src/session.c

   1 /* This file is part of Pazpar2.
   2    Copyright (C) 2006-2013 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     /* set session always. If may be 0 if client is not active */
 512     client_set_session(cl, se);
 513
 514     l = xmalloc(sizeof(*l));
 515     l->client = cl;
 516     l->next = se->clients_active;
 517     se->clients_active = l;
 518 }
 519
 520 static void session_reset_active_clients(struct session *se,
 521                                          struct client_list *new_list)
 522 {
 523     struct client_list *l;
 524
 525     session_enter(se, "session_reset_active_clients");
 526     l = se->clients_active;
 527     se->clients_active = new_list;
 528     session_leave(se, "session_reset_active_clients");
 529
 530     while (l)
 531     {
 532         struct client_list *l_next = l->next;
 533
 534         client_lock(l->client);
 535         client_set_session(l->client, 0); /* mark client inactive */
 536         client_unlock(l->client);
 537
 538         xfree(l);
 539         l = l_next;
 540     }
 541 }
 542
 543 static void session_remove_cached_clients(struct session *se)
 544 {
 545     struct client_list *l;
 546
 547     session_reset_active_clients(se, 0);
 548
 549     session_enter(se, "session_remove_cached_clients");
 550     l = se->clients_cached;
 551     se->clients_cached = 0;
 552     session_leave(se, "session_remove_cached_clients");
 553
 554     while (l)
 555     {
 556         struct client_list *l_next = l->next;
 557         client_lock(l->client);
 558         client_set_session(l->client, 0);
 559         client_set_database(l->client, 0);
 560         client_unlock(l->client);
 561         client_destroy(l->client);
 562         xfree(l);
 563         l = l_next;
 564     }
 565 }
 566
 567 // Associates a set of clients with a session;
 568 // Note: Session-databases represent databases with per-session
 569 // setting overrides
 570 static int select_targets(struct session *se, const char *filter)
 571 {
 572     return session_grep_databases(se, filter, select_targets_callback);
 573 }
 574
 575 int session_active_clients(struct session *s)
 576 {
 577     struct client_list *l;
 578     int res = 0;
 579
 580     for (l = s->clients_active; l; l = l->next)
 581         if (client_is_active(l->client))
 582             res++;
 583
 584     return res;
 585 }
 586
 587 int session_is_preferred_clients_ready(struct session *s)
 588 {
 589     struct client_list *l;
 590     int res = 0;
 591
 592     for (l = s->clients_active; l; l = l->next)
 593         if (client_is_active_preferred(l->client))
 594             res++;
 595     session_log(s, YLOG_DEBUG, "Has %d active preferred clients.", res);
 596     return res == 0;
 597 }
 598
 599 static void session_clear_set(struct session *se, struct reclist_sortparms *sp)
 600 {
 601     reclist_destroy(se->reclist);
 602     if (nmem_total(se->nmem))
 603         session_log(se, YLOG_DEBUG, "NMEN operation usage %zd",
 604                     nmem_total(se->nmem));
 605     nmem_reset(se->nmem);
 606     se->total_records = se->total_merged = 0;
 607     se->num_termlists = 0;
 608
 609     relevance_clear(se->relevance);
 610
 611     /* reset list of sorted results and clear to relevance search */
 612     se->sorted_results = nmem_malloc(se->nmem, sizeof(*se->sorted_results));
 613     se->sorted_results->name = nmem_strdup(se->nmem, sp->name);
 614     se->sorted_results->increasing = sp->increasing;
 615     se->sorted_results->type = sp->type;
 616     se->sorted_results->next = 0;
 617
 618     session_log(se, YLOG_DEBUG, "clear_set session_sort: field=%s increasing=%d type=%d configured",
 619             sp->name, sp->increasing, sp->type);
 620
 621     se->reclist = reclist_create(se->nmem);
 622 }
 623
 624 static void session_sort_unlocked(struct session *se,
 625                                   struct reclist_sortparms *sp,
 626                                   const char *mergekey,
 627                                   const char *rank)
 628 {
 629     struct client_list *l;
 630     const char *field = sp->name;
 631     int increasing = sp->increasing;
 632     int type  = sp->type;
 633     int clients_research = 0;
 634
 635     session_log(se, YLOG_DEBUG, "session_sort field=%s increasing=%d type=%d",
 636                 field, increasing, type);
 637
 638     if (rank && (!se->rank || strcmp(se->rank, rank)))
 639     {
 640         /* new rank must research/reingest anyway */
 641         assert(rank);
 642         xfree(se->rank);
 643         se->rank = *rank ? xstrdup(rank) : 0;
 644         clients_research = 1;
 645         session_log(se, YLOG_DEBUG, "session_sort: new rank = %s",
 646                     rank);
 647     }
 648     if (mergekey && (!se->mergekey || strcmp(se->mergekey, mergekey)))
 649     {
 650         /* new mergekey must research/reingest anyway */
 651         assert(mergekey);
 652         xfree(se->mergekey);
 653         se->mergekey = *mergekey ? xstrdup(mergekey) : 0;
 654         clients_research = 1;
 655         session_log(se, YLOG_DEBUG, "session_sort: new mergekey = %s",
 656                     mergekey);
 657     }
 658     if (clients_research == 0)
 659     {
 660         struct reclist_sortparms *sr;
 661         for (sr = se->sorted_results; sr; sr = sr->next)
 662             if (!reclist_sortparms_cmp(sr, sp))
 663                 break;
 664         if (sr)
 665         {
 666             session_log(se, YLOG_DEBUG, "session_sort: field=%s increasing=%d type=%d already fetched",
 667                         field, increasing, type);
 668             return;
 669         }
 670     }
 671     session_log(se, YLOG_DEBUG, "session_sort: field=%s increasing=%d type=%d must fetch",
 672                 field, increasing, type);
 673
 674     // We need to reset reclist on every sort that changes the records, not just for position
 675     // So if just one client requires new searching, we need to clear set.
 676     // Ask each of the client if sorting requires re-search due to native sort
 677     // If it does it will require us to
 678     for (l = se->clients_active; l; l = l->next)
 679     {
 680         struct client *cl = l->client;
 681         // Assume no re-search is required.
 682         client_parse_init(cl, 1);
 683         clients_research += client_parse_sort(cl, sp);
 684     }
 685     if (clients_research)
 686     {
 687         session_log(se, YLOG_DEBUG,
 688                     "session_sort: reset results due to %d clients researching",
 689                     clients_research);
 690         session_clear_set(se, sp);
 691     }
 692     else
 693     {
 694         // A new sorting based on same record set
 695         struct reclist_sortparms *sr = nmem_malloc(se->nmem, sizeof(*sr));
 696         sr->name = nmem_strdup(se->nmem, field);
 697         sr->increasing = increasing;
 698         sr->type = type;
 699         sr->next = se->sorted_results;
 700         se->sorted_results = sr;
 701         session_log(se, YLOG_DEBUG, "session_sort: no research/ingesting done");
 702         return ;
 703     }
 704     session_log(se, YLOG_DEBUG, "Re- search/ingesting for clients due to change in sort order");
 705
 706     for (l = se->clients_active; l; l = l->next)
 707     {
 708         struct client *cl = l->client;
 709         if (client_get_state(cl) == Client_Connecting ||
 710             client_get_state(cl) == Client_Idle ||
 711             client_get_state(cl) == Client_Working) {
 712             client_start_search(cl);
 713         }
 714         else
 715         {
 716             session_log(se, YLOG_DEBUG,
 717                         "session_sort: %s: No re-start/ingest in show. "
 718                         "Wrong client state: %d",
 719                         client_get_id(cl), client_get_state(cl));
 720         }
 721     }
 722 }
 723
 724 void session_sort(struct session *se, struct reclist_sortparms *sp,
 725                   const char *mergekey, const char *rank)
 726 {
 727     //session_enter(se, "session_sort");
 728     session_sort_unlocked(se, sp, mergekey, rank);
 729     //session_leave(se, "session_sort");
 730 }
 731
 732
 733 enum pazpar2_error_code session_search(struct session *se,
 734                                        const char *query,
 735                                        const char *startrecs,
 736                                        const char *maxrecs,
 737                                        const char *filter,
 738                                        const char *limit,
 739                                        const char **addinfo,
 740                                        const char **addinfo2,
 741                                        struct reclist_sortparms *sp,
 742                                        const char *mergekey,
 743                                        const char *rank)
 744 {
 745     int live_channels = 0;
 746     int no_working = 0;
 747     int no_failed_query = 0;
 748     int no_failed_limit = 0;
 749     struct client_list *l, *l0;
 750
 751     session_alert_watch(se, SESSION_WATCH_SHOW);
 752     session_alert_watch(se, SESSION_WATCH_BYTARGET);
 753     session_alert_watch(se, SESSION_WATCH_TERMLIST);
 754     session_alert_watch(se, SESSION_WATCH_SHOW_PREF);
 755
 756     session_log(se, YLOG_DEBUG, "Search");
 757
 758     *addinfo = 0;
 759
 760     if (se->settings_modified) {
 761         session_remove_cached_clients(se);
 762     }
 763     else
 764         session_reset_active_clients(se, 0);
 765
 766     session_enter(se, "session_search");
 767     se->settings_modified = 0;
 768
 769     if (mergekey)
 770     {
 771         xfree(se->mergekey);
 772         se->mergekey = *mergekey ? xstrdup(mergekey) : 0;
 773     }
 774     if (rank)
 775     {
 776         xfree(se->rank);
 777         se->rank = *rank ? xstrdup(rank) : 0;
 778     }
 779
 780     session_clear_set(se, sp);
 781     relevance_destroy(&se->relevance);
 782
 783     live_channels = select_targets(se, filter);
 784     if (!live_channels)
 785     {
 786         session_leave(se, "session_search");
 787         return PAZPAR2_NO_TARGETS;
 788     }
 789
 790     facet_limits_destroy(se->facet_limits);
 791     se->facet_limits = facet_limits_create(limit);
 792     if (!se->facet_limits)
 793     {
 794         *addinfo = "limit";
 795         session_leave(se, "session_search");
 796         return PAZPAR2_MALFORMED_PARAMETER_VALUE;
 797     }
 798
 799     l0 = se->clients_active;
 800     se->clients_active = 0;
 801     session_leave(se, "session_search");
 802
 803     for (l = l0; l; l = l->next)
 804     {
 805         int parse_ret;
 806         struct client *cl = l->client;
 807         client_parse_init(cl, 1);
 808         if (prepare_map(se, client_get_database(cl)) < 0)
 809             continue;
 810
 811         parse_ret = client_parse_query(cl, query, se->facet_limits, addinfo2);
 812         if (parse_ret == -1)
 813             no_failed_query++;
 814         else if (parse_ret == -2)
 815             no_failed_limit++;
 816         else if (parse_ret < 0)
 817             no_working++; /* other error, such as bad CCL map */
 818         else
 819         {
 820             client_parse_range(cl, startrecs, maxrecs);
 821             client_parse_sort(cl, sp);
 822             client_start_search(cl);
 823             no_working++;
 824         }
 825     }
 826     session_reset_active_clients(se, l0);
 827
 828     if (no_working == 0)
 829     {
 830         if (no_failed_query > 0)
 831         {
 832             *addinfo = "query";
 833             return PAZPAR2_MALFORMED_PARAMETER_VALUE;
 834         }
 835         else if (no_failed_limit > 0)
 836         {
 837             *addinfo = "limit";
 838             return PAZPAR2_MALFORMED_PARAMETER_VALUE;
 839         }
 840         else
 841             return PAZPAR2_NO_TARGETS;
 842     }
 843     return PAZPAR2_NO_ERROR;
 844 }
 845
 846 // Creates a new session_database object for a database
 847 static void session_init_databases_fun(void *context, struct database *db)
 848 {
 849     struct session *se = (struct session *) context;
 850     struct session_database *new = nmem_malloc(se->session_nmem, sizeof(*new));
 851     int i;
 852
 853     new->database = db;
 854
 855     new->map = 0;
 856     assert(db->settings);
 857     new->settings = nmem_malloc(se->session_nmem,
 858                                 sizeof(struct settings *) * db->num_settings);
 859     new->num_settings = db->num_settings;
 860     for (i = 0; i < db->num_settings; i++)
 861     {
 862         struct setting *setting = db->settings[i];
 863         new->settings[i] = setting;
 864     }
 865     new->next = se->databases;
 866     se->databases = new;
 867 }
 868
 869 // Doesn't free memory associated with sdb -- nmem takes care of that
 870 static void session_database_destroy(struct session_database *sdb)
 871 {
 872     sdb->map = 0;
 873 }
 874
 875 // Initialize session_database list -- this represents this session's view
 876 // of the database list -- subject to modification by the settings ws command
 877 void session_init_databases(struct session *se)
 878 {
 879     se->databases = 0;
 880     predef_grep_databases(se, se->service, session_init_databases_fun);
 881 }
 882
 883 // Probably session_init_databases_fun should be refactored instead of
 884 // called here.
 885 static struct session_database *load_session_database(struct session *se,
 886                                                       char *id)
 887 {
 888     struct database *db = new_database_inherit_settings(id, se->session_nmem, se->service->settings);
 889     session_init_databases_fun((void*) se, db);
 890
 891     // New sdb is head of se->databases list
 892     return se->databases;
 893 }
 894
 895 // Find an existing session database. If not found, load it
 896 static struct session_database *find_session_database(struct session *se,
 897                                                       char *id)
 898 {
 899     struct session_database *sdb;
 900
 901     for (sdb = se->databases; sdb; sdb = sdb->next)
 902         if (!strcmp(sdb->database->id, id))
 903             return sdb;
 904     return load_session_database(se, id);
 905 }
 906
 907 // Apply a session override to a database
 908 void session_apply_setting(struct session *se, char *dbname, char *setting,
 909                            char *value)
 910 {
 911     struct session_database *sdb = find_session_database(se, dbname);
 912     struct conf_service *service = se->service;
 913     struct setting *new = nmem_malloc(se->session_nmem, sizeof(*new));
 914     int offset = settings_create_offset(service, setting);
 915
 916     expand_settings_array(&sdb->settings, &sdb->num_settings, offset,
 917                           se->session_nmem);
 918     new->precedence = 0;
 919     new->target = dbname;
 920     new->name = setting;
 921     new->value = value;
 922     new->next = sdb->settings[offset];
 923     sdb->settings[offset] = new;
 924
 925     se->settings_modified = 1;
 926
 927     // Force later recompute of settings-driven data structures
 928     // (happens when a search starts and client connections are prepared)
 929     switch (offset)
 930     {
 931     case PZ_XSLT:
 932         if (sdb->map)
 933         {
 934             sdb->map = 0;
 935         }
 936         break;
 937     }
 938 }
 939
 940 void session_destroy(struct session *se)
 941 {
 942     struct session_database *sdb;
 943     session_log(se, YLOG_LOG, "destroy");
 944     session_use(-1);
 945     session_remove_cached_clients(se);
 946
 947     for (sdb = se->databases; sdb; sdb = sdb->next)
 948         session_database_destroy(sdb);
 949     normalize_cache_destroy(se->normalize_cache);
 950     relevance_destroy(&se->relevance);
 951     reclist_destroy(se->reclist);
 952     xfree(se->mergekey);
 953     xfree(se->rank);
 954     if (nmem_total(se->nmem))
 955         session_log(se, YLOG_DEBUG, "NMEN operation usage %zd", nmem_total(se->nmem));
 956     if (nmem_total(se->session_nmem))
 957         session_log(se, YLOG_DEBUG, "NMEN session usage %zd", nmem_total(se->session_nmem));
 958     facet_limits_destroy(se->facet_limits);
 959     nmem_destroy(se->nmem);
 960     service_destroy(se->service);
 961     yaz_mutex_destroy(&se->session_mutex);
 962 }
 963
 964 size_t session_get_memory_status(struct session *session) {
 965     size_t session_nmem;
 966     if (session == 0)
 967         return 0;
 968     session_enter(session, "session_get_memory_status");
 969     session_nmem = nmem_total(session->nmem);
 970     session_leave(session, "session_get_memory_status");
 971     return session_nmem;
 972 }
 973
 974
 975 struct session *new_session(NMEM nmem, struct conf_service *service,
 976                             unsigned session_id)
 977 {
 978     int i;
 979     struct session *session = nmem_malloc(nmem, sizeof(*session));
 980
 981     char tmp_str[50];
 982
 983     sprintf(tmp_str, "session#%u", session_id);
 984
 985     session->session_id = session_id;
 986     session_log(session, YLOG_DEBUG, "New");
 987     session->service = service;
 988     session->relevance = 0;
 989     session->total_records = 0;
 990     session->number_of_warnings_unknown_elements = 0;
 991     session->number_of_warnings_unknown_metadata = 0;
 992     session->num_termlists = 0;
 993     session->reclist = reclist_create(nmem);
 994     session->clients_active = 0;
 995     session->clients_cached = 0;
 996     session->settings_modified = 0;
 997     session->session_nmem = nmem;
 998     session->nmem = nmem_create();
 999     session->databases = 0;
1000     session->sorted_results = 0;
1001     session->facet_limits = 0;
1002     session->mergekey = 0;
1003     session->rank = 0;
1004
1005     for (i = 0; i <= SESSION_WATCH_MAX; i++)
1006     {
1007         session->watchlist[i].data = 0;
1008         session->watchlist[i].fun = 0;
1009     }
1010     session->normalize_cache = normalize_cache_create();
1011     session->session_mutex = 0;
1012     pazpar2_mutex_create(&session->session_mutex, tmp_str);
1013     session_log(session, YLOG_LOG, "create");
1014
1015     session_use(1);
1016     return session;
1017 }
1018
1019 const char * client_get_suggestions_xml(struct client *cl, WRBUF wrbuf);
1020
1021 static struct hitsbytarget *hitsbytarget_nb(struct session *se,
1022                                             int *count, NMEM nmem)
1023 {
1024     struct hitsbytarget *res = 0;
1025     struct client_list *l;
1026     size_t sz = 0;
1027
1028     for (l = se->clients_active; l; l = l->next)
1029         sz++;
1030
1031     res = nmem_malloc(nmem, sizeof(*res) * sz);
1032     *count = 0;
1033     for (l = se->clients_active; l; l = l->next)
1034     {
1035         struct client *cl = l->client;
1036         WRBUF w = wrbuf_alloc();
1037         const char *name = session_setting_oneval(client_get_database(cl),
1038                                                   PZ_NAME);
1039
1040         res[*count].id = client_get_id(cl);
1041         res[*count].name = *name ? name : "Unknown";
1042         res[*count].hits = client_get_hits(cl);
1043         res[*count].approximation = client_get_approximation(cl);
1044         res[*count].records = client_get_num_records(cl);
1045         res[*count].filtered = client_get_num_records_filtered(cl);
1046         res[*count].diagnostic =
1047             client_get_diagnostic(cl, &res[*count].message,
1048                                   &res[*count].addinfo);
1049         res[*count].state = client_get_state_str(cl);
1050         res[*count].connected  = client_get_connection(cl) ? 1 : 0;
1051         session_settings_dump(se, client_get_database(cl), w);
1052         res[*count].settings_xml = nmem_strdup(nmem, wrbuf_cstr(w));
1053         wrbuf_rewind(w);
1054         wrbuf_puts(w, "");
1055         res[*count].suggestions_xml = nmem_strdup(nmem, client_get_suggestions_xml(cl, w));
1056         wrbuf_destroy(w);
1057         (*count)++;
1058     }
1059     return res;
1060 }
1061
1062 struct hitsbytarget *get_hitsbytarget(struct session *se, int *count, NMEM nmem)
1063 {
1064     struct hitsbytarget *p;
1065     session_enter(se, "get_hitsbytarget");
1066     p = hitsbytarget_nb(se, count, nmem);
1067     session_leave(se, "get_hitsbytarget");
1068     return p;
1069 }
1070
1071 // Compares two hitsbytarget nodes by hitcount
1072 static int cmp_ht(const void *p1, const void *p2)
1073 {
1074     const struct hitsbytarget *h1 = p1;
1075     const struct hitsbytarget *h2 = p2;
1076     return h2->hits - h1->hits;
1077 }
1078
1079 // Compares two hitsbytarget nodes by hitcount
1080 static int cmp_ht_approx(const void *p1, const void *p2)
1081 {
1082     const struct hitsbytarget *h1 = p1;
1083     const struct hitsbytarget *h2 = p2;
1084     return h2->approximation - h1->approximation;
1085 }
1086
1087 static int targets_termlist_nb(WRBUF wrbuf, struct session *se, int num,
1088                                NMEM nmem, int version)
1089 {
1090     struct hitsbytarget *ht;
1091     int count, i;
1092
1093     ht = hitsbytarget_nb(se, &count, nmem);
1094     if (version >= 2)
1095         qsort(ht, count, sizeof(struct hitsbytarget), cmp_ht_approx);
1096     else
1097         qsort(ht, count, sizeof(struct hitsbytarget), cmp_ht);
1098     for (i = 0; i < count && i < num && ht[i].hits > 0; i++)
1099     {
1100
1101         // do only print terms which have display names
1102
1103         wrbuf_puts(wrbuf, "<term>\n");
1104
1105         wrbuf_puts(wrbuf, "<id>");
1106         wrbuf_xmlputs(wrbuf, ht[i].id);
1107         wrbuf_puts(wrbuf, "</id>\n");
1108
1109         wrbuf_puts(wrbuf, "<name>");
1110         if (!ht[i].name || !ht[i].name[0])
1111             wrbuf_xmlputs(wrbuf, "NO TARGET NAME");
1112         else
1113             wrbuf_xmlputs(wrbuf, ht[i].name);
1114         wrbuf_puts(wrbuf, "</name>\n");
1115
1116         wrbuf_printf(wrbuf, "<frequency>" ODR_INT_PRINTF "</frequency>\n",
1117                      ht[i].hits);
1118
1119         if (version >= 2) {
1120             // Should not print if we know it isn't a approximation.
1121             wrbuf_printf(wrbuf, "<approximation>" ODR_INT_PRINTF "</approximation>\n", ht[i].approximation);
1122             wrbuf_printf(wrbuf, "<records>%d</records>\n", ht[i].records - ht[i].filtered);
1123             wrbuf_printf(wrbuf, "<filtered>%d</filtered>\n", ht[i].filtered);
1124         }
1125
1126         wrbuf_puts(wrbuf, "<state>");
1127         wrbuf_xmlputs(wrbuf, ht[i].state);
1128         wrbuf_puts(wrbuf, "</state>\n");
1129
1130         wrbuf_printf(wrbuf, "<diagnostic>%d</diagnostic>\n",
1131                      ht[i].diagnostic);
1132         wrbuf_puts(wrbuf, "</term>\n");
1133     }
1134     return count;
1135 }
1136
1137 void perform_termlist(struct http_channel *c, struct session *se,
1138                       const char *name, int num, int version)
1139 {
1140     int i, j;
1141     NMEM nmem_tmp = nmem_create();
1142     char **names;
1143     int num_names = 0;
1144
1145     if (!name)
1146         name = "*";
1147
1148     nmem_strsplit(nmem_tmp, ",", name, &names, &num_names);
1149
1150     session_enter(se, "perform_termlist");
1151
1152     for (j = 0; j < num_names; j++)
1153     {
1154         const char *tname;
1155         int must_generate_empty = 1; /* bug 5350 */
1156
1157         for (i = 0; i < se->num_termlists; i++)
1158         {
1159             tname = se->termlists[i].name;
1160             if (!strcmp(names[j], tname) || !strcmp(names[j], "*"))
1161             {
1162                 struct termlist_score **p = 0;
1163                 int len;
1164
1165                 wrbuf_puts(c->wrbuf, "<list name=\"");
1166                 wrbuf_xmlputs(c->wrbuf, tname);
1167                 wrbuf_puts(c->wrbuf, "\">\n");
1168                 must_generate_empty = 0;
1169
1170                 p = termlist_highscore(se->termlists[i].termlist, &len,
1171                                        nmem_tmp);
1172                 if (p)
1173                 {
1174                     int i;
1175                     for (i = 0; i < len && i < num; i++)
1176                     {
1177                         // prevent sending empty term elements
1178                         if (!p[i]->display_term || !p[i]->display_term[0])
1179                             continue;
1180
1181                         wrbuf_puts(c->wrbuf, "<term>");
1182                         wrbuf_puts(c->wrbuf, "<name>");
1183                         wrbuf_xmlputs(c->wrbuf, p[i]->display_term);
1184                         wrbuf_puts(c->wrbuf, "</name>");
1185
1186                         wrbuf_printf(c->wrbuf,
1187                                      "<frequency>%d</frequency>",
1188                                      p[i]->frequency);
1189                         wrbuf_puts(c->wrbuf, "</term>\n");
1190                     }
1191                 }
1192                 wrbuf_puts(c->wrbuf, "</list>\n");
1193             }
1194         }
1195         tname = "xtargets";
1196         if (!strcmp(names[j], tname) || !strcmp(names[j], "*"))
1197         {
1198             wrbuf_puts(c->wrbuf, "<list name=\"");
1199             wrbuf_xmlputs(c->wrbuf, tname);
1200             wrbuf_puts(c->wrbuf, "\">\n");
1201
1202             targets_termlist_nb(c->wrbuf, se, num, c->nmem, version);
1203             wrbuf_puts(c->wrbuf, "</list>\n");
1204             must_generate_empty = 0;
1205         }
1206         if (must_generate_empty)
1207         {
1208             wrbuf_puts(c->wrbuf, "<list name=\"");
1209             wrbuf_xmlputs(c->wrbuf, names[j]);
1210             wrbuf_puts(c->wrbuf, "\"/>\n");
1211         }
1212     }
1213     session_leave(se, "perform_termlist");
1214     nmem_destroy(nmem_tmp);
1215 }
1216
1217 #ifdef MISSING_HEADERS
1218 void report_nmem_stats(void)
1219 {
1220     size_t in_use, is_free;
1221
1222     nmem_get_memory_in_use(&in_use);
1223     nmem_get_memory_free(&is_free);
1224
1225     yaz_log(YLOG_LOG, "nmem stat: use=%ld free=%ld",
1226             (long) in_use, (long) is_free);
1227 }
1228 #endif
1229
1230 struct record_cluster *show_single_start(struct session *se, const char *id,
1231                                          struct record_cluster **prev_r,
1232                                          struct record_cluster **next_r)
1233 {
1234     struct record_cluster *r = 0;
1235
1236     session_enter(se, "show_single_start");
1237     *prev_r = 0;
1238     *next_r = 0;
1239     reclist_limit(se->reclist, se, 1);
1240
1241     reclist_enter(se->reclist);
1242     while ((r = reclist_read_record(se->reclist)))
1243     {
1244         if (!strcmp(r->recid, id))
1245         {
1246             *next_r = reclist_read_record(se->reclist);
1247             break;
1248         }
1249         *prev_r = r;
1250     }
1251     reclist_leave(se->reclist);
1252     if (!r)
1253         session_leave(se, "show_single_start");
1254     return r;
1255 }
1256
1257 void show_single_stop(struct session *se, struct record_cluster *rec)
1258 {
1259     session_leave(se, "show_single_stop");
1260 }
1261
1262
1263 int session_fetch_more(struct session *se)
1264 {
1265     struct client_list *l;
1266     int ret = 0;
1267
1268     for (l = se->clients_active; l; l = l->next)
1269     {
1270         struct client *cl = l->client;
1271         if (client_get_state(cl) == Client_Idle)
1272         {
1273             if (client_fetch_more(cl))
1274             {
1275                 session_log(se, YLOG_LOG, "%s: more to fetch",
1276                             client_get_id(cl));
1277                 ret = 1;
1278             }
1279             else
1280             {
1281                 session_log(se, YLOG_LOG, "%s: no more to fetch",
1282                             client_get_id(cl));
1283             }
1284         }
1285         else
1286         {
1287             session_log(se, YLOG_LOG, "%s: no fetch due to state=%s",
1288                         client_get_id(cl), client_get_state_str(cl));
1289         }
1290
1291     }
1292     return ret;
1293 }
1294
1295 struct record_cluster **show_range_start(struct session *se,
1296                                          struct reclist_sortparms *sp,
1297                                          int start, int *num, int *total,
1298                                          Odr_int *sumhits, Odr_int *approx_hits,
1299                                          void (*show_records_ready)(void *data),
1300                                          struct http_channel *chan)
1301 {
1302     struct record_cluster **recs = 0;
1303     struct reclist_sortparms *spp;
1304     struct client_list *l;
1305     int i;
1306 #if USE_TIMING
1307     yaz_timing_t t = yaz_timing_create();
1308 #endif
1309     session_enter(se, "show_range_start");
1310     *sumhits = 0;
1311     *approx_hits = 0;
1312     *total = 0;
1313     reclist_limit(se->reclist, se, 0);
1314     if (se->relevance)
1315     {
1316         for (spp = sp; spp; spp = spp->next)
1317             if (spp->type == Metadata_type_relevance)
1318             {
1319                 relevance_prepare_read(se->relevance, se->reclist);
1320                 break;
1321             }
1322         for (l = se->clients_active; l; l = l->next) {
1323             *sumhits += client_get_hits(l->client);
1324             *approx_hits += client_get_approximation(l->client);
1325         }
1326     }
1327     reclist_sort(se->reclist, sp);
1328
1329     reclist_enter(se->reclist);
1330     *total = reclist_get_num_records(se->reclist);
1331
1332     for (l = se->clients_active; l; l = l->next)
1333         client_update_show_stat(l->client, 0);
1334
1335     for (i = 0; i < start; i++)
1336     {
1337         struct record_cluster *r = reclist_read_record(se->reclist);
1338         if (!r)
1339         {
1340             *num = 0;
1341             break;
1342         }
1343         else
1344         {
1345             struct record *rec = r->records;
1346             for (;rec; rec = rec->next)
1347                 client_update_show_stat(rec->client, 1);
1348         }
1349     }
1350     recs = nmem_malloc(se->nmem, (*num > 0 ? *num : 1) * sizeof(*recs));
1351     for (i = 0; i < *num; i++)
1352     {
1353         struct record_cluster *r = reclist_read_record(se->reclist);
1354         if (!r)
1355         {
1356             *num = i;
1357             break;
1358         }
1359         else
1360         {
1361             struct record *rec = r->records;
1362             for (;rec; rec = rec->next)
1363                 client_update_show_stat(rec->client, 1);
1364             recs[i] = r;
1365         }
1366     }
1367     reclist_leave(se->reclist);
1368 #if USE_TIMING
1369     yaz_timing_stop(t);
1370     session_log(se, YLOG_LOG, "show %6.5f %3.2f %3.2f",
1371             yaz_timing_get_real(t), yaz_timing_get_user(t),
1372             yaz_timing_get_sys(t));
1373     yaz_timing_destroy(&t);
1374 #endif
1375
1376     if (!session_fetch_more(se))
1377         session_log(se, YLOG_LOG, "can not fetch more");
1378     else
1379     {
1380         show_range_stop(se, recs);
1381         session_log(se, YLOG_LOG, "fetching more in progress");
1382         if (session_set_watch(se, SESSION_WATCH_SHOW,
1383                               show_records_ready, chan, chan))
1384         {
1385             session_log(se, YLOG_WARN, "Ignoring show block");
1386             session_enter(se, "show_range_start");
1387         }
1388         else
1389         {
1390             session_log(se, YLOG_LOG, "session watch OK");
1391             return 0;
1392         }
1393     }
1394     return recs;
1395 }
1396
1397 void show_range_stop(struct session *se, struct record_cluster **recs)
1398 {
1399     session_leave(se, "show_range_stop");
1400 }
1401
1402 void statistics(struct session *se, struct statistics *stat)
1403 {
1404     struct client_list *l;
1405     int count = 0;
1406
1407     memset(stat, 0, sizeof(*stat));
1408     stat->num_hits = 0;
1409     for (l = se->clients_active; l; l = l->next)
1410     {
1411         struct client *cl = l->client;
1412         if (!client_get_connection(cl))
1413             stat->num_no_connection++;
1414         stat->num_hits += client_get_hits(cl);
1415         switch (client_get_state(cl))
1416         {
1417         case Client_Connecting: stat->num_connecting++; break;
1418         case Client_Working: stat->num_working++; break;
1419         case Client_Idle: stat->num_idle++; break;
1420         case Client_Failed: stat->num_failed++; break;
1421         case Client_Error: stat->num_error++; break;
1422         default: break;
1423         }
1424         count++;
1425     }
1426     stat->num_records = se->total_records;
1427
1428     stat->num_clients = count;
1429 }
1430
1431 static struct record_metadata *record_metadata_init(
1432     NMEM nmem, const char *value, enum conf_metadata_type type,
1433     struct _xmlAttr *attr)
1434 {
1435     struct record_metadata *rec_md = record_metadata_create(nmem);
1436     struct record_metadata_attr **attrp = &rec_md->attributes;
1437
1438     for (; attr; attr = attr->next)
1439     {
1440         if (attr->children && attr->children->content)
1441         {
1442             if (strcmp((const char *) attr->name, "type")
1443                 && strcmp((const char *) attr->name, "empty"))
1444             {  /* skip the "type" + "empty" attribute..
1445                   The "Type" is already part of the element in output
1446                   (md-%s) and so repeating it here is redundant */
1447                 *attrp = nmem_malloc(nmem, sizeof(**attrp));
1448                 (*attrp)->name =
1449                     nmem_strdup(nmem, (const char *) attr->name);
1450                 (*attrp)->value =
1451                     nmem_strdup(nmem, (const char *) attr->children->content);
1452                 attrp = &(*attrp)->next;
1453             }
1454         }
1455     }
1456     *attrp = 0;
1457
1458     switch (type)
1459     {
1460     case Metadata_type_generic:
1461     case Metadata_type_skiparticle:
1462         rec_md->data.text.disp =
1463             normalize7bit_generic(nmem_strdup(nmem, value), " ,/.:([");
1464         rec_md->data.text.sort = 0;
1465         rec_md->data.text.snippet = 0;
1466         break;
1467     case Metadata_type_year:
1468     case Metadata_type_date:
1469     {
1470         int first, last;
1471         int longdate = 0;
1472
1473         if (type == Metadata_type_date)
1474             longdate = 1;
1475         if (extract7bit_dates((char *) value, &first, &last, longdate) < 0)
1476             return 0;
1477
1478         rec_md->data.number.min = first;
1479         rec_md->data.number.max = last;
1480     }
1481     break;
1482     case Metadata_type_float:
1483         rec_md->data.fnumber = atof(value);
1484         break;
1485     case Metadata_type_relevance:
1486     case Metadata_type_position:
1487         return 0;
1488     }
1489     return rec_md;
1490 }
1491
1492 static void mergekey_norm_wr(pp2_charset_fact_t charsets,
1493                              WRBUF norm_wr, const char *value)
1494 {
1495     const char *norm_str;
1496     pp2_charset_token_t prt =
1497         pp2_charset_token_create(charsets, "mergekey");
1498
1499     pp2_charset_token_first(prt, value, 0);
1500     while ((norm_str = pp2_charset_token_next(prt)))
1501     {
1502         if (*norm_str)
1503         {
1504             if (wrbuf_len(norm_wr))
1505                 wrbuf_puts(norm_wr, " ");
1506             wrbuf_puts(norm_wr, norm_str);
1507         }
1508     }
1509     pp2_charset_token_destroy(prt);
1510 }
1511
1512 static int get_mergekey_from_doc(xmlDoc *doc, xmlNode *root, const char *name,
1513                                  struct conf_service *service, WRBUF norm_wr)
1514 {
1515     xmlNode *n;
1516     int no_found = 0;
1517     for (n = root->children; n; n = n->next)
1518     {
1519         if (n->type != XML_ELEMENT_NODE)
1520             continue;
1521         if (!strcmp((const char *) n->name, "metadata"))
1522         {
1523             xmlChar *type = xmlGetProp(n, (xmlChar *) "type");
1524             if (type == NULL) {
1525                 yaz_log(YLOG_FATAL, "Missing type attribute on metadata element. Skipping!");
1526             }
1527             else if (!strcmp(name, (const char *) type))
1528             {
1529                 xmlChar *value = xmlNodeListGetString(doc, n->children, 1);
1530                 if (value && *value)
1531                 {
1532                     if (wrbuf_len(norm_wr) > 0)
1533                         wrbuf_puts(norm_wr, " ");
1534                     wrbuf_puts(norm_wr, name);
1535                     mergekey_norm_wr(service->charsets, norm_wr,
1536                                      (const char *) value);
1537                     no_found++;
1538                 }
1539                 if (value)
1540                     xmlFree(value);
1541             }
1542             xmlFree(type);
1543         }
1544     }
1545     return no_found;
1546 }
1547
1548 static const char *get_mergekey(xmlDoc *doc, xmlNode *root,
1549                                 struct client *cl, int record_no,
1550                                 struct conf_service *service, NMEM nmem,
1551                                 const char *session_mergekey)
1552 {
1553     char *mergekey_norm = 0;
1554     WRBUF norm_wr = wrbuf_alloc();
1555     xmlChar *mergekey;
1556
1557     if (session_mergekey)
1558     {
1559         int i, num = 0;
1560         char **values = 0;
1561         nmem_strsplit_escape2(nmem, ",", session_mergekey, &values,
1562                               &num, 1, '\\', 1);
1563
1564         for (i = 0; i < num; i++)
1565             get_mergekey_from_doc(doc, root, values[i], service, norm_wr);
1566     }
1567     else if ((mergekey = xmlGetProp(root, (xmlChar *) "mergekey")))
1568     {
1569         mergekey_norm_wr(service->charsets, norm_wr, (const char *) mergekey);
1570         xmlFree(mergekey);
1571     }
1572     else
1573     {
1574         /* no mergekey defined in XSL. Look for mergekey metadata instead */
1575         int field_id;
1576         for (field_id = 0; field_id < service->num_metadata; field_id++)
1577         {
1578             struct conf_metadata *ser_md = &service->metadata[field_id];
1579             if (ser_md->mergekey != Metadata_mergekey_no)
1580             {
1581                 int r = get_mergekey_from_doc(doc, root, ser_md->name,
1582                                               service, norm_wr);
1583                 if (r == 0 && ser_md->mergekey == Metadata_mergekey_required)
1584                 {
1585                     /* no mergekey on this one and it is required..
1586                        Generate unique key instead */
1587                     wrbuf_rewind(norm_wr);
1588                     break;
1589                 }
1590             }
1591         }
1592     }
1593
1594     /* generate unique key if none is not generated already or is empty */
1595     if (wrbuf_len(norm_wr) == 0)
1596     {
1597         wrbuf_printf(norm_wr, "position: %s-%d",
1598                      client_get_id(cl), record_no);
1599     }
1600     else
1601     {
1602         const char *lead = "content: ";
1603         wrbuf_insert(norm_wr, 0, lead, strlen(lead));
1604     }
1605     if (wrbuf_len(norm_wr) > 0)
1606         mergekey_norm = nmem_strdup(nmem, wrbuf_cstr(norm_wr));
1607     wrbuf_destroy(norm_wr);
1608     return mergekey_norm;
1609 }
1610
1611 /** \brief see if metadata for pz:recordfilter exists
1612     \param root xml root element of normalized record
1613     \param sdb session database for client
1614     \retval 0 if there is no metadata for pz:recordfilter
1615     \retval 1 if there is metadata for pz:recordfilter
1616
1617     If there is no pz:recordfilter defined, this function returns 1
1618     as well.
1619 */
1620
1621 static int check_record_filter(xmlNode *root, struct session_database *sdb)
1622 {
1623     int match = 0;
1624     xmlNode *n;
1625     const char *s;
1626     s = session_setting_oneval(sdb, PZ_RECORDFILTER);
1627
1628     if (!s || !*s)
1629         return 1;
1630
1631     for (n = root->children; n; n = n->next)
1632     {
1633         if (n->type != XML_ELEMENT_NODE)
1634             continue;
1635         if (!strcmp((const char *) n->name, "metadata"))
1636         {
1637             xmlChar *type = xmlGetProp(n, (xmlChar *) "type");
1638             if (type)
1639             {
1640                 size_t len;
1641                 int substring;
1642                 const char *eq;
1643
1644                 if ((eq = strchr(s, '=')))
1645                     substring = 0;
1646                 else if ((eq = strchr(s, '~')))
1647                     substring = 1;
1648                 if (eq)
1649                     len = eq - s;
1650                 else
1651                     len = strlen(s);
1652                 if (len == strlen((const char *)type) &&
1653                     !memcmp((const char *) type, s, len))
1654                 {
1655                     xmlChar *value = xmlNodeGetContent(n);
1656                     if (value && *value)
1657                     {
1658                         if (!eq ||
1659                             (substring && strstr((const char *) value, eq+1)) ||
1660                             (!substring && !strcmp((const char *) value, eq + 1)))
1661                             match = 1;
1662                     }
1663                     xmlFree(value);
1664                 }
1665                 xmlFree(type);
1666             }
1667         }
1668     }
1669     return match;
1670 }
1671
1672 static int ingest_to_cluster(struct client *cl,
1673                              xmlDoc *xdoc,
1674                              xmlNode *root,
1675                              int record_no,
1676                              struct record_metadata_attr *mergekey);
1677
1678 static int ingest_sub_record(struct client *cl, xmlDoc *xdoc, xmlNode *root,
1679                              int record_no, NMEM nmem,
1680                              struct session_database *sdb,
1681                              struct record_metadata_attr *mergekeys)
1682 {
1683     int ret = 0;
1684     struct session *se = client_get_session(cl);
1685     struct conf_service *service = se->service;
1686
1687     insert_settings_values(sdb, xdoc, root, service);
1688
1689     if (!check_record_filter(root, sdb))
1690     {
1691         session_log(se, YLOG_LOG,
1692                     "Filtered out record no %d from %s",
1693                     record_no, sdb->database->id);
1694         return 0;
1695     }
1696     session_enter(se, "ingest_sub_record");
1697     if (client_get_session(cl) == se && se->relevance)
1698         ret = ingest_to_cluster(cl, xdoc, root, record_no, mergekeys);
1699     session_leave(se, "ingest_sub_record");
1700
1701     return ret;
1702 }
1703
1704 /** \brief ingest XML record
1705     \param cl client holds the result set for record
1706     \param rec record buffer (0 terminated)
1707     \param record_no record position (1, 2, ..)
1708     \param nmem working NMEM
1709     \retval 0 OK
1710     \retval -1 failure
1711     \retval -2 Filtered
1712 */
1713 int ingest_record(struct client *cl, const char *rec,
1714                   int record_no, NMEM nmem)
1715 {
1716     struct session *se = client_get_session(cl);
1717     struct session_database *sdb = client_get_database(cl);
1718     struct conf_service *service = se->service;
1719     xmlDoc *xdoc = normalize_record(se, sdb, service, rec, nmem);
1720     int r = 0;
1721     xmlNode *root;
1722
1723     if (!xdoc)
1724         return -1;
1725
1726     if (global_parameters.dump_records)
1727     {
1728         session_log(se, YLOG_LOG, "Normalized record from %s",
1729                     sdb->database->id);
1730         log_xml_doc(xdoc);
1731     }
1732
1733     root = xmlDocGetRootElement(xdoc);
1734
1735     if (!strcmp((const char *) root->name, "cluster"))
1736     {
1737         int no_merge_keys = 0;
1738         int no_merge_dups = 0;
1739         xmlNode *sroot;
1740         struct record_metadata_attr *mk = 0;
1741
1742         for (sroot = root->children; sroot; sroot = sroot->next)
1743             if (sroot->type == XML_ELEMENT_NODE &&
1744                 !strcmp((const char *) sroot->name, "record"))
1745             {
1746                 struct record_metadata_attr **mkp;
1747                 const char *mergekey_norm =
1748                     get_mergekey(xdoc, sroot, cl, record_no, service, nmem,
1749                                  se->mergekey);
1750                 if (!mergekey_norm)
1751                 {
1752                     r = -1;
1753                     break;
1754                 }
1755                 for (mkp = &mk; *mkp; mkp = &(*mkp)->next)
1756                     if (!strcmp((*mkp)->value, mergekey_norm))
1757                         break;
1758                 if (!*mkp)
1759                 {
1760                     *mkp = (struct record_metadata_attr*)
1761                         nmem_malloc(nmem, sizeof(**mkp));
1762                     (*mkp)->name = 0;
1763                     (*mkp)->value = nmem_strdup(nmem, mergekey_norm);
1764                     (*mkp)->next = 0;
1765                     no_merge_keys++;
1766                 }
1767                 else
1768                     no_merge_dups++;
1769             }
1770         if (no_merge_keys > 1 || no_merge_dups > 0)
1771         {
1772             yaz_log(YLOG_LOG, "Got %d mergekeys, %d dups for position %d",
1773                     no_merge_keys, no_merge_dups, record_no);
1774         }
1775         for (sroot = root->children; !r && sroot; sroot = sroot->next)
1776             if (sroot->type == XML_ELEMENT_NODE &&
1777                 !strcmp((const char *) sroot->name, "record"))
1778             {
1779                 r = ingest_sub_record(cl, xdoc, sroot, record_no, nmem, sdb,
1780                                       mk);
1781             }
1782     }
1783     else if (!strcmp((const char *) root->name, "record"))
1784     {
1785         const char *mergekey_norm =
1786             get_mergekey(xdoc, root, cl, record_no, service, nmem,
1787                          se->mergekey);
1788         if (mergekey_norm)
1789         {
1790             struct record_metadata_attr *mk = (struct record_metadata_attr*)
1791                 nmem_malloc(nmem, sizeof(*mk));
1792             mk->name = 0;
1793             mk->value = nmem_strdup(nmem, mergekey_norm);
1794             mk->next = 0;
1795
1796             r = ingest_sub_record(cl, xdoc, root, record_no, nmem, sdb, mk);
1797         }
1798     }
1799     else
1800     {
1801         session_log(se, YLOG_WARN, "Bad pz root element: %s",
1802                     (const char *) root->name);
1803         r = -1;
1804     }
1805     xmlFreeDoc(xdoc);
1806     return r;
1807 }
1808
1809
1810 //    struct conf_metadata *ser_md = &service->metadata[md_field_id];
1811 //    struct record_metadata *rec_md = record->metadata[md_field_id];
1812 static int match_metadata_local(struct conf_service *service,
1813                                 struct conf_metadata *ser_md,
1814                                 struct record_metadata *rec_md0,
1815                                 char **values, int num_v)
1816 {
1817     int i;
1818     struct record_metadata *rec_md = rec_md0;
1819     WRBUF val_wr = 0;
1820     WRBUF text_wr = wrbuf_alloc();
1821     for (i = 0; i < num_v; )
1822     {
1823         if (rec_md)
1824         {
1825             if (ser_md->type == Metadata_type_year
1826                 || ser_md->type == Metadata_type_date)
1827             {
1828                 int y = atoi(values[i]);
1829                 if (y >= rec_md->data.number.min
1830                     && y <= rec_md->data.number.max)
1831                     break;
1832             }
1833             else
1834             {
1835                 if (!val_wr)
1836                 {
1837                     val_wr = wrbuf_alloc();
1838                     mergekey_norm_wr(service->charsets, val_wr, values[i]);
1839                 }
1840                 wrbuf_rewind(text_wr);
1841                 mergekey_norm_wr(service->charsets, text_wr,
1842                                  rec_md->data.text.disp);
1843                 if (!strcmp(wrbuf_cstr(val_wr), wrbuf_cstr(text_wr)))
1844                     break;
1845             }
1846             rec_md = rec_md->next;
1847         }
1848         else
1849         {
1850             rec_md = rec_md0;
1851             wrbuf_destroy(val_wr);
1852             val_wr = 0;
1853             i++;
1854         }
1855     }
1856     wrbuf_destroy(val_wr);
1857     wrbuf_destroy(text_wr);
1858     return i < num_v ? 1 : 0;
1859 }
1860
1861 int session_check_cluster_limit(struct session *se, struct record_cluster *rec)
1862 {
1863     int i;
1864     struct conf_service *service = se->service;
1865     int ret = 1;
1866     const char *name;
1867     const char *value;
1868     NMEM nmem_tmp = nmem_create();
1869
1870     for (i = 0; (name = facet_limits_get(se->facet_limits, i, &value)); i++)
1871     {
1872         int j;
1873         for (j = 0; j < service->num_metadata; j++)
1874         {
1875             struct conf_metadata *md = service->metadata + j;
1876             if (!strcmp(md->name, name) && md->limitcluster)
1877             {
1878                 char **values = 0;
1879                 int num = 0;
1880                 int md_field_id =
1881                     conf_service_metadata_field_id(service,
1882                                                    md->limitcluster);
1883
1884                 if (md_field_id < 0)
1885                 {
1886                     ret = 0;
1887                     break;
1888                 }
1889
1890                 nmem_strsplit_escape2(nmem_tmp, "|", value, &values,
1891                                       &num, 1, '\\', 1);
1892
1893                 if (!match_metadata_local(service,
1894                                           &service->metadata[md_field_id],
1895                                           rec->metadata[md_field_id],
1896                                           values, num))
1897                 {
1898                     ret = 0;
1899                     break;
1900                 }
1901             }
1902         }
1903     }
1904     nmem_destroy(nmem_tmp);
1905     return ret;
1906 }
1907
1908 // Skip record on non-zero
1909 static int check_limit_local(struct client *cl,
1910                              struct record *record,
1911                              int record_no)
1912 {
1913     int skip_record = 0;
1914     struct session *se = client_get_session(cl);
1915     struct conf_service *service = se->service;
1916     NMEM nmem_tmp = nmem_create();
1917     struct session_database *sdb = client_get_database(cl);
1918     int l = 0;
1919     while (!skip_record)
1920     {
1921         int md_field_id;
1922         char **values = 0;
1923         int num_v = 0;
1924         const char *name =
1925             client_get_facet_limit_local(cl, sdb, &l, nmem_tmp,
1926                                          &num_v, &values);
1927         if (!name)
1928             break;
1929
1930         if (!strcmp(name, "*"))
1931         {
1932             for (md_field_id = 0; md_field_id < service->num_metadata;
1933                  md_field_id++)
1934             {
1935                 if (match_metadata_local(
1936                         service,
1937                         &service->metadata[md_field_id],
1938                         record->metadata[md_field_id],
1939                         values, num_v))
1940                     break;
1941             }
1942             if (md_field_id == service->num_metadata)
1943                 skip_record = 1;
1944         }
1945         else
1946         {
1947             md_field_id = conf_service_metadata_field_id(service, name);
1948             if (md_field_id < 0)
1949             {
1950                 skip_record = 1;
1951                 break;
1952             }
1953             if (!match_metadata_local(
1954                     service,
1955                     &service->metadata[md_field_id],
1956                     record->metadata[md_field_id],
1957                     values, num_v))
1958             {
1959                 skip_record = 1;
1960             }
1961         }
1962     }
1963     nmem_destroy(nmem_tmp);
1964     return skip_record;
1965 }
1966
1967 static int ingest_to_cluster(struct client *cl,
1968                              xmlDoc *xdoc,
1969                              xmlNode *root,
1970                              int record_no,
1971                              struct record_metadata_attr *merge_keys)
1972 {
1973     xmlNode *n;
1974     xmlChar *type = 0;
1975     xmlChar *value = 0;
1976     struct session *se = client_get_session(cl);
1977     struct conf_service *service = se->service;
1978     int term_factor = 1;
1979     struct record_cluster *cluster;
1980     struct record_metadata **metadata0;
1981     struct session_database *sdb = client_get_database(cl);
1982     NMEM ingest_nmem = 0;
1983     char **rank_values = 0;
1984     int rank_num = 0;
1985     struct record *record = record_create(se->nmem,
1986                                           service->num_metadata,
1987                                           service->num_sortkeys, cl,
1988                                           record_no);
1989
1990     for (n = root->children; n; n = n->next)
1991     {
1992         if (type)
1993             xmlFree(type);
1994         if (value)
1995             xmlFree(value);
1996         type = value = 0;
1997
1998         if (n->type != XML_ELEMENT_NODE)
1999             continue;
2000         if (!strcmp((const char *) n->name, "metadata"))
2001         {
2002             struct conf_metadata *ser_md = 0;
2003             struct record_metadata **wheretoput = 0;
2004             struct record_metadata *rec_md = 0;
2005             int md_field_id = -1;
2006
2007             type = xmlGetProp(n, (xmlChar *) "type");
2008             value = xmlNodeListGetString(xdoc, n->children, 1);
2009             if (!type)
2010                 continue;
2011             if (!value || !*value)
2012             {
2013                 xmlChar *empty = xmlGetProp(n, (xmlChar *) "empty");
2014                 if (!empty)
2015                     continue;
2016                 if (value)
2017                     xmlFree(value);
2018                 value = empty;
2019             }
2020             md_field_id
2021                 = conf_service_metadata_field_id(service, (const char *) type);
2022             if (md_field_id < 0)
2023             {
2024                 if (se->number_of_warnings_unknown_metadata == 0)
2025                 {
2026                     session_log(se, YLOG_WARN,
2027                             "Ignoring unknown metadata element: %s", type);
2028                 }
2029                 se->number_of_warnings_unknown_metadata++;
2030                 continue;
2031             }
2032
2033             ser_md = &service->metadata[md_field_id];
2034
2035             // non-merged metadata
2036             rec_md = record_metadata_init(se->nmem, (const char *) value,
2037                                           ser_md->type, n->properties);
2038             if (!rec_md)
2039             {
2040                 session_log(se, YLOG_WARN, "bad metadata data '%s' "
2041                             "for element '%s'", value, type);
2042                 continue;
2043             }
2044
2045             if (ser_md->type == Metadata_type_generic)
2046             {
2047                 WRBUF w = wrbuf_alloc();
2048                 if (relevance_snippet(se->relevance,
2049                                       (char*) value, ser_md->name, w))
2050                     rec_md->data.text.snippet = nmem_strdup(se->nmem,
2051                                                             wrbuf_cstr(w));
2052                 wrbuf_destroy(w);
2053             }
2054
2055
2056             wheretoput = &record->metadata[md_field_id];
2057             while (*wheretoput)
2058                 wheretoput = &(*wheretoput)->next;
2059             *wheretoput = rec_md;
2060         }
2061     }
2062
2063     if (check_limit_local(cl, record, record_no))
2064     {
2065         session_log(se, YLOG_LOG, "Facet filtered out record no %d from %s",
2066                     record_no, sdb->database->id);
2067         if (type)
2068             xmlFree(type);
2069         if (value)
2070             xmlFree(value);
2071         return -2;
2072     }
2073     cluster = reclist_insert(se->reclist, se->relevance, service, record,
2074                              merge_keys, &se->total_merged);
2075     if (!cluster)
2076     {
2077         if (type)
2078             xmlFree(type);
2079         if (value)
2080             xmlFree(value);
2081         return 0; // complete match with existing record
2082     }
2083
2084     {
2085         const char *use_term_factor_str =
2086             session_setting_oneval(sdb, PZ_TERMLIST_TERM_FACTOR);
2087         if (use_term_factor_str && use_term_factor_str[0] == '1')
2088         {
2089             int maxrecs = client_get_maxrecs(cl);
2090             int hits = (int) client_get_hits(cl);
2091             term_factor = MAX(hits, maxrecs) /  MAX(1, maxrecs);
2092             assert(term_factor >= 1);
2093             session_log(se, YLOG_DEBUG, "Using term factor: %d (%d / %d)",
2094                         term_factor, MAX(hits, maxrecs), MAX(1, maxrecs));
2095         }
2096     }
2097
2098     if (global_parameters.dump_records)
2099         session_log(se, YLOG_LOG, "Cluster id %s from %s (#%d)", cluster->recid,
2100                     sdb->database->id, record_no);
2101
2102     // original metadata, to check if first existence of a field
2103     metadata0 = xmalloc(sizeof(*metadata0) * service->num_metadata);
2104     memcpy(metadata0, cluster->metadata,
2105            sizeof(*metadata0) * service->num_metadata);
2106
2107     ingest_nmem = nmem_create();
2108     if (se->rank)
2109     {
2110         yaz_log(YLOG_LOG, "local in sort : %s", se->rank);
2111         nmem_strsplit_escape2(ingest_nmem, ",", se->rank, &rank_values,
2112                               &rank_num, 1, '\\', 1);
2113     }
2114
2115     // now parsing XML record and adding data to cluster or record metadata
2116     for (n = root->children; n; n = n->next)
2117     {
2118         pp2_charset_token_t prt;
2119         if (type)
2120             xmlFree(type);
2121         if (value)
2122             xmlFree(value);
2123         type = value = 0;
2124
2125         if (n->type != XML_ELEMENT_NODE)
2126             continue;
2127         if (!strcmp((const char *) n->name, "metadata"))
2128         {
2129             struct conf_metadata *ser_md = 0;
2130             struct conf_sortkey *ser_sk = 0;
2131             struct record_metadata **wheretoput = 0;
2132             struct record_metadata *rec_md = 0;
2133             int md_field_id = -1;
2134             int sk_field_id = -1;
2135             const char *rank = 0;
2136             xmlChar *xml_rank = 0;
2137
2138             type = xmlGetProp(n, (xmlChar *) "type");
2139             value = xmlNodeListGetString(xdoc, n->children, 1);
2140
2141             if (!type || !value || !*value)
2142                 continue;
2143
2144             md_field_id
2145                 = conf_service_metadata_field_id(service, (const char *) type);
2146             if (md_field_id < 0)
2147                 continue;
2148
2149             ser_md = &service->metadata[md_field_id];
2150
2151             if (ser_md->sortkey_offset >= 0)
2152             {
2153                 sk_field_id = ser_md->sortkey_offset;
2154                 ser_sk = &service->sortkeys[sk_field_id];
2155             }
2156
2157             // merged metadata
2158             rec_md = record_metadata_init(se->nmem, (const char *) value,
2159                                           ser_md->type, 0);
2160
2161             // see if the field was not in cluster already (from beginning)
2162
2163             if (!rec_md)
2164                 continue;
2165
2166             if (rank_num)
2167             {
2168                 int i;
2169                 for (i = 0; i < rank_num; i++)
2170                 {
2171                     const char *val = rank_values[i];
2172                     const char *cp = strchr(val, '=');
2173                     if (!cp)
2174                         continue;
2175                     if ((cp - val) == strlen((const char *) type)
2176                         && !memcmp(val, type, cp - val))
2177                     {
2178                         rank = cp + 1;
2179                         break;
2180                     }
2181                 }
2182             }
2183             else
2184             {
2185                 xml_rank = xmlGetProp(n, (xmlChar *) "rank");
2186                 rank = xml_rank ? (const char *) xml_rank : ser_md->rank;
2187             }
2188
2189             wheretoput = &cluster->metadata[md_field_id];
2190
2191             if (ser_md->merge == Metadata_merge_first)
2192             {
2193                 if (!metadata0[md_field_id])
2194                 {
2195                     while (*wheretoput)
2196                         wheretoput = &(*wheretoput)->next;
2197                     *wheretoput = rec_md;
2198                 }
2199             }
2200             else if (ser_md->merge == Metadata_merge_unique)
2201             {
2202                 while (*wheretoput)
2203                 {
2204                     if (!strcmp((const char *) (*wheretoput)->data.text.disp,
2205                                 rec_md->data.text.disp))
2206                         break;
2207                     wheretoput = &(*wheretoput)->next;
2208                 }
2209                 if (!*wheretoput)
2210                     *wheretoput = rec_md;
2211             }
2212             else if (ser_md->merge == Metadata_merge_longest)
2213             {
2214                 if (!*wheretoput
2215                     || strlen(rec_md->data.text.disp)
2216                     > strlen((*wheretoput)->data.text.disp))
2217                 {
2218                     *wheretoput = rec_md;
2219                     if (ser_sk)
2220                     {
2221                         const char *sort_str = 0;
2222                         int skip_article =
2223                             ser_sk->type == Metadata_type_skiparticle;
2224
2225                         if (!cluster->sortkeys[sk_field_id])
2226                             cluster->sortkeys[sk_field_id] =
2227                                 nmem_malloc(se->nmem,
2228                                             sizeof(union data_types));
2229
2230                         prt =
2231                             pp2_charset_token_create(service->charsets, "sort");
2232
2233                         pp2_charset_token_first(prt, rec_md->data.text.disp,
2234                                                 skip_article);
2235
2236                         pp2_charset_token_next(prt);
2237
2238                         sort_str = pp2_get_sort(prt);
2239
2240                         cluster->sortkeys[sk_field_id]->text.disp =
2241                             rec_md->data.text.disp;
2242                         if (!sort_str)
2243                         {
2244                             sort_str = rec_md->data.text.disp;
2245                             session_log(se, YLOG_WARN,
2246                                     "Could not make sortkey. Bug #1858");
2247                         }
2248                         cluster->sortkeys[sk_field_id]->text.sort =
2249                             nmem_strdup(se->nmem, sort_str);
2250                         pp2_charset_token_destroy(prt);
2251                     }
2252                 }
2253             }
2254             else if (ser_md->merge == Metadata_merge_all)
2255             {
2256                 while (*wheretoput)
2257                     wheretoput = &(*wheretoput)->next;
2258                 *wheretoput = rec_md;
2259             }
2260             else if (ser_md->merge == Metadata_merge_range)
2261             {
2262                 if (!*wheretoput)
2263                 {
2264                     *wheretoput = rec_md;
2265                     if (ser_sk)
2266                         cluster->sortkeys[sk_field_id]
2267                             = &rec_md->data;
2268                 }
2269                 else
2270                 {
2271                     int this_min = rec_md->data.number.min;
2272                     int this_max = rec_md->data.number.max;
2273                     if (this_min < (*wheretoput)->data.number.min)
2274                         (*wheretoput)->data.number.min = this_min;
2275                     if (this_max > (*wheretoput)->data.number.max)
2276                         (*wheretoput)->data.number.max = this_max;
2277                 }
2278             }
2279
2280             // ranking of _all_ fields enabled ...
2281             if (rank)
2282             {
2283                 relevance_countwords(se->relevance, cluster,
2284                                      (char *) value, rank, ser_md->name);
2285             }
2286             // construct facets ... unless the client already has reported them
2287             if (ser_md->termlist && !client_has_facet(cl, (char *) type))
2288             {
2289                 if (ser_md->type == Metadata_type_year)
2290                 {
2291                     char year[64];
2292                     sprintf(year, "%d", rec_md->data.number.max);
2293
2294                     add_facet(se, (char *) type, year, term_factor);
2295                     if (rec_md->data.number.max != rec_md->data.number.min)
2296                     {
2297                         sprintf(year, "%d", rec_md->data.number.min);
2298                         add_facet(se, (char *) type, year, term_factor);
2299                     }
2300                 }
2301                 else
2302                     add_facet(se, (char *) type, (char *) value, term_factor);
2303             }
2304
2305             // cleaning up
2306             if (xml_rank)
2307                 xmlFree(xml_rank);
2308             xmlFree(type);
2309             xmlFree(value);
2310             type = value = 0;
2311         }
2312         else
2313         {
2314             if (se->number_of_warnings_unknown_elements == 0)
2315                 session_log(se, YLOG_WARN,
2316                         "Unexpected element in internal record: %s", n->name);
2317             se->number_of_warnings_unknown_elements++;
2318         }
2319     }
2320     if (type)
2321         xmlFree(type);
2322     if (value)
2323         xmlFree(value);
2324
2325     nmem_destroy(ingest_nmem);
2326     xfree(metadata0);
2327     relevance_donerecord(se->relevance, cluster);
2328     se->total_records++;
2329
2330     return 0;
2331 }
2332
2333 void session_log(struct session *s, int level, const char *fmt, ...)
2334 {
2335     char buf[1024];
2336     va_list ap;
2337     va_start(ap, fmt);
2338
2339     yaz_vsnprintf(buf, sizeof(buf)-30, fmt, ap);
2340     yaz_log(level, "Session %u: %s", s ? s->session_id : 0, buf);
2341
2342     va_end(ap);
2343 }
2344
2345 /*
2346  * Local variables:
2347  * c-basic-offset: 4
2348  * c-file-style: "Stroustrup"
2349  * indent-tabs-mode: nil
2350  * End:
2351  * vim: shiftwidth=4 tabstop=8 expandtab
2352  */
2353