1744         mergekey_norm = nmem_strdup(nmem, wrbuf_cstr(norm_wr));
1745     wrbuf_destroy(norm_wr);
1746     return mergekey_norm;
1747 }
1748
1749 /** \brief see if metadata for pz:recordfilter exists
1750     \param root xml root element of normalized record
1751     \param sdb session database for client
1752     \retval 0 if there is no metadata for pz:recordfilter
1753     \retval 1 if there is metadata for pz:recordfilter
1754
1755     If there is no pz:recordfilter defined, this function returns 1
1756     as well.
1757 */
1758
1759 static int check_record_filter(xmlNode *root, struct session_database *sdb)
1760 {
1761     int match = 0;
1762     xmlNode *n;
1763     const char *s;
1764     s = session_setting_oneval(sdb, PZ_RECORDFILTER);
1765
1766     if (!s || !*s)
1767         return 1;
1768
1769     for (n = root->children; n; n = n->next)
1770     {
1771         if (n->type != XML_ELEMENT_NODE)
1772             continue;
1773         if (!strcmp((const char *) n->name, "metadata"))
1774         {
1775             const char *type = yaz_xml_get_prop(n, "type");
1776             if (type)
1777             {
1778                 size_t len;
1779                 int substring;
1780                 const char *eq;
1781
1782                 if ((eq = strchr(s, '=')))
1783                     substring = 0;
1784                 else if ((eq = strchr(s, '~')))
1785                     substring = 1;
1786                 if (eq)
1787                     len = eq - s;
1788                 else
1789                     len = strlen(s);
1790                 if (len == strlen((const char *)type) &&
1791                     !memcmp((const char *) type, s, len))
1792                 {
1793                     xmlChar *value = xmlNodeGetContent(n);
1794                     if (value && *value)
1795                     {
1796                         if (!eq ||
1797                             (substring && strstr((const char *) value, eq+1)) ||
1798                             (!substring && !strcmp((const char *) value, eq + 1)))
1799                             match = 1;
1800                     }
1801                     xmlFree(value);
1802                 }
1803             }
1804         }
1805     }
1806     return match;
1807 }
1808
1809 static int ingest_to_cluster(struct client *cl,
1810                              WRBUF wrbuf_disp,
1811                              WRBUF wrbuf_norm,
1812                              xmlDoc *xdoc,
1813                              xmlNode *root,
1814                              int record_no,
1815                              struct record_metadata_attr *mergekey);
1816
1817 static int ingest_sub_record(struct client *cl, xmlDoc *xdoc, xmlNode *root,
1818                              int record_no, NMEM nmem,
1819                              struct session_database *sdb,
1820                              struct record_metadata_attr *mergekeys)
1821 {
1822     int ret = 0;
1823     struct session *se = client_get_session(cl);
1824     WRBUF wrbuf_disp, wrbuf_norm;
1825
1826     if (!check_record_filter(root, sdb))
1827     {
1828         session_log(se, YLOG_LOG,
1829                     "Filtered out record no %d from %s",
1830                     record_no, sdb->database->id);
1831         return 0;
1832     }
1833     wrbuf_disp = wrbuf_alloc();
1834     wrbuf_norm = wrbuf_alloc();
1835     session_enter(se, "ingest_sub_record");
1836     if (client_get_session(cl) == se && se->relevance)
1837         ret = ingest_to_cluster(cl, wrbuf_disp, wrbuf_norm,
1838                                 xdoc, root, record_no, mergekeys);
1839     session_leave(se, "ingest_sub_record");
1840     wrbuf_destroy(wrbuf_norm);
1841     wrbuf_destroy(wrbuf_disp);
1842     return ret;
1843 }
1844
1845 /** \brief ingest XML record
1846     \param cl client holds the result set for record
1847     \param rec record buffer (0 terminated)
1848     \param record_no record position (1, 2, ..)
1849     \param nmem working NMEM
1850     \retval 0 OK
1851     \retval -1 failure
1852     \retval -2 Filtered
1853 */
1854 int ingest_record(struct client *cl, const char *rec,
1855                   int record_no, NMEM nmem)
1856 {
1857     struct session *se = client_get_session(cl);
1858     struct session_database *sdb = client_get_database(cl);
1859     struct conf_service *service = se->service;
1860     xmlDoc *xdoc = normalize_record(se, sdb, service, rec, nmem);
1861     int r = ingest_xml_record(cl, xdoc, record_no, nmem, 0);
1862     client_store_xdoc(cl, record_no, xdoc);
1863     return r;
1864 }
1865
1866 int ingest_xml_record(struct client *cl, xmlDoc *xdoc,
1867                       int record_no, NMEM nmem, int cached_copy)
1868 {
1869     struct session *se = client_get_session(cl);
1870     struct session_database *sdb = client_get_database(cl);
1871     struct conf_service *service = se->service;
1872     xmlNode *root;
1873     int r = 0;
1874     if (!xdoc)
1875         return -1;
1876
1877     if (global_parameters.dump_records)
1878     {
1879         session_log(se, YLOG_LOG, "Normalized record from %s",
1880                     sdb->database->id);
1881         log_xml_doc(xdoc);
1882     }
1883
1884     root = xmlDocGetRootElement(xdoc);
1885
1886     if (!strcmp((const char *) root->name, "cluster"))
1887     {
1888         int no_merge_keys = 0;
1889         int no_merge_dups = 0;
1890         xmlNode *sroot;
1891         struct record_metadata_attr *mk = 0;
1892
1893         for (sroot = root->children; sroot; sroot = sroot->next)
1894             if (sroot->type == XML_ELEMENT_NODE &&
1895                 !strcmp((const char *) sroot->name, "record"))
1896             {
1897                 struct record_metadata_attr **mkp;
1898                 const char *mergekey_norm =
1899                     get_mergekey(xdoc, sroot, cl, record_no, service, nmem,
1900                                  se->mergekey);
1901                 if (!mergekey_norm)
1902                 {
1903                     r = -1;
1904                     break;
1905                 }
1906                 for (mkp = &mk; *mkp; mkp = &(*mkp)->next)
1907                     if (!strcmp((*mkp)->value, mergekey_norm))
1908                         break;
1909                 if (!*mkp)
1910                 {
1911                     *mkp = (struct record_metadata_attr*)
1912                         nmem_malloc(nmem, sizeof(**mkp));
1913                     (*mkp)->name = 0;
1914                     (*mkp)->value = nmem_strdup(nmem, mergekey_norm);
1915                     (*mkp)->next = 0;
1916                     no_merge_keys++;
1917                 }
1918                 else
1919                     no_merge_dups++;
1920             }
1921         if (no_merge_keys > 1 || no_merge_dups > 0)
1922         {
1923             yaz_log(YLOG_LOG, "Got %d mergekeys, %d dups for position %d",
1924                     no_merge_keys, no_merge_dups, record_no);
1925         }
1926         for (sroot = root->children; !r && sroot; sroot = sroot->next)
1927             if (sroot->type == XML_ELEMENT_NODE &&
1928                 !strcmp((const char *) sroot->name, "record"))
1929             {
1930                 if (!cached_copy)
1931                     insert_settings_values(sdb, xdoc, root, service);
1932                 r = ingest_sub_record(cl, xdoc, sroot, record_no, nmem, sdb,
1933                                       mk);
1934             }
1935     }
1936     else if (!strcmp((const char *) root->name, "record"))
1937     {
1938         const char *mergekey_norm =
1939             get_mergekey(xdoc, root, cl, record_no, service, nmem,
1940                          se->mergekey);
1941         if (mergekey_norm)
1942         {
1943             struct record_metadata_attr *mk = (struct record_metadata_attr*)
1944                 nmem_malloc(nmem, sizeof(*mk));
1945             mk->name = 0;
1946             mk->value = nmem_strdup(nmem, mergekey_norm);
1947             mk->next = 0;
1948
1949             if (!cached_copy)
1950                 insert_settings_values(sdb, xdoc, root, service);
1951             r = ingest_sub_record(cl, xdoc, root, record_no, nmem, sdb, mk);
1952         }
1953     }
1954     else
1955     {
1956         session_log(se, YLOG_WARN, "Bad pz root element: %s",
1957                     (const char *) root->name);
1958         r = -1;
1959     }
1960     return r;
1961 }
1962
1963
1964 //    struct conf_metadata *ser_md = &service->metadata[md_field_id];
1965 //    struct record_metadata *rec_md = record->metadata[md_field_id];
1966 static int match_metadata_local(struct conf_service *service,
1967                                 struct conf_metadata *ser_md,
1968                                 struct record_metadata *rec_md0,
1969                                 char **values, int num_v)
1970 {
1971     int i;
1972     struct record_metadata *rec_md = rec_md0;
1973     WRBUF val_wr = 0;
1974     WRBUF text_wr = wrbuf_alloc();
1975     for (i = 0; i < num_v; )
1976     {
1977         if (rec_md)
1978         {
1979             if (ser_md->type == Metadata_type_year
1980                 || ser_md->type == Metadata_type_date)
1981             {
1982                 int y = atoi(values[i]);
1983                 if (y >= rec_md->data.number.min
1984                     && y <= rec_md->data.number.max)
1985                     break;
1986             }
1987             else
1988             {
1989                 if (!val_wr)
1990                 {
1991                     val_wr = wrbuf_alloc();
1992                     mergekey_norm_wr(service->charsets, val_wr, values[i]);
1993                 }
1994                 wrbuf_rewind(text_wr);
1995                 mergekey_norm_wr(service->charsets, text_wr,
1996                                  rec_md->data.text.disp);
1997                 if (!strcmp(wrbuf_cstr(val_wr), wrbuf_cstr(text_wr)))
1998                     break;
1999             }
2000             rec_md = rec_md->next;
2001         }
2002         else
2003         {
2004             rec_md = rec_md0;
2005             wrbuf_destroy(val_wr);
2006             val_wr = 0;
2007             i++;
2008         }
2009     }
2010     wrbuf_destroy(val_wr);
2011     wrbuf_destroy(text_wr);
2012     return i < num_v ? 1 : 0;
2013 }
2014
2015 int session_check_cluster_limit(struct session *se, struct record_cluster *rec)
2016 {
2017     int i;
2018     struct conf_service *service = se->service;
2019     int ret = 1;
2020     const char *name;
2021     const char *value;
2022     NMEM nmem_tmp = nmem_create();
2023
2024     for (i = 0; (name = facet_limits_get(se->facet_limits, i, &value)); i++)
2025     {
2026         int j;
2027         for (j = 0; j < service->num_metadata; j++)
2028         {
2029             struct conf_metadata *md = service->metadata + j;
2030             if (!strcmp(md->name, name) && md->limitcluster)
2031             {
2032                 char **values = 0;
2033                 int num = 0;
2034                 int md_field_id =
2035                     conf_service_metadata_field_id(service,
2036                                                    md->limitcluster);
2037
2038                 if (md_field_id < 0)
2039                 {
2040                     ret = 0;
2041                     break;
2042                 }
2043
2044                 nmem_strsplit_escape2(nmem_tmp, "|", value, &values,
2045                                       &num, 1, '\\', 1);
2046
2047                 if (!match_metadata_local(service,
2048                                           &service->metadata[md_field_id],
2049                                           rec->metadata[md_field_id],
2050                                           values, num))
2051                 {
2052                     ret = 0;
2053                     break;
2054                 }
2055             }
2056         }
2057     }
2058     nmem_destroy(nmem_tmp);
2059     return ret;
2060 }
2061
2062 // Skip record on non-zero
2063 static int check_limit_local(struct client *cl,
2064                              struct record *record,
2065                              int record_no)
2066 {
2067     int skip_record = 0;
2068     struct session *se = client_get_session(cl);
2069     struct conf_service *service = se->service;
2070     NMEM nmem_tmp = nmem_create();
2071     struct session_database *sdb = client_get_database(cl);
2072     int l = 0;
2073     while (!skip_record)
2074     {
2075         int md_field_id;
2076         char **values = 0;
2077         int num_v = 0;
2078         const char *name =
2079             client_get_facet_limit_local(cl, sdb, &l, nmem_tmp,
2080                                          &num_v, &values);
2081         if (!name)
2082             break;
2083
2084         if (!strcmp(name, "*"))
2085         {
2086             for (md_field_id = 0; md_field_id < service->num_metadata;
2087                  md_field_id++)
2088             {
2089                 if (match_metadata_local(
2090                         service,
2091                         &service->metadata[md_field_id],
2092                         record->metadata[md_field_id],
2093                         values, num_v))
2094                     break;
2095             }
2096             if (md_field_id == service->num_metadata)
2097                 skip_record = 1;
2098         }
2099         else
2100         {
2101             md_field_id = conf_service_metadata_field_id(service, name);
2102             if (md_field_id < 0)
2103             {
2104                 skip_record = 1;
2105                 break;
2106             }
2107             if (!match_metadata_local(
2108                     service,
2109                     &service->metadata[md_field_id],
2110                     record->metadata[md_field_id],
2111                     values, num_v))
2112             {
2113                 skip_record = 1;
2114             }
2115         }
2116     }
2117     nmem_destroy(nmem_tmp);
2118     return skip_record;
2119 }
2120
2121 static int ingest_to_cluster(struct client *cl,
2122                              WRBUF wrbuf_disp,
2123                              WRBUF wrbuf_norm,
2124                              xmlDoc *xdoc,
2125                              xmlNode *root,
2126                              int record_no,
2127                              struct record_metadata_attr *merge_keys)
2128 {
2129     xmlNode *n;
2130     struct session *se = client_get_session(cl);
2131     struct conf_service *service = se->service;
2132     int term_factor = 1;
2133     struct record_cluster *cluster;
2134     struct record_metadata **metadata0;
2135     struct session_database *sdb = client_get_database(cl);
2136     NMEM ingest_nmem = 0;
2137     char **rank_values = 0;
2138     int rank_num = 0;
2139     struct record *record = record_create(se->nmem,
2140                                           service->num_metadata,
2141                                           service->num_sortkeys, cl,
2142                                           record_no);
2143
2144     for (n = root->children; n; n = n->next)
2145     {
2146         if (n->type != XML_ELEMENT_NODE)
2147             continue;
2148         if (!strcmp((const char *) n->name, "metadata"))
2149         {
2150             struct conf_metadata *ser_md = 0;
2151             struct record_metadata **wheretoput = 0;
2152             struct record_metadata *rec_md = 0;
2153             int md_field_id = -1;
2154             xmlChar *value0;
2155             const char *type = yaz_xml_get_prop(n, "type");
2156
2157             if (!type)
2158                 continue;
2159
2160             md_field_id
2161                 = conf_service_metadata_field_id(service, (const char *) type);
2162             if (md_field_id < 0)
2163             {
2164                 if (se->number_of_warnings_unknown_metadata == 0)
2165                 {
2166                     session_log(se, YLOG_WARN,
2167                             "Ignoring unknown metadata element: %s", type);
2168                 }
2169                 se->number_of_warnings_unknown_metadata++;
2170                 continue;
2171             }
2172
2173             wrbuf_rewind(wrbuf_disp);
2174             value0 = xmlNodeListGetString(xdoc, n->children, 1);
2175             if (!value0 || !*value0)
2176             {
2177                 const char *empty = yaz_xml_get_prop(n, "empty");
2178                 if (!empty)
2179                     continue;
2180                 wrbuf_puts(wrbuf_disp, (const char *) empty);
2181             }
2182             else
2183             {
2184                 wrbuf_puts(wrbuf_disp, (const char *) value0);
2185             }
2186             if (value0)
2187                 xmlFree(value0);
2188             ser_md = &service->metadata[md_field_id];
2189
2190             // non-merged metadata
2191             rec_md = record_metadata_init(se->nmem, wrbuf_cstr(wrbuf_disp), 0,
2192                                           ser_md->type, n->properties);
2193             if (!rec_md)
2194             {
2195                 session_log(se, YLOG_WARN, "bad metadata data '%s' "
2196                             "for element '%s'", wrbuf_cstr(wrbuf_disp), type);
2197                 continue;
2198             }
2199
2200             if (ser_md->type == Metadata_type_generic)
2201             {
2202                 WRBUF w = wrbuf_alloc();
2203                 if (relevance_snippet(se->relevance,
2204                                       wrbuf_cstr(wrbuf_disp), ser_md->name, w))
2205                     rec_md->data.text.snippet = nmem_strdup(se->nmem,
2206                                                             wrbuf_cstr(w));
2207                 wrbuf_destroy(w);
2208             }
2209
2210
2211             wheretoput = &record->metadata[md_field_id];
2212             while (*wheretoput)
2213                 wheretoput = &(*wheretoput)->next;
2214             *wheretoput = rec_md;
2215         }
2216     }
2217
2218     if (check_limit_local(cl, record, record_no))
2219     {
2220         return -2;
2221     }
2222     cluster = reclist_insert(se->reclist, se->relevance, service, record,
2223                              merge_keys, &se->total_merged);
2224     if (!cluster)
2225     {
2226         return 0; // complete match with existing record
2227     }
2228
2229     {
2230         const char *use_term_factor_str =
2231             session_setting_oneval(sdb, PZ_TERMLIST_TERM_FACTOR);
2232         if (use_term_factor_str && use_term_factor_str[0] == '1')
2233         {
2234             int maxrecs = client_get_maxrecs(cl);
2235             int hits = (int) client_get_hits(cl);
2236             term_factor = MAX(hits, maxrecs) /  MAX(1, maxrecs);
2237             assert(term_factor >= 1);
2238             session_log(se, YLOG_DEBUG, "Using term factor: %d (%d / %d)",
2239                         term_factor, MAX(hits, maxrecs), MAX(1, maxrecs));
2240         }
2241     }
2242
2243     if (global_parameters.dump_records)
2244         session_log(se, YLOG_LOG, "Cluster id %s from %s (#%d)", cluster->recid,
2245                     sdb->database->id, record_no);
2246
2247     // original metadata, to check if first existence of a field
2248     metadata0 = xmalloc(sizeof(*metadata0) * service->num_metadata);
2249     memcpy(metadata0, cluster->metadata,
2250            sizeof(*metadata0) * service->num_metadata);
2251
2252     ingest_nmem = nmem_create();
2253     if (se->rank)
2254     {
2255         yaz_log(YLOG_LOG, "local in sort : %s", se->rank);
2256         nmem_strsplit_escape2(ingest_nmem, ",", se->rank, &rank_values,
2257                               &rank_num, 1, '\\', 1);
2258     }
2259
2260     // now parsing XML record and adding data to cluster or record metadata
2261     for (n = root->children; n; n = n->next)
2262     {
2263         if (n->type != XML_ELEMENT_NODE)
2264             continue;
2265         if (!strcmp((const char *) n->name, "metadata"))
2266         {
2267             struct conf_metadata *ser_md = 0;
2268             struct conf_sortkey *ser_sk = 0;
2269             struct record_metadata **wheretoput = 0;
2270             struct record_metadata *rec_md = 0;
2271             int md_field_id = -1;
2272             int sk_field_id = -1;
2273             const char *rank = 0;
2274             const char *xml_rank = 0;
2275             const char *type = 0;
2276             xmlChar *value0;
2277
2278             type = yaz_xml_get_prop(n, "type");
2279             if (!type)
2280                 continue;
2281
2282             md_field_id
2283                 = conf_service_metadata_field_id(service, (const char *) type);
2284             if (md_field_id < 0)
2285                 continue;
2286
2287             ser_md = &service->metadata[md_field_id];
2288
2289             if (ser_md->sortkey_offset >= 0)
2290             {
2291                 sk_field_id = ser_md->sortkey_offset;
2292                 ser_sk = &service->sortkeys[sk_field_id];
2293             }
2294
2295             wrbuf_rewind(wrbuf_disp);
2296             wrbuf_rewind(wrbuf_norm);
2297
2298             value0 = xmlNodeListGetString(xdoc, n->children, 1);
2299             if (!value0 || !*value0)
2300             {
2301                 if (value0)
2302                     xmlFree(value0);
2303                 continue;
2304             }
2305
2306             if (ser_md->icurule)
2307             {
2308                 run_icu(se, ser_md->icurule, (const char *) value0,
2309                         wrbuf_norm, wrbuf_disp);
2310                 yaz_log(YLOG_LOG, "run_icu input=%s norm=%s disp=%s",
2311                         (const char *) value0,
2312                         wrbuf_cstr(wrbuf_norm), wrbuf_cstr(wrbuf_disp));
2313                 rec_md = record_metadata_init(se->nmem, wrbuf_cstr(wrbuf_disp),
2314                                               wrbuf_cstr(wrbuf_norm),
2315                                               ser_md->type, 0);
2316             }
2317             else
2318             {
2319                 wrbuf_puts(wrbuf_disp, (const char *) value0);
2320                 rec_md = record_metadata_init(se->nmem, wrbuf_cstr(wrbuf_disp),
2321                                               0,
2322                                               ser_md->type, 0);
2323             }
2324
2325             xmlFree(value0);
2326
2327             // see if the field was not in cluster already (from beginning)
2328             if (!rec_md)
2329                 continue;
2330
2331             if (rank_num)
2332             {
2333                 int i;
2334                 for (i = 0; i < rank_num; i++)
2335                 {
2336                     const char *val = rank_values[i];
2337                     const char *cp = strchr(val, '=');
2338                     if (!cp)
2339                         continue;
2340                     if ((cp - val) == strlen((const char *) type)
2341                         && !memcmp(val, type, cp - val))
2342                     {
2343                         rank = cp + 1;
2344                         break;
2345                     }
2346                 }
2347             }
2348             else
2349             {
2350                 xml_rank = yaz_xml_get_prop(n, "rank");
2351                 rank = xml_rank ? (const char *) xml_rank : ser_md->rank;
2352             }
2353
2354             wheretoput = &cluster->metadata[md_field_id];
2355
2356             if (ser_md->merge == Metadata_merge_first)
2357             {
2358                 if (!metadata0[md_field_id])
2359                 {
2360                     while (*wheretoput)
2361                         wheretoput = &(*wheretoput)->next;
2362                     *wheretoput = rec_md;
2363                 }
2364             }
2365             else if (ser_md->merge == Metadata_merge_unique)
2366             {
2367                 while (*wheretoput)
2368                 {
2369                     if (!strcmp((const char *) (*wheretoput)->data.text.norm,
2370                                 rec_md->data.text.norm))
2371                         break;
2372                     wheretoput = &(*wheretoput)->next;
2373                 }
2374                 if (!*wheretoput)
2375                     *wheretoput = rec_md;
2376             }
2377             else if (ser_md->merge == Metadata_merge_longest)
2378             {
2379                 if (!*wheretoput
2380                     || strlen(rec_md->data.text.norm)
2381                     > strlen((*wheretoput)->data.text.norm))
2382                 {
2383                     *wheretoput = rec_md;
2384                     if (ser_sk)
2385                     {
2386                         pp2_charset_token_t prt;
2387                         const char *sort_str = 0;
2388                         int skip_article =
2389                             ser_sk->type == Metadata_type_skiparticle;
2390
2391                         if (!cluster->sortkeys[sk_field_id])
2392                             cluster->sortkeys[sk_field_id] =
2393                                 nmem_malloc(se->nmem,
2394                                             sizeof(union data_types));
2395
2396                         prt =
2397                             pp2_charset_token_create(service->charsets, "sort");
2398
2399                         pp2_charset_token_first(prt, rec_md->data.text.disp,
2400                                                 skip_article);
2401
2402                         pp2_charset_token_next(prt);
2403
2404                         sort_str = pp2_get_sort(prt);
2405
2406                         cluster->sortkeys[sk_field_id]->text.disp =
2407                             rec_md->data.text.disp;
2408                         if (!sort_str)
2409                         {
2410                             sort_str = rec_md->data.text.disp;
2411                             session_log(se, YLOG_WARN,
2412                                     "Could not make sortkey. Bug #1858");
2413                         }
2414                         cluster->sortkeys[sk_field_id]->text.sort =
2415                             nmem_strdup(se->nmem, sort_str);
2416                         pp2_charset_token_destroy(prt);
2417                     }
2418                 }
2419             }
2420             else if (ser_md->merge == Metadata_merge_all)
2421             {
2422                 while (*wheretoput)
2423                     wheretoput = &(*wheretoput)->next;
2424                 *wheretoput = rec_md;
2425             }
2426             else if (ser_md->merge == Metadata_merge_range)
2427             {
2428                 if (!*wheretoput)
2429                 {
2430                     *wheretoput = rec_md;
2431                     if (ser_sk)
2432                         cluster->sortkeys[sk_field_id]
2433                             = &rec_md->data;
2434                 }
2435                 else
2436                 {
2437                     int this_min = rec_md->data.number.min;
2438                     int this_max = rec_md->data.number.max;
2439                     if (this_min < (*wheretoput)->data.number.min)
2440                         (*wheretoput)->data.number.min = this_min;
2441                     if (this_max > (*wheretoput)->data.number.max)
2442                         (*wheretoput)->data.number.max = this_max;
2443                 }
2444             }
2445
2446             // ranking of _all_ fields enabled ...
2447             if (rank)
2448             {
2449                 relevance_countwords(se->relevance, cluster,
2450                                      wrbuf_cstr(wrbuf_disp),
2451                                      rank, ser_md->name);
2452             }
2453             // construct facets ... unless the client already has reported them
2454             if (ser_md->termlist && !client_has_facet(cl, (char *) type))
2455             {
2456                 if (ser_md->type == Metadata_type_year)
2457                 {
2458                     char year[64];
2459                     sprintf(year, "%d", rec_md->data.number.max);
2460
2461                     add_facet(se, (char *) type, year, term_factor, cl);
2462                     if (rec_md->data.number.max != rec_md->data.number.min)
2463                     {
2464                         sprintf(year, "%d", rec_md->data.number.min);
2465                         add_facet(se, (char *) type, year, term_factor, cl);
2466                     }
2467                 }
2468                 else
2469                     add_facet(se, type, wrbuf_cstr(wrbuf_disp), term_factor, cl);
2470             }
2471         }
2472         else
2473         {
2474             if (se->number_of_warnings_unknown_elements == 0)
2475                 session_log(se, YLOG_WARN,
2476                         "Unexpected element in internal record: %s", n->name);
2477             se->number_of_warnings_unknown_elements++;
2478         }
2479     }
2480     nmem_destroy(ingest_nmem);
2481     xfree(metadata0);
2482     relevance_donerecord(se->relevance, cluster);
2483     se->total_records++;
2484
2485     return 0;
2486 }
2487
2488 void session_log(struct session *s, int level, const char *fmt, ...)
2489 {
2490     char buf[1024];
2491     va_list ap;
2492     va_start(ap, fmt);
2493
2494     yaz_vsnprintf(buf, sizeof(buf)-30, fmt, ap);
2495     yaz_log(level, "Session %u: %s", s ? s->session_id : 0, buf);
2496
2497     va_end(ap);
2498 }
2499
2500 /*
2501  * Local variables:
2502  * c-basic-offset: 4
2503  * c-file-style: "Stroustrup"
2504  * indent-tabs-mode: nil
2505  * End:
2506  * vim: shiftwidth=4 tabstop=8 expandtab
2507  */
2508