src/session.c

   1 /* This file is part of Pazpar2.
   2    Copyright (C) Index Data
   3
   4 Pazpar2 is free software; you can redistribute it and/or modify it under
   5 the terms of the GNU General Public License as published by the Free
   6 Software Foundation; either version 2, or (at your option) any later
   7 version.
   8
   9 Pazpar2 is distributed in the hope that it will be useful, but WITHOUT ANY
  10 WARRANTY; without even the implied warranty of MERCHANTABILITY or
  11 FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License
  12 for more details.
  13
  14 You should have received a copy of the GNU General Public License
  15 along with this program; if not, write to the Free Software
  16 Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
  17
  18 */
  19
  20 /** \file session.c
  21     \brief high-level logic; mostly user sessions and settings
  22 */
  23
  24 #if HAVE_CONFIG_H
  25 #include <config.h>
  26 #endif
  27
  28 #include <time.h>
  29 #include <stdlib.h>
  30 #include <stdio.h>
  31 #include <string.h>
  32 #if HAVE_SYS_TIME_H
  33 #include <sys/time.h>
  34 #endif
  35 #if HAVE_UNISTD_H
  36 #include <unistd.h>
  37 #endif
  38 #ifdef WIN32
  39 #include <windows.h>
  40 #endif
  41 #include <signal.h>
  42 #include <ctype.h>
  43 #include <assert.h>
  44 #include <math.h>
  45
  46 #include <yaz/marcdisp.h>
  47 #include <yaz/comstack.h>
  48 #include <yaz/tcpip.h>
  49 #include <yaz/proto.h>
  50 #include <yaz/readconf.h>
  51 #include <yaz/pquery.h>
  52 #include <yaz/otherinfo.h>
  53 #include <yaz/yaz-util.h>
  54 #include <yaz/nmem.h>
  55 #include <yaz/query-charset.h>
  56 #include <yaz/querytowrbuf.h>
  57 #include <yaz/oid_db.h>
  58 #include <yaz/snprintf.h>
  59
  60 #define USE_TIMING 1
  61 #if USE_TIMING
  62 #include <yaz/timing.h>
  63 #endif
  64
  65 #include "ppmutex.h"
  66 #include "parameters.h"
  67 #include "session.h"
  68 #include "eventl.h"
  69 #include "http.h"
  70 #include "termlists.h"
  71 #include "reclists.h"
  72 #include "relevance.h"
  73 #include "database.h"
  74 #include "client.h"
  75 #include "settings.h"
  76 #include "normalize7bit.h"
  77
  78 #include <libxml/tree.h>
  79
  80 #define MAX_CHUNK 15
  81
  82 #define MAX(a,b) ((a)>(b)?(a):(b))
  83
  84 // Note: Some things in this structure will eventually move to configuration
  85 struct parameters global_parameters =
  86 {
  87     0,   // dump_records
  88     0,   // debug_mode
  89     0,   // predictable sessions
  90 };
  91
  92 struct client_list {
  93     struct client *client;
  94     struct client_list *next;
  95 };
  96
  97 /* session counting (1) , disable client counting (0) */
  98 static YAZ_MUTEX g_session_mutex = 0;
  99 static int no_sessions = 0;
 100
 101 static int session_use(int delta)
 102 {
 103     int sessions;
 104     if (!g_session_mutex)
 105         yaz_mutex_create(&g_session_mutex);
 106     yaz_mutex_enter(g_session_mutex);
 107     no_sessions += delta;
 108     sessions = no_sessions;
 109     yaz_mutex_leave(g_session_mutex);
 110     yaz_log(YLOG_DEBUG, "%s sessions=%d", delta == 0 ? "" :
 111             (delta > 0 ? "INC" : "DEC"), no_sessions);
 112     return sessions;
 113 }
 114
 115 int sessions_count(void)
 116 {
 117     return session_use(0);
 118 }
 119
 120 static void log_xml_doc(xmlDoc *doc)
 121 {
 122     FILE *lf = yaz_log_file();
 123     xmlChar *result = 0;
 124     int len = 0;
 125 #if LIBXML_VERSION >= 20600
 126     xmlDocDumpFormatMemory(doc, &result, &len, 1);
 127 #else
 128     xmlDocDumpMemory(doc, &result, &len);
 129 #endif
 130     if (lf && len)
 131     {
 132         (void) fwrite(result, 1, len, lf);
 133         fprintf(lf, "\n");
 134     }
 135     xmlFree(result);
 136 }
 137
 138 static void session_enter(struct session *s, const char *caller)
 139 {
 140     if (caller)
 141         session_log(s, YLOG_DEBUG, "Session lock by %s", caller);
 142     yaz_mutex_enter(s->session_mutex);
 143 }
 144
 145 static void session_leave(struct session *s, const char *caller)
 146 {
 147     yaz_mutex_leave(s->session_mutex);
 148     if (caller)
 149         session_log(s, YLOG_DEBUG, "Session unlock by %s", caller);
 150 }
 151
 152 static void session_normalize_facet(struct session *s,
 153                                     const char *type, const char *value,
 154                                     WRBUF display_wrbuf, WRBUF facet_wrbuf)
 155 {
 156     struct conf_service *service = s->service;
 157     pp2_charset_token_t prt;
 158     const char *facet_component;
 159     int i;
 160     const char *icu_chain_id = 0;
 161
 162     for (i = 0; i < service->num_metadata; i++)
 163         if (!strcmp((service->metadata + i)->name, type))
 164             icu_chain_id = (service->metadata + i)->facetrule;
 165     if (!icu_chain_id)
 166         icu_chain_id = "facet";
 167     prt = pp2_charset_token_create(service->charsets, icu_chain_id);
 168     if (!prt)
 169     {
 170         session_log(s, YLOG_FATAL,
 171                     "Unknown ICU chain '%s' for facet of type '%s'",
 172                 icu_chain_id, type);
 173         wrbuf_destroy(facet_wrbuf);
 174         wrbuf_destroy(display_wrbuf);
 175         return;
 176     }
 177     pp2_charset_token_first(prt, value, 0);
 178     while ((facet_component = pp2_charset_token_next(prt)))
 179     {
 180         const char *display_component;
 181         if (*facet_component)
 182         {
 183             if (wrbuf_len(facet_wrbuf))
 184                 wrbuf_puts(facet_wrbuf, " ");
 185             wrbuf_puts(facet_wrbuf, facet_component);
 186         }
 187         display_component = pp2_get_display(prt);
 188         if (display_component)
 189         {
 190             if (wrbuf_len(display_wrbuf))
 191                 wrbuf_puts(display_wrbuf, " ");
 192             wrbuf_puts(display_wrbuf, display_component);
 193         }
 194     }
 195     pp2_charset_token_destroy(prt);
 196 }
 197
 198 void add_facet(struct session *s, const char *type, const char *value, int count)
 199 {
 200     WRBUF facet_wrbuf = wrbuf_alloc();
 201     WRBUF display_wrbuf = wrbuf_alloc();
 202
 203     session_normalize_facet(s, type, value, display_wrbuf, facet_wrbuf);
 204
 205     if (wrbuf_len(facet_wrbuf))
 206     {
 207         int i;
 208         for (i = 0; i < s->num_termlists; i++)
 209             if (!strcmp(s->termlists[i].name, type))
 210                 break;
 211         if (i == s->num_termlists)
 212         {
 213             if (i == SESSION_MAX_TERMLISTS)
 214             {
 215                 session_log(s, YLOG_FATAL, "Too many termlists");
 216                 wrbuf_destroy(facet_wrbuf);
 217                 wrbuf_destroy(display_wrbuf);
 218                 return;
 219             }
 220
 221             s->termlists[i].name = nmem_strdup(s->nmem, type);
 222             s->termlists[i].termlist = termlist_create(s->nmem);
 223             s->num_termlists = i + 1;
 224         }
 225
 226 #if 0
 227         session_log(s, YLOG_LOG, "Facets for %s: %s norm:%s (%d)", type, value, wrbuf_cstr(facet_wrbuf), count);
 228 #endif
 229         termlist_insert(s->termlists[i].termlist, wrbuf_cstr(display_wrbuf),
 230                         wrbuf_cstr(facet_wrbuf), count);
 231     }
 232     wrbuf_destroy(facet_wrbuf);
 233     wrbuf_destroy(display_wrbuf);
 234 }
 235
 236 static xmlDoc *record_to_xml(struct session *se,
 237                              struct session_database *sdb, const char *rec)
 238 {
 239     struct database *db = sdb->database;
 240     xmlDoc *rdoc = 0;
 241
 242     rdoc = xmlParseMemory(rec, strlen(rec));
 243
 244     if (!rdoc)
 245     {
 246         session_log(se, YLOG_WARN, "Non-wellformed XML");
 247         return 0;
 248     }
 249
 250     if (global_parameters.dump_records)
 251     {
 252         session_log(se, YLOG_LOG, "Un-normalized record from %s", db->id);
 253         log_xml_doc(rdoc);
 254     }
 255
 256     return rdoc;
 257 }
 258
 259 #define MAX_XSLT_ARGS 16
 260
 261 // Add static values from session database settings if applicable
 262 static void insert_settings_parameters(struct session_database *sdb,
 263                                        struct conf_service *service,
 264                                        char **parms,
 265                                        NMEM nmem)
 266 {
 267     int i;
 268     int nparms = 0;
 269     int offset = 0;
 270
 271     for (i = 0; i < service->num_metadata; i++)
 272     {
 273         struct conf_metadata *md = &service->metadata[i];
 274         int setting;
 275
 276         if (md->setting == Metadata_setting_parameter &&
 277             (setting = settings_lookup_offset(service, md->name)) >= 0)
 278         {
 279             const char *val = session_setting_oneval(sdb, setting);
 280             if (val && nparms < MAX_XSLT_ARGS)
 281             {
 282                 char *buf;
 283                 int len = strlen(val);
 284                 buf = nmem_malloc(nmem, len + 3);
 285                 buf[0] = '\'';
 286                 strcpy(buf + 1, val);
 287                 buf[len+1] = '\'';
 288                 buf[len+2] = '\0';
 289                 parms[offset++] = md->name;
 290                 parms[offset++] = buf;
 291                 nparms++;
 292             }
 293         }
 294     }
 295     parms[offset] = 0;
 296 }
 297
 298 // Add static values from session database settings if applicable
 299 static void insert_settings_values(struct session_database *sdb, xmlDoc *doc,
 300                                    xmlNode *root,
 301                                    struct conf_service *service)
 302 {
 303     int i;
 304
 305     for (i = 0; i < service->num_metadata; i++)
 306     {
 307         struct conf_metadata *md = &service->metadata[i];
 308         int offset;
 309
 310         if (md->setting == Metadata_setting_postproc &&
 311             (offset = settings_lookup_offset(service, md->name)) >= 0)
 312         {
 313             const char *val = session_setting_oneval(sdb, offset);
 314             if (val)
 315             {
 316                 xmlNode *n = xmlNewTextChild(root, 0, (xmlChar *) "metadata",
 317                                              (xmlChar *) val);
 318                 xmlSetProp(n, (xmlChar *) "type", (xmlChar *) md->name);
 319             }
 320         }
 321     }
 322 }
 323
 324 static xmlDoc *normalize_record(struct session *se,
 325                                 struct session_database *sdb,
 326                                 struct conf_service *service,
 327                                 const char *rec, NMEM nmem)
 328 {
 329     xmlDoc *rdoc = record_to_xml(se, sdb, rec);
 330
 331     if (rdoc)
 332     {
 333         char *parms[MAX_XSLT_ARGS*2+1];
 334
 335         insert_settings_parameters(sdb, service, parms, nmem);
 336
 337         if (normalize_record_transform(sdb->map, &rdoc, (const char **)parms))
 338         {
 339             session_log(se, YLOG_WARN, "Normalize failed");
 340         }
 341     }
 342     return rdoc;
 343 }
 344
 345 void session_settings_dump(struct session *se,
 346                            struct session_database *db,
 347                            WRBUF w)
 348 {
 349     if (db->settings)
 350     {
 351         int i, num = db->num_settings;
 352         for (i = 0; i < num; i++)
 353         {
 354             struct setting *s = db->settings[i];
 355             for (;s ; s = s->next)
 356             {
 357                 wrbuf_puts(w, "<set name=\"");
 358                 wrbuf_xmlputs(w, s->name);
 359                 wrbuf_puts(w, "\" value=\"");
 360                 wrbuf_xmlputs(w, s->value);
 361                 wrbuf_puts(w, "\"/>");
 362             }
 363             if (db->settings[i])
 364                 wrbuf_puts(w, "\n");
 365         }
 366     }
 367 }
 368
 369 // Retrieve first defined value for 'name' for given database.
 370 // Will be extended to take into account user associated with session
 371 const char *session_setting_oneval(struct session_database *db, int offset)
 372 {
 373     if (offset >= db->num_settings || !db->settings[offset])
 374         return "";
 375     return db->settings[offset]->value;
 376 }
 377
 378 // Prepare XSLT stylesheets for record normalization
 379 // Structures are allocated on the session_wide nmem to avoid having
 380 // to recompute this for every search. This would lead
 381 // to leaking if a single session was to repeatedly change the PZ_XSLT
 382 // setting. However, this is not a realistic use scenario.
 383 static int prepare_map(struct session *se, struct session_database *sdb)
 384 {
 385     if (sdb->settings && !sdb->map)
 386     {
 387         const char *s;
 388
 389         if (sdb->settings[PZ_XSLT] &&
 390             (s = session_setting_oneval(sdb, PZ_XSLT)))
 391         {
 392             char auto_stylesheet[256];
 393
 394             if (!strcmp(s, "auto"))
 395             {
 396                 const char *request_syntax = session_setting_oneval(
 397                     sdb, PZ_REQUESTSYNTAX);
 398                 if (request_syntax)
 399                 {
 400                     char *cp;
 401                     yaz_snprintf(auto_stylesheet, sizeof(auto_stylesheet),
 402                                  "%s.xsl", request_syntax);
 403                     for (cp = auto_stylesheet; *cp; cp++)
 404                     {
 405                         /* deliberately only consider ASCII */
 406                         if (*cp > 32 && *cp < 127)
 407                             *cp = tolower(*cp);
 408                     }
 409                     s = auto_stylesheet;
 410                 }
 411                 else
 412                 {
 413                     session_log(se, YLOG_WARN,
 414                                 "No pz:requestsyntax for auto stylesheet");
 415                 }
 416             }
 417             sdb->map = normalize_cache_get(se->normalize_cache,
 418                                            se->service, s);
 419             if (!sdb->map)
 420                 return -1;
 421         }
 422     }
 423     return 0;
 424 }
 425
 426 // called if watch should be removed because http_channel is to be destroyed
 427 static void session_watch_cancel(void *data, struct http_channel *c,
 428                                  void *data2)
 429 {
 430     struct session_watchentry *ent = data;
 431
 432     ent->fun = 0;
 433     ent->data = 0;
 434     ent->obs = 0;
 435 }
 436
 437 // set watch. Returns 0=OK, -1 if watch is already set
 438 int session_set_watch(struct session *s, int what,
 439                       session_watchfun fun, void *data,
 440                       struct http_channel *chan)
 441 {
 442     int ret;
 443     session_enter(s, "session_set_watch");
 444     if (s->watchlist[what].fun)
 445         ret = -1;
 446     else
 447     {
 448
 449         s->watchlist[what].fun = fun;
 450         s->watchlist[what].data = data;
 451         s->watchlist[what].obs = http_add_observer(chan, &s->watchlist[what],
 452                                                    session_watch_cancel);
 453         ret = 0;
 454     }
 455     session_leave(s, "session_set_watch");
 456     return ret;
 457 }
 458
 459 void session_alert_watch(struct session *s, int what)
 460 {
 461     assert(s);
 462     session_enter(s, "session_alert_watch");
 463     if (s->watchlist[what].fun)
 464     {
 465         /* our watch is no longer associated with http_channel */
 466         void *data;
 467         session_watchfun fun;
 468
 469         http_remove_observer(s->watchlist[what].obs);
 470         fun  = s->watchlist[what].fun;
 471         data = s->watchlist[what].data;
 472
 473         /* reset watch before fun is invoked - in case fun wants to set
 474            it again */
 475         s->watchlist[what].fun = 0;
 476         s->watchlist[what].data = 0;
 477         s->watchlist[what].obs = 0;
 478
 479         session_leave(s, "session_alert_watch");
 480         session_log(s, YLOG_DEBUG,
 481                     "Alert Watch: %d calling function: %p", what, fun);
 482         fun(data);
 483     }
 484     else
 485         session_leave(s,"session_alert_watch");
 486 }
 487
 488 //callback for grep_databases
 489 static void select_targets_callback(struct session *se,
 490                                     struct session_database *db)
 491 {
 492     struct client *cl;
 493     struct client_list *l;
 494
 495     for (l = se->clients_cached; l; l = l->next)
 496         if (client_get_database(l->client) == db)
 497             break;
 498
 499     if (l)
 500         cl = l->client;
 501     else
 502     {
 503         cl = client_create(db->database->id);
 504         client_set_database(cl, db);
 505
 506         l = xmalloc(sizeof(*l));
 507         l->client = cl;
 508         l->next = se->clients_cached;
 509         se->clients_cached = l;
 510     }
 511     client_set_session(cl, se);
 512
 513     l = xmalloc(sizeof(*l));
 514     l->client = cl;
 515     l->next = se->clients_active;
 516     se->clients_active = l;
 517 }
 518
 519 static void session_reset_active_clients(struct session *se,
 520                                          struct client_list *new_list)
 521 {
 522     struct client_list *l;
 523
 524     session_enter(se, "session_reset_active_clients");
 525     l = se->clients_active;
 526     se->clients_active = new_list;
 527     session_leave(se, "session_reset_active_clients");
 528
 529     while (l)
 530     {
 531         struct client_list *l_next = l->next;
 532
 533         client_lock(l->client);
 534         client_set_session(l->client, 0); /* mark client inactive */
 535         client_unlock(l->client);
 536
 537         xfree(l);
 538         l = l_next;
 539     }
 540 }
 541
 542 static void session_remove_cached_clients(struct session *se)
 543 {
 544     struct client_list *l;
 545
 546     session_reset_active_clients(se, 0);
 547
 548     session_enter(se, "session_remove_cached_clients");
 549     l = se->clients_cached;
 550     se->clients_cached = 0;
 551     session_leave(se, "session_remove_cached_clients");
 552
 553     while (l)
 554     {
 555         struct client_list *l_next = l->next;
 556         client_lock(l->client);
 557         client_set_session(l->client, 0);
 558         client_set_database(l->client, 0);
 559         client_unlock(l->client);
 560         client_destroy(l->client);
 561         xfree(l);
 562         l = l_next;
 563     }
 564 }
 565
 566 // Associates a set of clients with a session;
 567 // Note: Session-databases represent databases with per-session
 568 // setting overrides
 569 static int select_targets(struct session *se, const char *filter)
 570 {
 571     return session_grep_databases(se, filter, select_targets_callback);
 572 }
 573
 574 int session_active_clients(struct session *s)
 575 {
 576     struct client_list *l;
 577     int res = 0;
 578
 579     for (l = s->clients_active; l; l = l->next)
 580         if (client_is_active(l->client))
 581             res++;
 582
 583     return res;
 584 }
 585
 586 int session_is_preferred_clients_ready(struct session *s)
 587 {
 588     struct client_list *l;
 589     int res = 0;
 590
 591     for (l = s->clients_active; l; l = l->next)
 592         if (client_is_active_preferred(l->client))
 593             res++;
 594     session_log(s, YLOG_DEBUG, "Has %d active preferred clients.", res);
 595     return res == 0;
 596 }
 597
 598 static void session_clear_set(struct session *se, struct reclist_sortparms *sp)
 599 {
 600     reclist_destroy(se->reclist);
 601     if (nmem_total(se->nmem))
 602         session_log(se, YLOG_DEBUG, "NMEN operation usage %zd",
 603                     nmem_total(se->nmem));
 604     nmem_reset(se->nmem);
 605     se->total_records = se->total_merged = 0;
 606     se->num_termlists = 0;
 607
 608     relevance_clear(se->relevance);
 609
 610     /* reset list of sorted results and clear to relevance search */
 611     se->sorted_results = nmem_malloc(se->nmem, sizeof(*se->sorted_results));
 612     se->sorted_results->name = nmem_strdup(se->nmem, sp->name);
 613     se->sorted_results->increasing = sp->increasing;
 614     se->sorted_results->type = sp->type;
 615     se->sorted_results->next = 0;
 616
 617     session_log(se, YLOG_DEBUG, "clear_set session_sort: field=%s increasing=%d type=%d configured",
 618             sp->name, sp->increasing, sp->type);
 619
 620     se->reclist = reclist_create(se->nmem);
 621 }
 622
 623 static void session_sort_unlocked(struct session *se,
 624                                   struct reclist_sortparms *sp,
 625                                   const char *mergekey,
 626                                   const char *rank)
 627 {
 628     struct client_list *l;
 629     const char *field = sp->name;
 630     int increasing = sp->increasing;
 631     int type  = sp->type;
 632     int clients_research = 0;
 633
 634     session_log(se, YLOG_DEBUG, "session_sort field=%s increasing=%d type=%d",
 635                 field, increasing, type);
 636
 637     if (rank && (!se->rank || strcmp(se->rank, rank)))
 638     {
 639         /* new rank must research/reingest anyway */
 640         assert(rank);
 641         xfree(se->rank);
 642         se->rank = *rank ? xstrdup(rank) : 0;
 643         clients_research = 1;
 644         session_log(se, YLOG_DEBUG, "session_sort: new rank = %s",
 645                     rank);
 646     }
 647     if (mergekey && (!se->mergekey || strcmp(se->mergekey, mergekey)))
 648     {
 649         /* new mergekey must research/reingest anyway */
 650         assert(mergekey);
 651         xfree(se->mergekey);
 652         se->mergekey = *mergekey ? xstrdup(mergekey) : 0;
 653         clients_research = 1;
 654         session_log(se, YLOG_DEBUG, "session_sort: new mergekey = %s",
 655                     mergekey);
 656     }
 657     if (clients_research == 0)
 658     {
 659         struct reclist_sortparms *sr;
 660         for (sr = se->sorted_results; sr; sr = sr->next)
 661             if (!reclist_sortparms_cmp(sr, sp))
 662                 break;
 663         if (sr)
 664         {
 665             session_log(se, YLOG_DEBUG, "session_sort: field=%s increasing=%d type=%d already fetched",
 666                         field, increasing, type);
 667             return;
 668         }
 669     }
 670     session_log(se, YLOG_DEBUG, "session_sort: field=%s increasing=%d type=%d must fetch",
 671                 field, increasing, type);
 672
 673     // We need to reset reclist on every sort that changes the records, not just for position
 674     // So if just one client requires new searching, we need to clear set.
 675     // Ask each of the client if sorting requires re-search due to native sort
 676     // If it does it will require us to
 677     for (l = se->clients_active; l; l = l->next)
 678     {
 679         struct client *cl = l->client;
 680         // Assume no re-search is required.
 681         client_parse_init(cl, 1);
 682         clients_research += client_parse_sort(cl, sp);
 683     }
 684     if (clients_research)
 685     {
 686         session_log(se, YLOG_DEBUG,
 687                     "session_sort: reset results due to %d clients researching",
 688                     clients_research);
 689         session_clear_set(se, sp);
 690     }
 691     else
 692     {
 693         // A new sorting based on same record set
 694         struct reclist_sortparms *sr = nmem_malloc(se->nmem, sizeof(*sr));
 695         sr->name = nmem_strdup(se->nmem, field);
 696         sr->increasing = increasing;
 697         sr->type = type;
 698         sr->next = se->sorted_results;
 699         se->sorted_results = sr;
 700         session_log(se, YLOG_DEBUG, "session_sort: no research/ingesting done");
 701         return ;
 702     }
 703     session_log(se, YLOG_DEBUG, "Re- search/ingesting for clients due to change in sort order");
 704
 705     for (l = se->clients_active; l; l = l->next)
 706     {
 707         struct client *cl = l->client;
 708         if (client_get_state(cl) == Client_Connecting ||
 709             client_get_state(cl) == Client_Idle ||
 710             client_get_state(cl) == Client_Working) {
 711             client_start_search(cl);
 712         }
 713         else
 714         {
 715             session_log(se, YLOG_DEBUG,
 716                         "session_sort: %s: No re-start/ingest in show. "
 717                         "Wrong client state: %d",
 718                         client_get_id(cl), client_get_state(cl));
 719         }
 720     }
 721 }
 722
 723 void session_sort(struct session *se, struct reclist_sortparms *sp,
 724                   const char *mergekey, const char *rank)
 725 {
 726     //session_enter(se, "session_sort");
 727     session_sort_unlocked(se, sp, mergekey, rank);
 728     //session_leave(se, "session_sort");
 729 }
 730
 731
 732 enum pazpar2_error_code session_search(struct session *se,
 733                                        const char *query,
 734                                        const char *startrecs,
 735                                        const char *maxrecs,
 736                                        const char *filter,
 737                                        const char *limit,
 738                                        const char **addinfo,
 739                                        const char **addinfo2,
 740                                        struct reclist_sortparms *sp,
 741                                        const char *mergekey,
 742                                        const char *rank)
 743 {
 744     int live_channels = 0;
 745     int no_working = 0;
 746     int no_failed_query = 0;
 747     int no_failed_limit = 0;
 748     struct client_list *l, *l0;
 749
 750     session_alert_watch(se, SESSION_WATCH_SHOW);
 751     session_alert_watch(se, SESSION_WATCH_BYTARGET);
 752     session_alert_watch(se, SESSION_WATCH_TERMLIST);
 753     session_alert_watch(se, SESSION_WATCH_SHOW_PREF);
 754
 755     session_log(se, YLOG_DEBUG, "Search");
 756
 757     *addinfo = 0;
 758
 759     if (se->settings_modified) {
 760         session_remove_cached_clients(se);
 761     }
 762     else
 763         session_reset_active_clients(se, 0);
 764
 765     session_enter(se, "session_search");
 766     se->settings_modified = 0;
 767
 768     if (mergekey)
 769     {
 770         xfree(se->mergekey);
 771         se->mergekey = *mergekey ? xstrdup(mergekey) : 0;
 772     }
 773     if (rank)
 774     {
 775         xfree(se->rank);
 776         se->rank = *rank ? xstrdup(rank) : 0;
 777     }
 778
 779     session_clear_set(se, sp);
 780     relevance_destroy(&se->relevance);
 781
 782     live_channels = select_targets(se, filter);
 783     if (!live_channels)
 784     {
 785         session_leave(se, "session_search");
 786         return PAZPAR2_NO_TARGETS;
 787     }
 788
 789     facet_limits_destroy(se->facet_limits);
 790     se->facet_limits = facet_limits_create(limit);
 791     if (!se->facet_limits)
 792     {
 793         *addinfo = "limit";
 794         session_leave(se, "session_search");
 795         return PAZPAR2_MALFORMED_PARAMETER_VALUE;
 796     }
 797
 798     l0 = se->clients_active;
 799     se->clients_active = 0;
 800     session_leave(se, "session_search");
 801
 802     for (l = l0; l; l = l->next)
 803     {
 804         int parse_ret;
 805         struct client *cl = l->client;
 806         client_parse_init(cl, 1);
 807         if (prepare_map(se, client_get_database(cl)) < 0)
 808             continue;
 809
 810         parse_ret = client_parse_query(cl, query, se->facet_limits, addinfo2);
 811         if (parse_ret == -1)
 812             no_failed_query++;
 813         else if (parse_ret == -2)
 814             no_failed_limit++;
 815         else if (parse_ret < 0)
 816             no_working++; /* other error, such as bad CCL map */
 817         else
 818         {
 819             client_parse_range(cl, startrecs, maxrecs);
 820             client_parse_sort(cl, sp);
 821             client_start_search(cl);
 822             no_working++;
 823         }
 824     }
 825     session_reset_active_clients(se, l0);
 826
 827     if (no_working == 0)
 828     {
 829         if (no_failed_query > 0)
 830         {
 831             *addinfo = "query";
 832             return PAZPAR2_MALFORMED_PARAMETER_VALUE;
 833         }
 834         else if (no_failed_limit > 0)
 835         {
 836             *addinfo = "limit";
 837             return PAZPAR2_MALFORMED_PARAMETER_VALUE;
 838         }
 839         else
 840             return PAZPAR2_NO_TARGETS;
 841     }
 842     return PAZPAR2_NO_ERROR;
 843 }
 844
 845 // Creates a new session_database object for a database
 846 static void session_init_databases_fun(void *context, struct database *db)
 847 {
 848     struct session *se = (struct session *) context;
 849     struct session_database *new = nmem_malloc(se->session_nmem, sizeof(*new));
 850     int i;
 851
 852     new->database = db;
 853
 854     new->map = 0;
 855     assert(db->settings);
 856     new->settings = nmem_malloc(se->session_nmem,
 857                                 sizeof(struct settings *) * db->num_settings);
 858     new->num_settings = db->num_settings;
 859     for (i = 0; i < db->num_settings; i++)
 860     {
 861         struct setting *setting = db->settings[i];
 862         new->settings[i] = setting;
 863     }
 864     new->next = se->databases;
 865     se->databases = new;
 866 }
 867
 868 // Doesn't free memory associated with sdb -- nmem takes care of that
 869 static void session_database_destroy(struct session_database *sdb)
 870 {
 871     sdb->map = 0;
 872 }
 873
 874 // Initialize session_database list -- this represents this session's view
 875 // of the database list -- subject to modification by the settings ws command
 876 void session_init_databases(struct session *se)
 877 {
 878     se->databases = 0;
 879     predef_grep_databases(se, se->service, session_init_databases_fun);
 880 }
 881
 882 // Probably session_init_databases_fun should be refactored instead of
 883 // called here.
 884 static struct session_database *load_session_database(struct session *se,
 885                                                       char *id)
 886 {
 887     struct database *db = new_database_inherit_settings(id, se->session_nmem, se->service->settings);
 888     session_init_databases_fun((void*) se, db);
 889
 890     // New sdb is head of se->databases list
 891     return se->databases;
 892 }
 893
 894 // Find an existing session database. If not found, load it
 895 static struct session_database *find_session_database(struct session *se,
 896                                                       char *id)
 897 {
 898     struct session_database *sdb;
 899
 900     for (sdb = se->databases; sdb; sdb = sdb->next)
 901         if (!strcmp(sdb->database->id, id))
 902             return sdb;
 903     return load_session_database(se, id);
 904 }
 905
 906 // Apply a session override to a database
 907 void session_apply_setting(struct session *se, char *dbname, char *setting,
 908                            char *value)
 909 {
 910     struct session_database *sdb = find_session_database(se, dbname);
 911     struct conf_service *service = se->service;
 912     struct setting *new = nmem_malloc(se->session_nmem, sizeof(*new));
 913     int offset = settings_create_offset(service, setting);
 914
 915     expand_settings_array(&sdb->settings, &sdb->num_settings, offset,
 916                           se->session_nmem);
 917     new->precedence = 0;
 918     new->target = dbname;
 919     new->name = setting;
 920     new->value = value;
 921     new->next = sdb->settings[offset];
 922     sdb->settings[offset] = new;
 923
 924     se->settings_modified = 1;
 925
 926     // Force later recompute of settings-driven data structures
 927     // (happens when a search starts and client connections are prepared)
 928     switch (offset)
 929     {
 930     case PZ_XSLT:
 931         if (sdb->map)
 932         {
 933             sdb->map = 0;
 934         }
 935         break;
 936     }
 937 }
 938
 939 void session_destroy(struct session *se)
 940 {
 941     struct session_database *sdb;
 942     session_log(se, YLOG_LOG, "destroy");
 943     session_use(-1);
 944     session_remove_cached_clients(se);
 945
 946     for (sdb = se->databases; sdb; sdb = sdb->next)
 947         session_database_destroy(sdb);
 948     normalize_cache_destroy(se->normalize_cache);
 949     relevance_destroy(&se->relevance);
 950     reclist_destroy(se->reclist);
 951     xfree(se->mergekey);
 952     xfree(se->rank);
 953     if (nmem_total(se->nmem))
 954         session_log(se, YLOG_DEBUG, "NMEN operation usage %zd", nmem_total(se->nmem));
 955     if (nmem_total(se->session_nmem))
 956         session_log(se, YLOG_DEBUG, "NMEN session usage %zd", nmem_total(se->session_nmem));
 957     facet_limits_destroy(se->facet_limits);
 958     nmem_destroy(se->nmem);
 959     service_destroy(se->service);
 960     yaz_mutex_destroy(&se->session_mutex);
 961 }
 962
 963 size_t session_get_memory_status(struct session *session) {
 964     size_t session_nmem;
 965     if (session == 0)
 966         return 0;
 967     session_enter(session, "session_get_memory_status");
 968     session_nmem = nmem_total(session->nmem);
 969     session_leave(session, "session_get_memory_status");
 970     return session_nmem;
 971 }
 972
 973
 974 struct session *new_session(NMEM nmem, struct conf_service *service,
 975                             unsigned session_id)
 976 {
 977     int i;
 978     struct session *session = nmem_malloc(nmem, sizeof(*session));
 979
 980     char tmp_str[50];
 981
 982     sprintf(tmp_str, "session#%u", session_id);
 983
 984     session->session_id = session_id;
 985     session_log(session, YLOG_DEBUG, "New");
 986     session->service = service;
 987     session->relevance = 0;
 988     session->total_records = 0;
 989     session->number_of_warnings_unknown_elements = 0;
 990     session->number_of_warnings_unknown_metadata = 0;
 991     session->num_termlists = 0;
 992     session->reclist = reclist_create(nmem);
 993     session->clients_active = 0;
 994     session->clients_cached = 0;
 995     session->settings_modified = 0;
 996     session->session_nmem = nmem;
 997     session->nmem = nmem_create();
 998     session->databases = 0;
 999     session->sorted_results = 0;
1000     session->facet_limits = 0;
1001     session->mergekey = 0;
1002     session->rank = 0;
1003
1004     for (i = 0; i <= SESSION_WATCH_MAX; i++)
1005     {
1006         session->watchlist[i].data = 0;
1007         session->watchlist[i].fun = 0;
1008     }
1009     session->normalize_cache = normalize_cache_create();
1010     session->session_mutex = 0;
1011     pazpar2_mutex_create(&session->session_mutex, tmp_str);
1012     session_log(session, YLOG_LOG, "create");
1013
1014     session_use(1);
1015     return session;
1016 }
1017
1018 const char * client_get_suggestions_xml(struct client *cl, WRBUF wrbuf);
1019
1020 static struct hitsbytarget *hitsbytarget_nb(struct session *se,
1021                                             int *count, NMEM nmem)
1022 {
1023     struct hitsbytarget *res = 0;
1024     struct client_list *l;
1025     size_t sz = 0;
1026
1027     for (l = se->clients_active; l; l = l->next)
1028         sz++;
1029
1030     res = nmem_malloc(nmem, sizeof(*res) * sz);
1031     *count = 0;
1032     for (l = se->clients_active; l; l = l->next)
1033     {
1034         struct client *cl = l->client;
1035         WRBUF w = wrbuf_alloc();
1036         const char *name = session_setting_oneval(client_get_database(cl),
1037                                                   PZ_NAME);
1038
1039         res[*count].id = client_get_id(cl);
1040         res[*count].name = *name ? name : "Unknown";
1041         res[*count].hits = client_get_hits(cl);
1042         res[*count].approximation = client_get_approximation(cl);
1043         res[*count].records = client_get_num_records(cl);
1044         res[*count].filtered = client_get_num_records_filtered(cl);
1045         res[*count].diagnostic =
1046             client_get_diagnostic(cl, &res[*count].message,
1047                                   &res[*count].addinfo);
1048         res[*count].state = client_get_state_str(cl);
1049         res[*count].connected  = client_get_connection(cl) ? 1 : 0;
1050         session_settings_dump(se, client_get_database(cl), w);
1051         res[*count].settings_xml = nmem_strdup(nmem, wrbuf_cstr(w));
1052         wrbuf_rewind(w);
1053         wrbuf_puts(w, "");
1054         res[*count].suggestions_xml = nmem_strdup(nmem, client_get_suggestions_xml(cl, w));
1055         wrbuf_destroy(w);
1056         (*count)++;
1057     }
1058     return res;
1059 }
1060
1061 struct hitsbytarget *get_hitsbytarget(struct session *se, int *count, NMEM nmem)
1062 {
1063     struct hitsbytarget *p;
1064     session_enter(se, "get_hitsbytarget");
1065     p = hitsbytarget_nb(se, count, nmem);
1066     session_leave(se, "get_hitsbytarget");
1067     return p;
1068 }
1069
1070 // Compares two hitsbytarget nodes by hitcount
1071 static int cmp_ht(const void *p1, const void *p2)
1072 {
1073     const struct hitsbytarget *h1 = p1;
1074     const struct hitsbytarget *h2 = p2;
1075     return h2->hits - h1->hits;
1076 }
1077
1078 // Compares two hitsbytarget nodes by hitcount
1079 static int cmp_ht_approx(const void *p1, const void *p2)
1080 {
1081     const struct hitsbytarget *h1 = p1;
1082     const struct hitsbytarget *h2 = p2;
1083     return h2->approximation - h1->approximation;
1084 }
1085
1086 static int targets_termlist_nb(WRBUF wrbuf, struct session *se, int num,
1087                                NMEM nmem, int version)
1088 {
1089     struct hitsbytarget *ht;
1090     int count, i;
1091
1092     ht = hitsbytarget_nb(se, &count, nmem);
1093     if (version >= 2)
1094         qsort(ht, count, sizeof(struct hitsbytarget), cmp_ht_approx);
1095     else
1096         qsort(ht, count, sizeof(struct hitsbytarget), cmp_ht);
1097     for (i = 0; i < count && i < num && ht[i].hits > 0; i++)
1098     {
1099
1100         // do only print terms which have display names
1101
1102         wrbuf_puts(wrbuf, "<term>\n");
1103
1104         wrbuf_puts(wrbuf, "<id>");
1105         wrbuf_xmlputs(wrbuf, ht[i].id);
1106         wrbuf_puts(wrbuf, "</id>\n");
1107
1108         wrbuf_puts(wrbuf, "<name>");
1109         if (!ht[i].name || !ht[i].name[0])
1110             wrbuf_xmlputs(wrbuf, "NO TARGET NAME");
1111         else
1112             wrbuf_xmlputs(wrbuf, ht[i].name);
1113         wrbuf_puts(wrbuf, "</name>\n");
1114
1115         wrbuf_printf(wrbuf, "<frequency>" ODR_INT_PRINTF "</frequency>\n",
1116                      ht[i].hits);
1117
1118         if (version >= 2) {
1119             // Should not print if we know it isn't a approximation.
1120             wrbuf_printf(wrbuf, "<approximation>" ODR_INT_PRINTF "</approximation>\n", ht[i].approximation);
1121             wrbuf_printf(wrbuf, "<records>%d</records>\n", ht[i].records - ht[i].filtered);
1122             wrbuf_printf(wrbuf, "<filtered>%d</filtered>\n", ht[i].filtered);
1123         }
1124
1125         wrbuf_puts(wrbuf, "<state>");
1126         wrbuf_xmlputs(wrbuf, ht[i].state);
1127         wrbuf_puts(wrbuf, "</state>\n");
1128
1129         wrbuf_printf(wrbuf, "<diagnostic>%d</diagnostic>\n",
1130                      ht[i].diagnostic);
1131         wrbuf_puts(wrbuf, "</term>\n");
1132     }
1133     return count;
1134 }
1135
1136 void perform_termlist(struct http_channel *c, struct session *se,
1137                       const char *name, int num, int version)
1138 {
1139     int i, j;
1140     NMEM nmem_tmp = nmem_create();
1141     char **names;
1142     int num_names = 0;
1143
1144     if (!name)
1145         name = "*";
1146
1147     nmem_strsplit(nmem_tmp, ",", name, &names, &num_names);
1148
1149     session_enter(se, "perform_termlist");
1150
1151     for (j = 0; j < num_names; j++)
1152     {
1153         const char *tname;
1154         int must_generate_empty = 1; /* bug 5350 */
1155
1156         for (i = 0; i < se->num_termlists; i++)
1157         {
1158             tname = se->termlists[i].name;
1159             if (!strcmp(names[j], tname) || !strcmp(names[j], "*"))
1160             {
1161                 struct termlist_score **p = 0;
1162                 int len;
1163
1164                 wrbuf_puts(c->wrbuf, "<list name=\"");
1165                 wrbuf_xmlputs(c->wrbuf, tname);
1166                 wrbuf_puts(c->wrbuf, "\">\n");
1167                 must_generate_empty = 0;
1168
1169                 p = termlist_highscore(se->termlists[i].termlist, &len,
1170                                        nmem_tmp);
1171                 if (p)
1172                 {
1173                     int i;
1174                     for (i = 0; i < len && i < num; i++)
1175                     {
1176                         // prevent sending empty term elements
1177                         if (!p[i]->display_term || !p[i]->display_term[0])
1178                             continue;
1179
1180                         wrbuf_puts(c->wrbuf, "<term>");
1181                         wrbuf_puts(c->wrbuf, "<name>");
1182                         wrbuf_xmlputs(c->wrbuf, p[i]->display_term);
1183                         wrbuf_puts(c->wrbuf, "</name>");
1184
1185                         wrbuf_printf(c->wrbuf,
1186                                      "<frequency>%d</frequency>",
1187                                      p[i]->frequency);
1188                         wrbuf_puts(c->wrbuf, "</term>\n");
1189                     }
1190                 }
1191                 wrbuf_puts(c->wrbuf, "</list>\n");
1192             }
1193         }
1194         tname = "xtargets";
1195         if (!strcmp(names[j], tname) || !strcmp(names[j], "*"))
1196         {
1197             wrbuf_puts(c->wrbuf, "<list name=\"");
1198             wrbuf_xmlputs(c->wrbuf, tname);
1199             wrbuf_puts(c->wrbuf, "\">\n");
1200
1201             targets_termlist_nb(c->wrbuf, se, num, c->nmem, version);
1202             wrbuf_puts(c->wrbuf, "</list>\n");
1203             must_generate_empty = 0;
1204         }
1205         if (must_generate_empty)
1206         {
1207             wrbuf_puts(c->wrbuf, "<list name=\"");
1208             wrbuf_xmlputs(c->wrbuf, names[j]);
1209             wrbuf_puts(c->wrbuf, "\"/>\n");
1210         }
1211     }
1212     session_leave(se, "perform_termlist");
1213     nmem_destroy(nmem_tmp);
1214 }
1215
1216 #ifdef MISSING_HEADERS
1217 void report_nmem_stats(void)
1218 {
1219     size_t in_use, is_free;
1220
1221     nmem_get_memory_in_use(&in_use);
1222     nmem_get_memory_free(&is_free);
1223
1224     yaz_log(YLOG_LOG, "nmem stat: use=%ld free=%ld",
1225             (long) in_use, (long) is_free);
1226 }
1227 #endif
1228
1229 struct record_cluster *show_single_start(struct session *se, const char *id,
1230                                          struct record_cluster **prev_r,
1231                                          struct record_cluster **next_r)
1232 {
1233     struct record_cluster *r = 0;
1234
1235     session_enter(se, "show_single_start");
1236     *prev_r = 0;
1237     *next_r = 0;
1238     reclist_limit(se->reclist, se, 1);
1239
1240     reclist_enter(se->reclist);
1241     while ((r = reclist_read_record(se->reclist)))
1242     {
1243         if (!strcmp(r->recid, id))
1244         {
1245             *next_r = reclist_read_record(se->reclist);
1246             break;
1247         }
1248         *prev_r = r;
1249     }
1250     reclist_leave(se->reclist);
1251     if (!r)
1252         session_leave(se, "show_single_start");
1253     return r;
1254 }
1255
1256 void show_single_stop(struct session *se, struct record_cluster *rec)
1257 {
1258     session_leave(se, "show_single_stop");
1259 }
1260
1261
1262 int session_fetch_more(struct session *se)
1263 {
1264     struct client_list *l;
1265     int ret = 0;
1266
1267     for (l = se->clients_active; l; l = l->next)
1268     {
1269         struct client *cl = l->client;
1270         if (client_get_state(cl) == Client_Idle)
1271         {
1272             if (client_fetch_more(cl))
1273             {
1274                 session_log(se, YLOG_LOG, "%s: more to fetch",
1275                             client_get_id(cl));
1276                 ret = 1;
1277             }
1278             else
1279             {
1280                 session_log(se, YLOG_LOG, "%s: no more to fetch",
1281                             client_get_id(cl));
1282             }
1283         }
1284         else
1285         {
1286             session_log(se, YLOG_LOG, "%s: no fetch due to state=%s",
1287                         client_get_id(cl), client_get_state_str(cl));
1288         }
1289
1290     }
1291     return ret;
1292 }
1293
1294 struct record_cluster **show_range_start(struct session *se,
1295                                          struct reclist_sortparms *sp,
1296                                          int start, int *num, int *total,
1297                                          Odr_int *sumhits, Odr_int *approx_hits,
1298                                          void (*show_records_ready)(void *data),
1299                                          struct http_channel *chan)
1300 {
1301     struct record_cluster **recs = 0;
1302     struct reclist_sortparms *spp;
1303     struct client_list *l;
1304     int i;
1305 #if USE_TIMING
1306     yaz_timing_t t = yaz_timing_create();
1307 #endif
1308     session_enter(se, "show_range_start");
1309     *sumhits = 0;
1310     *approx_hits = 0;
1311     *total = 0;
1312     reclist_limit(se->reclist, se, 0);
1313     if (se->relevance)
1314     {
1315         for (spp = sp; spp; spp = spp->next)
1316             if (spp->type == Metadata_type_relevance)
1317             {
1318                 relevance_prepare_read(se->relevance, se->reclist);
1319                 break;
1320             }
1321         for (l = se->clients_active; l; l = l->next) {
1322             *sumhits += client_get_hits(l->client);
1323             *approx_hits += client_get_approximation(l->client);
1324         }
1325     }
1326     reclist_sort(se->reclist, sp);
1327
1328     reclist_enter(se->reclist);
1329     *total = reclist_get_num_records(se->reclist);
1330
1331     for (l = se->clients_active; l; l = l->next)
1332         client_update_show_stat(l->client, 0);
1333
1334     for (i = 0; i < start; i++)
1335     {
1336         struct record_cluster *r = reclist_read_record(se->reclist);
1337         if (!r)
1338         {
1339             *num = 0;
1340             break;
1341         }
1342         else
1343         {
1344             struct record *rec = r->records;
1345             for (;rec; rec = rec->next)
1346                 client_update_show_stat(rec->client, 1);
1347         }
1348     }
1349     recs = nmem_malloc(se->nmem, (*num > 0 ? *num : 1) * sizeof(*recs));
1350     for (i = 0; i < *num; i++)
1351     {
1352         struct record_cluster *r = reclist_read_record(se->reclist);
1353         if (!r)
1354         {
1355             *num = i;
1356             break;
1357         }
1358         else
1359         {
1360             struct record *rec = r->records;
1361             for (;rec; rec = rec->next)
1362                 client_update_show_stat(rec->client, 1);
1363             recs[i] = r;
1364         }
1365     }
1366     reclist_leave(se->reclist);
1367 #if USE_TIMING
1368     yaz_timing_stop(t);
1369     session_log(se, YLOG_LOG, "show %6.5f %3.2f %3.2f",
1370             yaz_timing_get_real(t), yaz_timing_get_user(t),
1371             yaz_timing_get_sys(t));
1372     yaz_timing_destroy(&t);
1373 #endif
1374
1375     if (!session_fetch_more(se))
1376         session_log(se, YLOG_LOG, "can not fetch more");
1377     else
1378     {
1379         show_range_stop(se, recs);
1380         session_log(se, YLOG_LOG, "fetching more in progress");
1381         if (session_set_watch(se, SESSION_WATCH_SHOW,
1382                               show_records_ready, chan, chan))
1383         {
1384             session_log(se, YLOG_WARN, "Ignoring show block");
1385             session_enter(se, "show_range_start");
1386         }
1387         else
1388         {
1389             session_log(se, YLOG_LOG, "session watch OK");
1390             return 0;
1391         }
1392     }
1393     return recs;
1394 }
1395
1396 void show_range_stop(struct session *se, struct record_cluster **recs)
1397 {
1398     session_leave(se, "show_range_stop");
1399 }
1400
1401 void statistics(struct session *se, struct statistics *stat)
1402 {
1403     struct client_list *l;
1404     int count = 0;
1405
1406     memset(stat, 0, sizeof(*stat));
1407     stat->num_hits = 0;
1408     for (l = se->clients_active; l; l = l->next)
1409     {
1410         struct client *cl = l->client;
1411         if (!client_get_connection(cl))
1412             stat->num_no_connection++;
1413         stat->num_hits += client_get_hits(cl);
1414         switch (client_get_state(cl))
1415         {
1416         case Client_Connecting: stat->num_connecting++; break;
1417         case Client_Working: stat->num_working++; break;
1418         case Client_Idle: stat->num_idle++; break;
1419         case Client_Failed: stat->num_failed++; break;
1420         case Client_Error: stat->num_error++; break;
1421         default: break;
1422         }
1423         count++;
1424     }
1425     stat->num_records = se->total_records;
1426
1427     stat->num_clients = count;
1428 }
1429
1430 static struct record_metadata *record_metadata_init(
1431     NMEM nmem, const char *value, enum conf_metadata_type type,
1432     struct _xmlAttr *attr)
1433 {
1434     struct record_metadata *rec_md = record_metadata_create(nmem);
1435     struct record_metadata_attr **attrp = &rec_md->attributes;
1436
1437     for (; attr; attr = attr->next)
1438     {
1439         if (attr->children && attr->children->content)
1440         {
1441             if (strcmp((const char *) attr->name, "type")
1442                 && strcmp((const char *) attr->name, "empty"))
1443             {  /* skip the "type" + "empty" attribute..
1444                   The "Type" is already part of the element in output
1445                   (md-%s) and so repeating it here is redundant */
1446                 *attrp = nmem_malloc(nmem, sizeof(**attrp));
1447                 (*attrp)->name =
1448                     nmem_strdup(nmem, (const char *) attr->name);
1449                 (*attrp)->value =
1450                     nmem_strdup(nmem, (const char *) attr->children->content);
1451                 attrp = &(*attrp)->next;
1452             }
1453         }
1454     }
1455     *attrp = 0;
1456
1457     switch (type)
1458     {
1459     case Metadata_type_generic:
1460     case Metadata_type_skiparticle:
1461         if (strstr(value, "://")) /* looks like a URL */
1462             rec_md->data.text.disp = nmem_strdup(nmem, value);
1463         else
1464             rec_md->data.text.disp =
1465                 normalize7bit_generic(nmem_strdup(nmem, value), " ,/.:([");
1466         rec_md->data.text.sort = 0;
1467         rec_md->data.text.snippet = 0;
1468         break;
1469     case Metadata_type_year:
1470     case Metadata_type_date:
1471     {
1472         int first, last;
1473         int longdate = 0;
1474
1475         if (type == Metadata_type_date)
1476             longdate = 1;
1477         if (extract7bit_dates((char *) value, &first, &last, longdate) < 0)
1478             return 0;
1479
1480         rec_md->data.number.min = first;
1481         rec_md->data.number.max = last;
1482     }
1483     break;
1484     case Metadata_type_float:
1485         rec_md->data.fnumber = atof(value);
1486         break;
1487     case Metadata_type_relevance:
1488     case Metadata_type_position:
1489         return 0;
1490     }
1491     return rec_md;
1492 }
1493
1494 static void mergekey_norm_wr(pp2_charset_fact_t charsets,
1495                              WRBUF norm_wr, const char *value)
1496 {
1497     const char *norm_str;
1498     pp2_charset_token_t prt =
1499         pp2_charset_token_create(charsets, "mergekey");
1500
1501     pp2_charset_token_first(prt, value, 0);
1502     while ((norm_str = pp2_charset_token_next(prt)))
1503     {
1504         if (*norm_str)
1505         {
1506             if (wrbuf_len(norm_wr))
1507                 wrbuf_puts(norm_wr, " ");
1508             wrbuf_puts(norm_wr, norm_str);
1509         }
1510     }
1511     pp2_charset_token_destroy(prt);
1512 }
1513
1514 static int get_mergekey_from_doc(xmlDoc *doc, xmlNode *root, const char *name,
1515                                  struct conf_service *service, WRBUF norm_wr)
1516 {
1517     xmlNode *n;
1518     int no_found = 0;
1519     for (n = root->children; n; n = n->next)
1520     {
1521         if (n->type != XML_ELEMENT_NODE)
1522             continue;
1523         if (!strcmp((const char *) n->name, "metadata"))
1524         {
1525             xmlChar *type = xmlGetProp(n, (xmlChar *) "type");
1526             if (type == NULL) {
1527                 yaz_log(YLOG_FATAL, "Missing type attribute on metadata element. Skipping!");
1528             }
1529             else if (!strcmp(name, (const char *) type))
1530             {
1531                 xmlChar *value = xmlNodeListGetString(doc, n->children, 1);
1532                 if (value && *value)
1533                 {
1534                     if (wrbuf_len(norm_wr) > 0)
1535                         wrbuf_puts(norm_wr, " ");
1536                     wrbuf_puts(norm_wr, name);
1537                     mergekey_norm_wr(service->charsets, norm_wr,
1538                                      (const char *) value);
1539                     no_found++;
1540                 }
1541                 if (value)
1542                     xmlFree(value);
1543             }
1544             xmlFree(type);
1545         }
1546     }
1547     return no_found;
1548 }
1549
1550 static const char *get_mergekey(xmlDoc *doc, xmlNode *root,
1551                                 struct client *cl, int record_no,
1552                                 struct conf_service *service, NMEM nmem,
1553                                 const char *session_mergekey)
1554 {
1555     char *mergekey_norm = 0;
1556     WRBUF norm_wr = wrbuf_alloc();
1557     xmlChar *mergekey;
1558
1559     if (session_mergekey)
1560     {
1561         int i, num = 0;
1562         char **values = 0;
1563         nmem_strsplit_escape2(nmem, ",", session_mergekey, &values,
1564                               &num, 1, '\\', 1);
1565
1566         for (i = 0; i < num; i++)
1567             get_mergekey_from_doc(doc, root, values[i], service, norm_wr);
1568     }
1569     else if ((mergekey = xmlGetProp(root, (xmlChar *) "mergekey")))
1570     {
1571         mergekey_norm_wr(service->charsets, norm_wr, (const char *) mergekey);
1572         xmlFree(mergekey);
1573     }
1574     else
1575     {
1576         /* no mergekey defined in XSL. Look for mergekey metadata instead */
1577         int field_id;
1578         for (field_id = 0; field_id < service->num_metadata; field_id++)
1579         {
1580             struct conf_metadata *ser_md = &service->metadata[field_id];
1581             if (ser_md->mergekey != Metadata_mergekey_no)
1582             {
1583                 int r = get_mergekey_from_doc(doc, root, ser_md->name,
1584                                               service, norm_wr);
1585                 if (r == 0 && ser_md->mergekey == Metadata_mergekey_required)
1586                 {
1587                     /* no mergekey on this one and it is required..
1588                        Generate unique key instead */
1589                     wrbuf_rewind(norm_wr);
1590                     break;
1591                 }
1592             }
1593         }
1594     }
1595
1596     /* generate unique key if none is not generated already or is empty */
1597     if (wrbuf_len(norm_wr) == 0)
1598     {
1599         wrbuf_printf(norm_wr, "position: %s-%d",
1600                      client_get_id(cl), record_no);
1601     }
1602     else
1603     {
1604         const char *lead = "content: ";
1605         wrbuf_insert(norm_wr, 0, lead, strlen(lead));
1606     }
1607     if (wrbuf_len(norm_wr) > 0)
1608         mergekey_norm = nmem_strdup(nmem, wrbuf_cstr(norm_wr));
1609     wrbuf_destroy(norm_wr);
1610     return mergekey_norm;
1611 }
1612
1613 /** \brief see if metadata for pz:recordfilter exists
1614     \param root xml root element of normalized record
1615     \param sdb session database for client
1616     \retval 0 if there is no metadata for pz:recordfilter
1617     \retval 1 if there is metadata for pz:recordfilter
1618
1619     If there is no pz:recordfilter defined, this function returns 1
1620     as well.
1621 */
1622
1623 static int check_record_filter(xmlNode *root, struct session_database *sdb)
1624 {
1625     int match = 0;
1626     xmlNode *n;
1627     const char *s;
1628     s = session_setting_oneval(sdb, PZ_RECORDFILTER);
1629
1630     if (!s || !*s)
1631         return 1;
1632
1633     for (n = root->children; n; n = n->next)
1634     {
1635         if (n->type != XML_ELEMENT_NODE)
1636             continue;
1637         if (!strcmp((const char *) n->name, "metadata"))
1638         {
1639             xmlChar *type = xmlGetProp(n, (xmlChar *) "type");
1640             if (type)
1641             {
1642                 size_t len;
1643                 int substring;
1644                 const char *eq;
1645
1646                 if ((eq = strchr(s, '=')))
1647                     substring = 0;
1648                 else if ((eq = strchr(s, '~')))
1649                     substring = 1;
1650                 if (eq)
1651                     len = eq - s;
1652                 else
1653                     len = strlen(s);
1654                 if (len == strlen((const char *)type) &&
1655                     !memcmp((const char *) type, s, len))
1656                 {
1657                     xmlChar *value = xmlNodeGetContent(n);
1658                     if (value && *value)
1659                     {
1660                         if (!eq ||
1661                             (substring && strstr((const char *) value, eq+1)) ||
1662                             (!substring && !strcmp((const char *) value, eq + 1)))
1663                             match = 1;
1664                     }
1665                     xmlFree(value);
1666                 }
1667                 xmlFree(type);
1668             }
1669         }
1670     }
1671     return match;
1672 }
1673
1674 static int ingest_to_cluster(struct client *cl,
1675                              xmlDoc *xdoc,
1676                              xmlNode *root,
1677                              int record_no,
1678                              struct record_metadata_attr *mergekey);
1679
1680 static int ingest_sub_record(struct client *cl, xmlDoc *xdoc, xmlNode *root,
1681                              int record_no, NMEM nmem,
1682                              struct session_database *sdb,
1683                              struct record_metadata_attr *mergekeys)
1684 {
1685     int ret = 0;
1686     struct session *se = client_get_session(cl);
1687     struct conf_service *service = se->service;
1688
1689     insert_settings_values(sdb, xdoc, root, service);
1690
1691     if (!check_record_filter(root, sdb))
1692     {
1693         session_log(se, YLOG_LOG,
1694                     "Filtered out record no %d from %s",
1695                     record_no, sdb->database->id);
1696         return 0;
1697     }
1698     session_enter(se, "ingest_sub_record");
1699     if (client_get_session(cl) == se && se->relevance)
1700         ret = ingest_to_cluster(cl, xdoc, root, record_no, mergekeys);
1701     session_leave(se, "ingest_sub_record");
1702
1703     return ret;
1704 }
1705
1706 /** \brief ingest XML record
1707     \param cl client holds the result set for record
1708     \param rec record buffer (0 terminated)
1709     \param record_no record position (1, 2, ..)
1710     \param nmem working NMEM
1711     \retval 0 OK
1712     \retval -1 failure
1713     \retval -2 Filtered
1714 */
1715 int ingest_record(struct client *cl, const char *rec,
1716                   int record_no, NMEM nmem)
1717 {
1718     struct session *se = client_get_session(cl);
1719     struct session_database *sdb = client_get_database(cl);
1720     struct conf_service *service = se->service;
1721     xmlDoc *xdoc = normalize_record(se, sdb, service, rec, nmem);
1722     int r = 0;
1723     xmlNode *root;
1724
1725     if (!xdoc)
1726         return -1;
1727
1728     if (global_parameters.dump_records)
1729     {
1730         session_log(se, YLOG_LOG, "Normalized record from %s",
1731                     sdb->database->id);
1732         log_xml_doc(xdoc);
1733     }
1734
1735     root = xmlDocGetRootElement(xdoc);
1736
1737     if (!strcmp((const char *) root->name, "cluster"))
1738     {
1739         int no_merge_keys = 0;
1740         int no_merge_dups = 0;
1741         xmlNode *sroot;
1742         struct record_metadata_attr *mk = 0;
1743
1744         for (sroot = root->children; sroot; sroot = sroot->next)
1745             if (sroot->type == XML_ELEMENT_NODE &&
1746                 !strcmp((const char *) sroot->name, "record"))
1747             {
1748                 struct record_metadata_attr **mkp;
1749                 const char *mergekey_norm =
1750                     get_mergekey(xdoc, sroot, cl, record_no, service, nmem,
1751                                  se->mergekey);
1752                 if (!mergekey_norm)
1753                 {
1754                     r = -1;
1755                     break;
1756                 }
1757                 for (mkp = &mk; *mkp; mkp = &(*mkp)->next)
1758                     if (!strcmp((*mkp)->value, mergekey_norm))
1759                         break;
1760                 if (!*mkp)
1761                 {
1762                     *mkp = (struct record_metadata_attr*)
1763                         nmem_malloc(nmem, sizeof(**mkp));
1764                     (*mkp)->name = 0;
1765                     (*mkp)->value = nmem_strdup(nmem, mergekey_norm);
1766                     (*mkp)->next = 0;
1767                     no_merge_keys++;
1768                 }
1769                 else
1770                     no_merge_dups++;
1771             }
1772         if (no_merge_keys > 1 || no_merge_dups > 0)
1773         {
1774             yaz_log(YLOG_LOG, "Got %d mergekeys, %d dups for position %d",
1775                     no_merge_keys, no_merge_dups, record_no);
1776         }
1777         for (sroot = root->children; !r && sroot; sroot = sroot->next)
1778             if (sroot->type == XML_ELEMENT_NODE &&
1779                 !strcmp((const char *) sroot->name, "record"))
1780             {
1781                 r = ingest_sub_record(cl, xdoc, sroot, record_no, nmem, sdb,
1782                                       mk);
1783             }
1784     }
1785     else if (!strcmp((const char *) root->name, "record"))
1786     {
1787         const char *mergekey_norm =
1788             get_mergekey(xdoc, root, cl, record_no, service, nmem,
1789                          se->mergekey);
1790         if (mergekey_norm)
1791         {
1792             struct record_metadata_attr *mk = (struct record_metadata_attr*)
1793                 nmem_malloc(nmem, sizeof(*mk));
1794             mk->name = 0;
1795             mk->value = nmem_strdup(nmem, mergekey_norm);
1796             mk->next = 0;
1797
1798             r = ingest_sub_record(cl, xdoc, root, record_no, nmem, sdb, mk);
1799         }
1800     }
1801     else
1802     {
1803         session_log(se, YLOG_WARN, "Bad pz root element: %s",
1804                     (const char *) root->name);
1805         r = -1;
1806     }
1807     xmlFreeDoc(xdoc);
1808     return r;
1809 }
1810
1811
1812 //    struct conf_metadata *ser_md = &service->metadata[md_field_id];
1813 //    struct record_metadata *rec_md = record->metadata[md_field_id];
1814 static int match_metadata_local(struct conf_service *service,
1815                                 struct conf_metadata *ser_md,
1816                                 struct record_metadata *rec_md0,
1817                                 char **values, int num_v)
1818 {
1819     int i;
1820     struct record_metadata *rec_md = rec_md0;
1821     WRBUF val_wr = 0;
1822     WRBUF text_wr = wrbuf_alloc();
1823     for (i = 0; i < num_v; )
1824     {
1825         if (rec_md)
1826         {
1827             if (ser_md->type == Metadata_type_year
1828                 || ser_md->type == Metadata_type_date)
1829             {
1830                 int y = atoi(values[i]);
1831                 if (y >= rec_md->data.number.min
1832                     && y <= rec_md->data.number.max)
1833                     break;
1834             }
1835             else
1836             {
1837                 if (!val_wr)
1838                 {
1839                     val_wr = wrbuf_alloc();
1840                     mergekey_norm_wr(service->charsets, val_wr, values[i]);
1841                 }
1842                 wrbuf_rewind(text_wr);
1843                 mergekey_norm_wr(service->charsets, text_wr,
1844                                  rec_md->data.text.disp);
1845                 if (!strcmp(wrbuf_cstr(val_wr), wrbuf_cstr(text_wr)))
1846                     break;
1847             }
1848             rec_md = rec_md->next;
1849         }
1850         else
1851         {
1852             rec_md = rec_md0;
1853             wrbuf_destroy(val_wr);
1854             val_wr = 0;
1855             i++;
1856         }
1857     }
1858     wrbuf_destroy(val_wr);
1859     wrbuf_destroy(text_wr);
1860     return i < num_v ? 1 : 0;
1861 }
1862
1863 int session_check_cluster_limit(struct session *se, struct record_cluster *rec)
1864 {
1865     int i;
1866     struct conf_service *service = se->service;
1867     int ret = 1;
1868     const char *name;
1869     const char *value;
1870     NMEM nmem_tmp = nmem_create();
1871
1872     for (i = 0; (name = facet_limits_get(se->facet_limits, i, &value)); i++)
1873     {
1874         int j;
1875         for (j = 0; j < service->num_metadata; j++)
1876         {
1877             struct conf_metadata *md = service->metadata + j;
1878             if (!strcmp(md->name, name) && md->limitcluster)
1879             {
1880                 char **values = 0;
1881                 int num = 0;
1882                 int md_field_id =
1883                     conf_service_metadata_field_id(service,
1884                                                    md->limitcluster);
1885
1886                 if (md_field_id < 0)
1887                 {
1888                     ret = 0;
1889                     break;
1890                 }
1891
1892                 nmem_strsplit_escape2(nmem_tmp, "|", value, &values,
1893                                       &num, 1, '\\', 1);
1894
1895                 if (!match_metadata_local(service,
1896                                           &service->metadata[md_field_id],
1897                                           rec->metadata[md_field_id],
1898                                           values, num))
1899                 {
1900                     ret = 0;
1901                     break;
1902                 }
1903             }
1904         }
1905     }
1906     nmem_destroy(nmem_tmp);
1907     return ret;
1908 }
1909
1910 // Skip record on non-zero
1911 static int check_limit_local(struct client *cl,
1912                              struct record *record,
1913                              int record_no)
1914 {
1915     int skip_record = 0;
1916     struct session *se = client_get_session(cl);
1917     struct conf_service *service = se->service;
1918     NMEM nmem_tmp = nmem_create();
1919     struct session_database *sdb = client_get_database(cl);
1920     int l = 0;
1921     while (!skip_record)
1922     {
1923         int md_field_id;
1924         char **values = 0;
1925         int num_v = 0;
1926         const char *name =
1927             client_get_facet_limit_local(cl, sdb, &l, nmem_tmp,
1928                                          &num_v, &values);
1929         if (!name)
1930             break;
1931
1932         if (!strcmp(name, "*"))
1933         {
1934             for (md_field_id = 0; md_field_id < service->num_metadata;
1935                  md_field_id++)
1936             {
1937                 if (match_metadata_local(
1938                         service,
1939                         &service->metadata[md_field_id],
1940                         record->metadata[md_field_id],
1941                         values, num_v))
1942                     break;
1943             }
1944             if (md_field_id == service->num_metadata)
1945                 skip_record = 1;
1946         }
1947         else
1948         {
1949             md_field_id = conf_service_metadata_field_id(service, name);
1950             if (md_field_id < 0)
1951             {
1952                 skip_record = 1;
1953                 break;
1954             }
1955             if (!match_metadata_local(
1956                     service,
1957                     &service->metadata[md_field_id],
1958                     record->metadata[md_field_id],
1959                     values, num_v))
1960             {
1961                 skip_record = 1;
1962             }
1963         }
1964     }
1965     nmem_destroy(nmem_tmp);
1966     return skip_record;
1967 }
1968
1969 static int ingest_to_cluster(struct client *cl,
1970                              xmlDoc *xdoc,
1971                              xmlNode *root,
1972                              int record_no,
1973                              struct record_metadata_attr *merge_keys)
1974 {
1975     xmlNode *n;
1976     xmlChar *type = 0;
1977     xmlChar *value = 0;
1978     struct session *se = client_get_session(cl);
1979     struct conf_service *service = se->service;
1980     int term_factor = 1;
1981     struct record_cluster *cluster;
1982     struct record_metadata **metadata0;
1983     struct session_database *sdb = client_get_database(cl);
1984     NMEM ingest_nmem = 0;
1985     char **rank_values = 0;
1986     int rank_num = 0;
1987     struct record *record = record_create(se->nmem,
1988                                           service->num_metadata,
1989                                           service->num_sortkeys, cl,
1990                                           record_no);
1991
1992     for (n = root->children; n; n = n->next)
1993     {
1994         if (type)
1995             xmlFree(type);
1996         if (value)
1997             xmlFree(value);
1998         type = value = 0;
1999
2000         if (n->type != XML_ELEMENT_NODE)
2001             continue;
2002         if (!strcmp((const char *) n->name, "metadata"))
2003         {
2004             struct conf_metadata *ser_md = 0;
2005             struct record_metadata **wheretoput = 0;
2006             struct record_metadata *rec_md = 0;
2007             int md_field_id = -1;
2008
2009             type = xmlGetProp(n, (xmlChar *) "type");
2010             value = xmlNodeListGetString(xdoc, n->children, 1);
2011             if (!type)
2012                 continue;
2013             if (!value || !*value)
2014             {
2015                 xmlChar *empty = xmlGetProp(n, (xmlChar *) "empty");
2016                 if (!empty)
2017                     continue;
2018                 if (value)
2019                     xmlFree(value);
2020                 value = empty;
2021             }
2022             md_field_id
2023                 = conf_service_metadata_field_id(service, (const char *) type);
2024             if (md_field_id < 0)
2025             {
2026                 if (se->number_of_warnings_unknown_metadata == 0)
2027                 {
2028                     session_log(se, YLOG_WARN,
2029                             "Ignoring unknown metadata element: %s", type);
2030                 }
2031                 se->number_of_warnings_unknown_metadata++;
2032                 continue;
2033             }
2034
2035             ser_md = &service->metadata[md_field_id];
2036
2037             // non-merged metadata
2038             rec_md = record_metadata_init(se->nmem, (const char *) value,
2039                                           ser_md->type, n->properties);
2040             if (!rec_md)
2041             {
2042                 session_log(se, YLOG_WARN, "bad metadata data '%s' "
2043                             "for element '%s'", value, type);
2044                 continue;
2045             }
2046
2047             if (ser_md->type == Metadata_type_generic)
2048             {
2049                 WRBUF w = wrbuf_alloc();
2050                 if (relevance_snippet(se->relevance,
2051                                       (char*) value, ser_md->name, w))
2052                     rec_md->data.text.snippet = nmem_strdup(se->nmem,
2053                                                             wrbuf_cstr(w));
2054                 wrbuf_destroy(w);
2055             }
2056
2057
2058             wheretoput = &record->metadata[md_field_id];
2059             while (*wheretoput)
2060                 wheretoput = &(*wheretoput)->next;
2061             *wheretoput = rec_md;
2062         }
2063     }
2064
2065     if (check_limit_local(cl, record, record_no))
2066     {
2067         session_log(se, YLOG_LOG, "Facet filtered out record no %d from %s",
2068                     record_no, sdb->database->id);
2069         if (type)
2070             xmlFree(type);
2071         if (value)
2072             xmlFree(value);
2073         return -2;
2074     }
2075     cluster = reclist_insert(se->reclist, se->relevance, service, record,
2076                              merge_keys, &se->total_merged);
2077     if (!cluster)
2078     {
2079         if (type)
2080             xmlFree(type);
2081         if (value)
2082             xmlFree(value);
2083         return 0; // complete match with existing record
2084     }
2085
2086     {
2087         const char *use_term_factor_str =
2088             session_setting_oneval(sdb, PZ_TERMLIST_TERM_FACTOR);
2089         if (use_term_factor_str && use_term_factor_str[0] == '1')
2090         {
2091             int maxrecs = client_get_maxrecs(cl);
2092             int hits = (int) client_get_hits(cl);
2093             term_factor = MAX(hits, maxrecs) /  MAX(1, maxrecs);
2094             assert(term_factor >= 1);
2095             session_log(se, YLOG_DEBUG, "Using term factor: %d (%d / %d)",
2096                         term_factor, MAX(hits, maxrecs), MAX(1, maxrecs));
2097         }
2098     }
2099
2100     if (global_parameters.dump_records)
2101         session_log(se, YLOG_LOG, "Cluster id %s from %s (#%d)", cluster->recid,
2102                     sdb->database->id, record_no);
2103
2104     // original metadata, to check if first existence of a field
2105     metadata0 = xmalloc(sizeof(*metadata0) * service->num_metadata);
2106     memcpy(metadata0, cluster->metadata,
2107            sizeof(*metadata0) * service->num_metadata);
2108
2109     ingest_nmem = nmem_create();
2110     if (se->rank)
2111     {
2112         yaz_log(YLOG_LOG, "local in sort : %s", se->rank);
2113         nmem_strsplit_escape2(ingest_nmem, ",", se->rank, &rank_values,
2114                               &rank_num, 1, '\\', 1);
2115     }
2116
2117     // now parsing XML record and adding data to cluster or record metadata
2118     for (n = root->children; n; n = n->next)
2119     {
2120         pp2_charset_token_t prt;
2121         if (type)
2122             xmlFree(type);
2123         if (value)
2124             xmlFree(value);
2125         type = value = 0;
2126
2127         if (n->type != XML_ELEMENT_NODE)
2128             continue;
2129         if (!strcmp((const char *) n->name, "metadata"))
2130         {
2131             struct conf_metadata *ser_md = 0;
2132             struct conf_sortkey *ser_sk = 0;
2133             struct record_metadata **wheretoput = 0;
2134             struct record_metadata *rec_md = 0;
2135             int md_field_id = -1;
2136             int sk_field_id = -1;
2137             const char *rank = 0;
2138             xmlChar *xml_rank = 0;
2139
2140             type = xmlGetProp(n, (xmlChar *) "type");
2141             value = xmlNodeListGetString(xdoc, n->children, 1);
2142
2143             if (!type || !value || !*value)
2144                 continue;
2145
2146             md_field_id
2147                 = conf_service_metadata_field_id(service, (const char *) type);
2148             if (md_field_id < 0)
2149                 continue;
2150
2151             ser_md = &service->metadata[md_field_id];
2152
2153             if (ser_md->sortkey_offset >= 0)
2154             {
2155                 sk_field_id = ser_md->sortkey_offset;
2156                 ser_sk = &service->sortkeys[sk_field_id];
2157             }
2158
2159             // merged metadata
2160             rec_md = record_metadata_init(se->nmem, (const char *) value,
2161                                           ser_md->type, 0);
2162
2163             // see if the field was not in cluster already (from beginning)
2164
2165             if (!rec_md)
2166                 continue;
2167
2168             if (rank_num)
2169             {
2170                 int i;
2171                 for (i = 0; i < rank_num; i++)
2172                 {
2173                     const char *val = rank_values[i];
2174                     const char *cp = strchr(val, '=');
2175                     if (!cp)
2176                         continue;
2177                     if ((cp - val) == strlen((const char *) type)
2178                         && !memcmp(val, type, cp - val))
2179                     {
2180                         rank = cp + 1;
2181                         break;
2182                     }
2183                 }
2184             }
2185             else
2186             {
2187                 xml_rank = xmlGetProp(n, (xmlChar *) "rank");
2188                 rank = xml_rank ? (const char *) xml_rank : ser_md->rank;
2189             }
2190
2191             wheretoput = &cluster->metadata[md_field_id];
2192
2193             if (ser_md->merge == Metadata_merge_first)
2194             {
2195                 if (!metadata0[md_field_id])
2196                 {
2197                     while (*wheretoput)
2198                         wheretoput = &(*wheretoput)->next;
2199                     *wheretoput = rec_md;
2200                 }
2201             }
2202             else if (ser_md->merge == Metadata_merge_unique)
2203             {
2204                 while (*wheretoput)
2205                 {
2206                     if (!strcmp((const char *) (*wheretoput)->data.text.disp,
2207                                 rec_md->data.text.disp))
2208                         break;
2209                     wheretoput = &(*wheretoput)->next;
2210                 }
2211                 if (!*wheretoput)
2212                     *wheretoput = rec_md;
2213             }
2214             else if (ser_md->merge == Metadata_merge_longest)
2215             {
2216                 if (!*wheretoput
2217                     || strlen(rec_md->data.text.disp)
2218                     > strlen((*wheretoput)->data.text.disp))
2219                 {
2220                     *wheretoput = rec_md;
2221                     if (ser_sk)
2222                     {
2223                         const char *sort_str = 0;
2224                         int skip_article =
2225                             ser_sk->type == Metadata_type_skiparticle;
2226
2227                         if (!cluster->sortkeys[sk_field_id])
2228                             cluster->sortkeys[sk_field_id] =
2229                                 nmem_malloc(se->nmem,
2230                                             sizeof(union data_types));
2231
2232                         prt =
2233                             pp2_charset_token_create(service->charsets, "sort");
2234
2235                         pp2_charset_token_first(prt, rec_md->data.text.disp,
2236                                                 skip_article);
2237
2238                         pp2_charset_token_next(prt);
2239
2240                         sort_str = pp2_get_sort(prt);
2241
2242                         cluster->sortkeys[sk_field_id]->text.disp =
2243                             rec_md->data.text.disp;
2244                         if (!sort_str)
2245                         {
2246                             sort_str = rec_md->data.text.disp;
2247                             session_log(se, YLOG_WARN,
2248                                     "Could not make sortkey. Bug #1858");
2249                         }
2250                         cluster->sortkeys[sk_field_id]->text.sort =
2251                             nmem_strdup(se->nmem, sort_str);
2252                         pp2_charset_token_destroy(prt);
2253                     }
2254                 }
2255             }
2256             else if (ser_md->merge == Metadata_merge_all)
2257             {
2258                 while (*wheretoput)
2259                     wheretoput = &(*wheretoput)->next;
2260                 *wheretoput = rec_md;
2261             }
2262             else if (ser_md->merge == Metadata_merge_range)
2263             {
2264                 if (!*wheretoput)
2265                 {
2266                     *wheretoput = rec_md;
2267                     if (ser_sk)
2268                         cluster->sortkeys[sk_field_id]
2269                             = &rec_md->data;
2270                 }
2271                 else
2272                 {
2273                     int this_min = rec_md->data.number.min;
2274                     int this_max = rec_md->data.number.max;
2275                     if (this_min < (*wheretoput)->data.number.min)
2276                         (*wheretoput)->data.number.min = this_min;
2277                     if (this_max > (*wheretoput)->data.number.max)
2278                         (*wheretoput)->data.number.max = this_max;
2279                 }
2280             }
2281
2282             // ranking of _all_ fields enabled ...
2283             if (rank)
2284             {
2285                 relevance_countwords(se->relevance, cluster,
2286                                      (char *) value, rank, ser_md->name);
2287             }
2288             // construct facets ... unless the client already has reported them
2289             if (ser_md->termlist && !client_has_facet(cl, (char *) type))
2290             {
2291                 if (ser_md->type == Metadata_type_year)
2292                 {
2293                     char year[64];
2294                     sprintf(year, "%d", rec_md->data.number.max);
2295
2296                     add_facet(se, (char *) type, year, term_factor);
2297                     if (rec_md->data.number.max != rec_md->data.number.min)
2298                     {
2299                         sprintf(year, "%d", rec_md->data.number.min);
2300                         add_facet(se, (char *) type, year, term_factor);
2301                     }
2302                 }
2303                 else
2304                     add_facet(se, (char *) type, (char *) value, term_factor);
2305             }
2306
2307             // cleaning up
2308             if (xml_rank)
2309                 xmlFree(xml_rank);
2310             xmlFree(type);
2311             xmlFree(value);
2312             type = value = 0;
2313         }
2314         else
2315         {
2316             if (se->number_of_warnings_unknown_elements == 0)
2317                 session_log(se, YLOG_WARN,
2318                         "Unexpected element in internal record: %s", n->name);
2319             se->number_of_warnings_unknown_elements++;
2320         }
2321     }
2322     if (type)
2323         xmlFree(type);
2324     if (value)
2325         xmlFree(value);
2326
2327     nmem_destroy(ingest_nmem);
2328     xfree(metadata0);
2329     relevance_donerecord(se->relevance, cluster);
2330     se->total_records++;
2331
2332     return 0;
2333 }
2334
2335 void session_log(struct session *s, int level, const char *fmt, ...)
2336 {
2337     char buf[1024];
2338     va_list ap;
2339     va_start(ap, fmt);
2340
2341     yaz_vsnprintf(buf, sizeof(buf)-30, fmt, ap);
2342     yaz_log(level, "Session %u: %s", s ? s->session_id : 0, buf);
2343
2344     va_end(ap);
2345 }
2346
2347 /*
2348  * Local variables:
2349  * c-basic-offset: 4
2350  * c-file-style: "Stroustrup"
2351  * indent-tabs-mode: nil
2352  * End:
2353  * vim: shiftwidth=4 tabstop=8 expandtab
2354  */
2355