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 0
  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;
 750
 751     session_log(se, YLOG_DEBUG, "Search");
 752
 753     *addinfo = 0;
 754
 755     session_enter(se, "session_search0");
 756     if (se->clients_starting)
 757     {
 758         session_leave(se, "session_search0");
 759         return PAZPAR2_NO_ERROR;
 760     }
 761     se->clients_starting = 1;
 762     session_leave(se, "session_search0");
 763
 764     if (se->settings_modified) {
 765         session_remove_cached_clients(se);
 766     }
 767     else
 768         session_reset_active_clients(se, 0);
 769
 770     session_enter(se, "session_search");
 771     se->settings_modified = 0;
 772
 773     if (mergekey)
 774     {
 775         xfree(se->mergekey);
 776         se->mergekey = *mergekey ? xstrdup(mergekey) : 0;
 777     }
 778     if (rank)
 779     {
 780         xfree(se->rank);
 781         se->rank = *rank ? xstrdup(rank) : 0;
 782     }
 783
 784     session_clear_set(se, sp);
 785     relevance_destroy(&se->relevance);
 786
 787     live_channels = select_targets(se, filter);
 788     if (!live_channels)
 789     {
 790         session_leave(se, "session_search");
 791         se->clients_starting = 0;
 792         return PAZPAR2_NO_TARGETS;
 793     }
 794
 795     facet_limits_destroy(se->facet_limits);
 796     se->facet_limits = facet_limits_create(limit);
 797     if (!se->facet_limits)
 798     {
 799         *addinfo = "limit";
 800         session_leave(se, "session_search");
 801         se->clients_starting = 0;
 802         return PAZPAR2_MALFORMED_PARAMETER_VALUE;
 803     }
 804
 805     session_leave(se, "session_search");
 806
 807     session_alert_watch(se, SESSION_WATCH_SHOW);
 808     session_alert_watch(se, SESSION_WATCH_BYTARGET);
 809     session_alert_watch(se, SESSION_WATCH_TERMLIST);
 810     session_alert_watch(se, SESSION_WATCH_SHOW_PREF);
 811
 812     for (l = se->clients_active; l; l = l->next)
 813     {
 814         int parse_ret;
 815         struct client *cl = l->client;
 816         client_parse_init(cl, 1);
 817         if (prepare_map(se, client_get_database(cl)) < 0)
 818             continue;
 819
 820         parse_ret = client_parse_query(cl, query, se->facet_limits, addinfo2);
 821         if (parse_ret == -1)
 822             no_failed_query++;
 823         else if (parse_ret == -2)
 824             no_failed_limit++;
 825         else if (parse_ret < 0)
 826             no_working++; /* other error, such as bad CCL map */
 827         else
 828         {
 829             client_parse_range(cl, startrecs, maxrecs);
 830             client_parse_sort(cl, sp);
 831             client_start_search(cl);
 832             no_working++;
 833         }
 834     }
 835     session_enter(se, "session_search2");
 836     se->clients_starting = 0;
 837     session_leave(se, "session_search2");
 838     if (no_working == 0)
 839     {
 840         if (no_failed_query > 0)
 841         {
 842             *addinfo = "query";
 843             return PAZPAR2_MALFORMED_PARAMETER_VALUE;
 844         }
 845         else if (no_failed_limit > 0)
 846         {
 847             *addinfo = "limit";
 848             return PAZPAR2_MALFORMED_PARAMETER_VALUE;
 849         }
 850         else
 851             return PAZPAR2_NO_TARGETS;
 852     }
 853     return PAZPAR2_NO_ERROR;
 854 }
 855
 856 // Creates a new session_database object for a database
 857 static void session_init_databases_fun(void *context, struct database *db)
 858 {
 859     struct session *se = (struct session *) context;
 860     struct session_database *new = nmem_malloc(se->session_nmem, sizeof(*new));
 861     int i;
 862
 863     new->database = db;
 864
 865     new->map = 0;
 866     assert(db->settings);
 867     new->settings = nmem_malloc(se->session_nmem,
 868                                 sizeof(struct settings *) * db->num_settings);
 869     new->num_settings = db->num_settings;
 870     for (i = 0; i < db->num_settings; i++)
 871     {
 872         struct setting *setting = db->settings[i];
 873         new->settings[i] = setting;
 874     }
 875     new->next = se->databases;
 876     se->databases = new;
 877 }
 878
 879 // Doesn't free memory associated with sdb -- nmem takes care of that
 880 static void session_database_destroy(struct session_database *sdb)
 881 {
 882     sdb->map = 0;
 883 }
 884
 885 // Initialize session_database list -- this represents this session's view
 886 // of the database list -- subject to modification by the settings ws command
 887 void session_init_databases(struct session *se)
 888 {
 889     se->databases = 0;
 890     predef_grep_databases(se, se->service, session_init_databases_fun);
 891 }
 892
 893 // Probably session_init_databases_fun should be refactored instead of
 894 // called here.
 895 static struct session_database *load_session_database(struct session *se,
 896                                                       const char *id)
 897 {
 898     struct database *db = new_database_inherit_settings(id, se->session_nmem, se->service->settings);
 899     session_init_databases_fun((void*) se, db);
 900
 901     // New sdb is head of se->databases list
 902     return se->databases;
 903 }
 904
 905 // Find an existing session database. If not found, load it
 906 static struct session_database *find_session_database(struct session *se,
 907                                                       const char *id)
 908 {
 909     struct session_database *sdb;
 910
 911     for (sdb = se->databases; sdb; sdb = sdb->next)
 912         if (!strcmp(sdb->database->id, id))
 913             return sdb;
 914     return load_session_database(se, id);
 915 }
 916
 917 // Apply a session override to a database
 918 void session_apply_setting(struct session *se, const char *dbname,
 919                            const char *name, const char *value)
 920 {
 921     session_enter(se, "session_apply_setting");
 922     {
 923         struct session_database *sdb = find_session_database(se, dbname);
 924         struct conf_service *service = se->service;
 925         struct setting *s;
 926         int offset = settings_create_offset(service, name);
 927
 928         expand_settings_array(&sdb->settings, &sdb->num_settings, offset,
 929                               se->session_nmem);
 930         // Force later recompute of settings-driven data structures
 931         // (happens when a search starts and client connections are prepared)
 932         if (offset == PZ_XSLT)
 933             sdb->map = 0;
 934         se->settings_modified = 1;
 935         for (s = sdb->settings[offset]; s; s = s->next)
 936             if (!strcmp(s->name, name) &&
 937                 dbname && s->target && !strcmp(dbname, s->target))
 938                 break;
 939         if (!s)
 940         {
 941             s = nmem_malloc(se->session_nmem, sizeof(*s));
 942             s->precedence = 0;
 943             s->target = nmem_strdup(se->session_nmem, dbname);
 944             s->name = nmem_strdup(se->session_nmem, name);
 945             s->next = sdb->settings[offset];
 946             sdb->settings[offset] = s;
 947         }
 948         s->value = nmem_strdup(se->session_nmem, value);
 949     }
 950     session_leave(se, "session_apply_setting");
 951 }
 952
 953 void session_destroy(struct session *se)
 954 {
 955     struct session_database *sdb;
 956     session_log(se, YLOG_LOG, "destroy");
 957     session_use(-1);
 958     session_remove_cached_clients(se);
 959
 960     for (sdb = se->databases; sdb; sdb = sdb->next)
 961         session_database_destroy(sdb);
 962     normalize_cache_destroy(se->normalize_cache);
 963     relevance_destroy(&se->relevance);
 964     reclist_destroy(se->reclist);
 965     xfree(se->mergekey);
 966     xfree(se->rank);
 967     if (nmem_total(se->nmem))
 968         session_log(se, YLOG_DEBUG, "NMEN operation usage %zd", nmem_total(se->nmem));
 969     if (nmem_total(se->session_nmem))
 970         session_log(se, YLOG_DEBUG, "NMEN session usage %zd", nmem_total(se->session_nmem));
 971     facet_limits_destroy(se->facet_limits);
 972     nmem_destroy(se->nmem);
 973     service_destroy(se->service);
 974     yaz_mutex_destroy(&se->session_mutex);
 975 }
 976
 977 size_t session_get_memory_status(struct session *session) {
 978     size_t session_nmem;
 979     if (session == 0)
 980         return 0;
 981     session_enter(session, "session_get_memory_status");
 982     session_nmem = nmem_total(session->nmem);
 983     session_leave(session, "session_get_memory_status");
 984     return session_nmem;
 985 }
 986
 987
 988 struct session *new_session(NMEM nmem, struct conf_service *service,
 989                             unsigned session_id)
 990 {
 991     int i;
 992     struct session *session = nmem_malloc(nmem, sizeof(*session));
 993
 994     char tmp_str[50];
 995
 996     sprintf(tmp_str, "session#%u", session_id);
 997
 998     session->session_id = session_id;
 999     session_log(session, YLOG_DEBUG, "New");
1000     session->service = service;
1001     session->relevance = 0;
1002     session->total_records = 0;
1003     session->number_of_warnings_unknown_elements = 0;
1004     session->number_of_warnings_unknown_metadata = 0;
1005     session->num_termlists = 0;
1006     session->reclist = reclist_create(nmem);
1007     session->clients_active = 0;
1008     session->clients_cached = 0;
1009     session->settings_modified = 0;
1010     session->session_nmem = nmem;
1011     session->nmem = nmem_create();
1012     session->databases = 0;
1013     session->sorted_results = 0;
1014     session->facet_limits = 0;
1015     session->mergekey = 0;
1016     session->rank = 0;
1017     session->clients_starting = 0;
1018
1019     for (i = 0; i <= SESSION_WATCH_MAX; i++)
1020     {
1021         session->watchlist[i].data = 0;
1022         session->watchlist[i].fun = 0;
1023     }
1024     session->normalize_cache = normalize_cache_create();
1025     session->session_mutex = 0;
1026     pazpar2_mutex_create(&session->session_mutex, tmp_str);
1027     session_log(session, YLOG_LOG, "create");
1028
1029     session_use(1);
1030     return session;
1031 }
1032
1033 const char * client_get_suggestions_xml(struct client *cl, WRBUF wrbuf);
1034
1035 static struct hitsbytarget *hitsbytarget_nb(struct session *se,
1036                                             int *count, NMEM nmem)
1037 {
1038     struct hitsbytarget *res = 0;
1039     struct client_list *l;
1040     size_t sz = 0;
1041
1042     for (l = se->clients_active; l; l = l->next)
1043         sz++;
1044
1045     res = nmem_malloc(nmem, sizeof(*res) * sz);
1046     *count = 0;
1047     for (l = se->clients_active; l; l = l->next)
1048     {
1049         struct client *cl = l->client;
1050         WRBUF w = wrbuf_alloc();
1051         const char *name = session_setting_oneval(client_get_database(cl),
1052                                                   PZ_NAME);
1053         res[*count].id = client_get_id(cl);
1054         res[*count].name = *name ? name : "Unknown";
1055         res[*count].hits = client_get_hits(cl);
1056         res[*count].approximation = client_get_approximation(cl);
1057         res[*count].records = client_get_num_records(cl,
1058                                                      &res[*count].filtered,
1059                                                      0, 0);
1060         res[*count].diagnostic =
1061             client_get_diagnostic(cl, &res[*count].message,
1062                                   &res[*count].addinfo);
1063         res[*count].state = client_get_state_str(cl);
1064         res[*count].connected  = client_get_connection(cl) ? 1 : 0;
1065         session_settings_dump(se, client_get_database(cl), w);
1066         res[*count].settings_xml = nmem_strdup(nmem, wrbuf_cstr(w));
1067         wrbuf_rewind(w);
1068         wrbuf_puts(w, "");
1069         res[*count].suggestions_xml = nmem_strdup(nmem, client_get_suggestions_xml(cl, w));
1070         wrbuf_destroy(w);
1071         (*count)++;
1072     }
1073     return res;
1074 }
1075
1076 struct hitsbytarget *get_hitsbytarget(struct session *se, int *count, NMEM nmem)
1077 {
1078     struct hitsbytarget *p;
1079     session_enter(se, "get_hitsbytarget");
1080     p = hitsbytarget_nb(se, count, nmem);
1081     session_leave(se, "get_hitsbytarget");
1082     return p;
1083 }
1084
1085 // Compares two hitsbytarget nodes by hitcount
1086 static int cmp_ht(const void *p1, const void *p2)
1087 {
1088     const struct hitsbytarget *h1 = p1;
1089     const struct hitsbytarget *h2 = p2;
1090     return h2->hits - h1->hits;
1091 }
1092
1093 // Compares two hitsbytarget nodes by hitcount
1094 static int cmp_ht_approx(const void *p1, const void *p2)
1095 {
1096     const struct hitsbytarget *h1 = p1;
1097     const struct hitsbytarget *h2 = p2;
1098     return h2->approximation - h1->approximation;
1099 }
1100
1101 static int targets_termlist_nb(WRBUF wrbuf, struct session *se, int num,
1102                                NMEM nmem, int version)
1103 {
1104     struct hitsbytarget *ht;
1105     int count, i;
1106
1107     ht = hitsbytarget_nb(se, &count, nmem);
1108     if (version >= 2)
1109         qsort(ht, count, sizeof(struct hitsbytarget), cmp_ht_approx);
1110     else
1111         qsort(ht, count, sizeof(struct hitsbytarget), cmp_ht);
1112     for (i = 0; i < count && i < num && ht[i].hits > 0; i++)
1113     {
1114
1115         // do only print terms which have display names
1116
1117         wrbuf_puts(wrbuf, "<term>\n");
1118
1119         wrbuf_puts(wrbuf, "<id>");
1120         wrbuf_xmlputs(wrbuf, ht[i].id);
1121         wrbuf_puts(wrbuf, "</id>\n");
1122
1123         wrbuf_puts(wrbuf, "<name>");
1124         if (!ht[i].name || !ht[i].name[0])
1125             wrbuf_xmlputs(wrbuf, "NO TARGET NAME");
1126         else
1127             wrbuf_xmlputs(wrbuf, ht[i].name);
1128         wrbuf_puts(wrbuf, "</name>\n");
1129
1130         wrbuf_printf(wrbuf, "<frequency>" ODR_INT_PRINTF "</frequency>\n",
1131                      ht[i].hits);
1132
1133         if (version >= 2) {
1134             // Should not print if we know it isn't a approximation.
1135             wrbuf_printf(wrbuf, "<approximation>" ODR_INT_PRINTF "</approximation>\n", ht[i].approximation);
1136             wrbuf_printf(wrbuf, "<records>%d</records>\n", ht[i].records - ht[i].filtered);
1137             wrbuf_printf(wrbuf, "<filtered>%d</filtered>\n", ht[i].filtered);
1138         }
1139
1140         wrbuf_puts(wrbuf, "<state>");
1141         wrbuf_xmlputs(wrbuf, ht[i].state);
1142         wrbuf_puts(wrbuf, "</state>\n");
1143
1144         wrbuf_printf(wrbuf, "<diagnostic>%d</diagnostic>\n",
1145                      ht[i].diagnostic);
1146         wrbuf_puts(wrbuf, "</term>\n");
1147     }
1148     return count;
1149 }
1150
1151 void perform_termlist(struct http_channel *c, struct session *se,
1152                       const char *name, int num, int version)
1153 {
1154     int i, j;
1155     NMEM nmem_tmp = nmem_create();
1156     char **names;
1157     int num_names = 0;
1158
1159     if (!name)
1160         name = "*";
1161
1162     nmem_strsplit(nmem_tmp, ",", name, &names, &num_names);
1163
1164     session_enter(se, "perform_termlist");
1165
1166     for (j = 0; j < num_names; j++)
1167     {
1168         const char *tname;
1169         int must_generate_empty = 1; /* bug 5350 */
1170
1171         for (i = 0; i < se->num_termlists; i++)
1172         {
1173             tname = se->termlists[i].name;
1174             if (!strcmp(names[j], tname) || !strcmp(names[j], "*"))
1175             {
1176                 struct termlist_score **p = 0;
1177                 int len;
1178
1179                 wrbuf_puts(c->wrbuf, "<list name=\"");
1180                 wrbuf_xmlputs(c->wrbuf, tname);
1181                 wrbuf_puts(c->wrbuf, "\">\n");
1182                 must_generate_empty = 0;
1183
1184                 p = termlist_highscore(se->termlists[i].termlist, &len,
1185                                        nmem_tmp);
1186                 if (p)
1187                 {
1188                     int i;
1189                     for (i = 0; i < len && i < num; i++)
1190                     {
1191                         // prevent sending empty term elements
1192                         if (!p[i]->display_term || !p[i]->display_term[0])
1193                             continue;
1194
1195                         wrbuf_puts(c->wrbuf, "<term>");
1196                         wrbuf_puts(c->wrbuf, "<name>");
1197                         wrbuf_xmlputs(c->wrbuf, p[i]->display_term);
1198                         wrbuf_puts(c->wrbuf, "</name>");
1199
1200                         wrbuf_printf(c->wrbuf,
1201                                      "<frequency>%d</frequency>",
1202                                      p[i]->frequency);
1203                         wrbuf_puts(c->wrbuf, "</term>\n");
1204                     }
1205                 }
1206                 wrbuf_puts(c->wrbuf, "</list>\n");
1207             }
1208         }
1209         tname = "xtargets";
1210         if (!strcmp(names[j], tname) || !strcmp(names[j], "*"))
1211         {
1212             wrbuf_puts(c->wrbuf, "<list name=\"");
1213             wrbuf_xmlputs(c->wrbuf, tname);
1214             wrbuf_puts(c->wrbuf, "\">\n");
1215
1216             targets_termlist_nb(c->wrbuf, se, num, c->nmem, version);
1217             wrbuf_puts(c->wrbuf, "</list>\n");
1218             must_generate_empty = 0;
1219         }
1220         if (must_generate_empty)
1221         {
1222             wrbuf_puts(c->wrbuf, "<list name=\"");
1223             wrbuf_xmlputs(c->wrbuf, names[j]);
1224             wrbuf_puts(c->wrbuf, "\"/>\n");
1225         }
1226     }
1227     session_leave(se, "perform_termlist");
1228     nmem_destroy(nmem_tmp);
1229 }
1230
1231 #ifdef MISSING_HEADERS
1232 void report_nmem_stats(void)
1233 {
1234     size_t in_use, is_free;
1235
1236     nmem_get_memory_in_use(&in_use);
1237     nmem_get_memory_free(&is_free);
1238
1239     yaz_log(YLOG_LOG, "nmem stat: use=%ld free=%ld",
1240             (long) in_use, (long) is_free);
1241 }
1242 #endif
1243
1244 struct record_cluster *show_single_start(struct session *se, const char *id,
1245                                          struct record_cluster **prev_r,
1246                                          struct record_cluster **next_r)
1247 {
1248     struct record_cluster *r = 0;
1249
1250     session_enter(se, "show_single_start");
1251     *prev_r = 0;
1252     *next_r = 0;
1253     reclist_limit(se->reclist, se, 1);
1254
1255     reclist_enter(se->reclist);
1256     while ((r = reclist_read_record(se->reclist)))
1257     {
1258         if (!strcmp(r->recid, id))
1259         {
1260             *next_r = reclist_read_record(se->reclist);
1261             break;
1262         }
1263         *prev_r = r;
1264     }
1265     reclist_leave(se->reclist);
1266     if (!r)
1267         session_leave(se, "show_single_start");
1268     return r;
1269 }
1270
1271 void show_single_stop(struct session *se, struct record_cluster *rec)
1272 {
1273     session_leave(se, "show_single_stop");
1274 }
1275
1276
1277 int session_fetch_more(struct session *se)
1278 {
1279     struct client_list *l;
1280     int ret = 0;
1281
1282     for (l = se->clients_active; l; l = l->next)
1283     {
1284         struct client *cl = l->client;
1285         if (client_get_state(cl) == Client_Idle)
1286         {
1287             if (client_fetch_more(cl))
1288             {
1289                 session_log(se, YLOG_LOG, "%s: more to fetch",
1290                             client_get_id(cl));
1291                 ret = 1;
1292             }
1293             else
1294             {
1295                 int filtered;
1296                 int ingest_failures;
1297                 int record_failures;
1298                 int num = client_get_num_records(
1299                     cl, &filtered, &ingest_failures, &record_failures);
1300
1301                 session_log(se, YLOG_LOG, "%s: hits=" ODR_INT_PRINTF
1302                             " fetched=%d filtered=%d",
1303                             client_get_id(cl),
1304                             client_get_hits(cl),
1305                             num, filtered);
1306                 if (ingest_failures || record_failures)
1307                 {
1308                     session_log(se, YLOG_WARN, "%s:"
1309                                 " ingest failures=%d record failures=%d",
1310                                 client_get_id(cl),
1311                                 ingest_failures, record_failures);
1312                 }
1313             }
1314         }
1315         else
1316         {
1317             session_log(se, YLOG_LOG, "%s: no fetch due to state=%s",
1318                         client_get_id(cl), client_get_state_str(cl));
1319         }
1320
1321     }
1322     return ret;
1323 }
1324
1325 struct record_cluster **show_range_start(struct session *se,
1326                                          struct reclist_sortparms *sp,
1327                                          int start, int *num, int *total,
1328                                          Odr_int *sumhits, Odr_int *approx_hits,
1329                                          void (*show_records_ready)(void *data),
1330                                          struct http_channel *chan)
1331 {
1332     struct record_cluster **recs = 0;
1333     struct reclist_sortparms *spp;
1334     struct client_list *l;
1335     int i;
1336 #if USE_TIMING
1337     yaz_timing_t t = yaz_timing_create();
1338 #endif
1339     session_enter(se, "show_range_start");
1340     *sumhits = 0;
1341     *approx_hits = 0;
1342     *total = 0;
1343     reclist_limit(se->reclist, se, 0);
1344     if (se->relevance)
1345     {
1346         for (spp = sp; spp; spp = spp->next)
1347             if (spp->type == Metadata_type_relevance)
1348             {
1349                 relevance_prepare_read(se->relevance, se->reclist);
1350                 break;
1351             }
1352         for (l = se->clients_active; l; l = l->next) {
1353             *sumhits += client_get_hits(l->client);
1354             *approx_hits += client_get_approximation(l->client);
1355         }
1356     }
1357     reclist_sort(se->reclist, sp);
1358
1359     reclist_enter(se->reclist);
1360     *total = reclist_get_num_records(se->reclist);
1361
1362     for (l = se->clients_active; l; l = l->next)
1363         client_update_show_stat(l->client, 0);
1364
1365     for (i = 0; i < start; i++)
1366     {
1367         struct record_cluster *r = reclist_read_record(se->reclist);
1368         if (!r)
1369         {
1370             *num = 0;
1371             break;
1372         }
1373         else
1374         {
1375             struct record *rec = r->records;
1376             for (;rec; rec = rec->next)
1377                 client_update_show_stat(rec->client, 1);
1378         }
1379     }
1380     recs = nmem_malloc(se->nmem, (*num > 0 ? *num : 1) * sizeof(*recs));
1381     for (i = 0; i < *num; i++)
1382     {
1383         struct record_cluster *r = reclist_read_record(se->reclist);
1384         if (!r)
1385         {
1386             *num = i;
1387             break;
1388         }
1389         else
1390         {
1391             struct record *rec = r->records;
1392             for (;rec; rec = rec->next)
1393                 client_update_show_stat(rec->client, 1);
1394             recs[i] = r;
1395         }
1396     }
1397     reclist_leave(se->reclist);
1398 #if USE_TIMING
1399     yaz_timing_stop(t);
1400     session_log(se, YLOG_LOG, "show %6.5f %3.2f %3.2f",
1401             yaz_timing_get_real(t), yaz_timing_get_user(t),
1402             yaz_timing_get_sys(t));
1403     yaz_timing_destroy(&t);
1404 #endif
1405
1406     if (!session_fetch_more(se))
1407         session_log(se, YLOG_LOG, "can not fetch more");
1408     else
1409     {
1410         show_range_stop(se, recs);
1411         session_log(se, YLOG_LOG, "fetching more in progress");
1412         if (session_set_watch(se, SESSION_WATCH_SHOW,
1413                               show_records_ready, chan, chan))
1414         {
1415             session_log(se, YLOG_WARN, "Ignoring show block");
1416             session_enter(se, "show_range_start");
1417         }
1418         else
1419         {
1420             session_log(se, YLOG_LOG, "session watch OK");
1421             return 0;
1422         }
1423     }
1424     return recs;
1425 }
1426
1427 void show_range_stop(struct session *se, struct record_cluster **recs)
1428 {
1429     session_leave(se, "show_range_stop");
1430 }
1431
1432 void statistics(struct session *se, struct statistics *stat)
1433 {
1434     struct client_list *l;
1435     int count = 0;
1436
1437     memset(stat, 0, sizeof(*stat));
1438     stat->num_hits = 0;
1439     for (l = se->clients_active; l; l = l->next)
1440     {
1441         struct client *cl = l->client;
1442         if (!client_get_connection(cl))
1443             stat->num_no_connection++;
1444         stat->num_hits += client_get_hits(cl);
1445         switch (client_get_state(cl))
1446         {
1447         case Client_Connecting: stat->num_connecting++; break;
1448         case Client_Working: stat->num_working++; break;
1449         case Client_Idle: stat->num_idle++; break;
1450         case Client_Failed: stat->num_failed++; break;
1451         case Client_Error: stat->num_error++; break;
1452         default: break;
1453         }
1454         count++;
1455     }
1456     stat->num_records = se->total_records;
1457
1458     stat->num_clients = count;
1459 }
1460
1461 static struct record_metadata *record_metadata_init(
1462     NMEM nmem, const char *value, enum conf_metadata_type type,
1463     struct _xmlAttr *attr)
1464 {
1465     struct record_metadata *rec_md = record_metadata_create(nmem);
1466     struct record_metadata_attr **attrp = &rec_md->attributes;
1467
1468     for (; attr; attr = attr->next)
1469     {
1470         if (attr->children && attr->children->content)
1471         {
1472             if (strcmp((const char *) attr->name, "type")
1473                 && strcmp((const char *) attr->name, "empty"))
1474             {  /* skip the "type" + "empty" attribute..
1475                   The "Type" is already part of the element in output
1476                   (md-%s) and so repeating it here is redundant */
1477                 *attrp = nmem_malloc(nmem, sizeof(**attrp));
1478                 (*attrp)->name =
1479                     nmem_strdup(nmem, (const char *) attr->name);
1480                 (*attrp)->value =
1481                     nmem_strdup(nmem, (const char *) attr->children->content);
1482                 attrp = &(*attrp)->next;
1483             }
1484         }
1485     }
1486     *attrp = 0;
1487
1488     switch (type)
1489     {
1490     case Metadata_type_generic:
1491     case Metadata_type_skiparticle:
1492         if (strstr(value, "://")) /* looks like a URL */
1493             rec_md->data.text.disp = nmem_strdup(nmem, value);
1494         else
1495             rec_md->data.text.disp =
1496                 normalize7bit_generic(nmem_strdup(nmem, value), " ,/.:([");
1497         rec_md->data.text.sort = 0;
1498         rec_md->data.text.snippet = 0;
1499         break;
1500     case Metadata_type_year:
1501     case Metadata_type_date:
1502     {
1503         int first, last;
1504         int longdate = 0;
1505
1506         if (type == Metadata_type_date)
1507             longdate = 1;
1508         if (extract7bit_dates((char *) value, &first, &last, longdate) < 0)
1509             return 0;
1510
1511         rec_md->data.number.min = first;
1512         rec_md->data.number.max = last;
1513     }
1514     break;
1515     case Metadata_type_float:
1516         rec_md->data.fnumber = atof(value);
1517         break;
1518     case Metadata_type_relevance:
1519     case Metadata_type_position:
1520         return 0;
1521     }
1522     return rec_md;
1523 }
1524
1525 static void mergekey_norm_wr(pp2_charset_fact_t charsets,
1526                              WRBUF norm_wr, const char *value)
1527 {
1528     const char *norm_str;
1529     pp2_charset_token_t prt =
1530         pp2_charset_token_create(charsets, "mergekey");
1531
1532     pp2_charset_token_first(prt, value, 0);
1533     while ((norm_str = pp2_charset_token_next(prt)))
1534     {
1535         if (*norm_str)
1536         {
1537             if (wrbuf_len(norm_wr))
1538                 wrbuf_puts(norm_wr, " ");
1539             wrbuf_puts(norm_wr, norm_str);
1540         }
1541     }
1542     pp2_charset_token_destroy(prt);
1543 }
1544
1545 static int get_mergekey_from_doc(xmlDoc *doc, xmlNode *root, const char *name,
1546                                  struct conf_service *service, WRBUF norm_wr)
1547 {
1548     xmlNode *n;
1549     int no_found = 0;
1550     for (n = root->children; n; n = n->next)
1551     {
1552         if (n->type != XML_ELEMENT_NODE)
1553             continue;
1554         if (!strcmp((const char *) n->name, "metadata"))
1555         {
1556             xmlChar *type = xmlGetProp(n, (xmlChar *) "type");
1557             if (type == NULL) {
1558                 yaz_log(YLOG_FATAL, "Missing type attribute on metadata element. Skipping!");
1559             }
1560             else if (!strcmp(name, (const char *) type))
1561             {
1562                 xmlChar *value = xmlNodeListGetString(doc, n->children, 1);
1563                 if (value && *value)
1564                 {
1565                     if (wrbuf_len(norm_wr) > 0)
1566                         wrbuf_puts(norm_wr, " ");
1567                     wrbuf_puts(norm_wr, name);
1568                     mergekey_norm_wr(service->charsets, norm_wr,
1569                                      (const char *) value);
1570                     no_found++;
1571                 }
1572                 if (value)
1573                     xmlFree(value);
1574             }
1575             xmlFree(type);
1576         }
1577     }
1578     return no_found;
1579 }
1580
1581 static const char *get_mergekey(xmlDoc *doc, xmlNode *root,
1582                                 struct client *cl, int record_no,
1583                                 struct conf_service *service, NMEM nmem,
1584                                 const char *session_mergekey)
1585 {
1586     char *mergekey_norm = 0;
1587     WRBUF norm_wr = wrbuf_alloc();
1588     xmlChar *mergekey;
1589
1590     if (session_mergekey)
1591     {
1592         int i, num = 0;
1593         char **values = 0;
1594         nmem_strsplit_escape2(nmem, ",", session_mergekey, &values,
1595                               &num, 1, '\\', 1);
1596
1597         for (i = 0; i < num; i++)
1598             get_mergekey_from_doc(doc, root, values[i], service, norm_wr);
1599     }
1600     else if ((mergekey = xmlGetProp(root, (xmlChar *) "mergekey")))
1601     {
1602         mergekey_norm_wr(service->charsets, norm_wr, (const char *) mergekey);
1603         xmlFree(mergekey);
1604     }
1605     else
1606     {
1607         /* no mergekey defined in XSL. Look for mergekey metadata instead */
1608         int field_id;
1609         for (field_id = 0; field_id < service->num_metadata; field_id++)
1610         {
1611             struct conf_metadata *ser_md = &service->metadata[field_id];
1612             if (ser_md->mergekey != Metadata_mergekey_no)
1613             {
1614                 int r = get_mergekey_from_doc(doc, root, ser_md->name,
1615                                               service, norm_wr);
1616                 if (r == 0 && ser_md->mergekey == Metadata_mergekey_required)
1617                 {
1618                     /* no mergekey on this one and it is required..
1619                        Generate unique key instead */
1620                     wrbuf_rewind(norm_wr);
1621                     break;
1622                 }
1623             }
1624         }
1625     }
1626
1627     /* generate unique key if none is not generated already or is empty */
1628     if (wrbuf_len(norm_wr) == 0)
1629     {
1630         wrbuf_printf(norm_wr, "position: %s-%d",
1631                      client_get_id(cl), record_no);
1632     }
1633     else
1634     {
1635         const char *lead = "content: ";
1636         wrbuf_insert(norm_wr, 0, lead, strlen(lead));
1637     }
1638     if (wrbuf_len(norm_wr) > 0)
1639         mergekey_norm = nmem_strdup(nmem, wrbuf_cstr(norm_wr));
1640     wrbuf_destroy(norm_wr);
1641     return mergekey_norm;
1642 }
1643
1644 /** \brief see if metadata for pz:recordfilter exists
1645     \param root xml root element of normalized record
1646     \param sdb session database for client
1647     \retval 0 if there is no metadata for pz:recordfilter
1648     \retval 1 if there is metadata for pz:recordfilter
1649
1650     If there is no pz:recordfilter defined, this function returns 1
1651     as well.
1652 */
1653
1654 static int check_record_filter(xmlNode *root, struct session_database *sdb)
1655 {
1656     int match = 0;
1657     xmlNode *n;
1658     const char *s;
1659     s = session_setting_oneval(sdb, PZ_RECORDFILTER);
1660
1661     if (!s || !*s)
1662         return 1;
1663
1664     for (n = root->children; n; n = n->next)
1665     {
1666         if (n->type != XML_ELEMENT_NODE)
1667             continue;
1668         if (!strcmp((const char *) n->name, "metadata"))
1669         {
1670             xmlChar *type = xmlGetProp(n, (xmlChar *) "type");
1671             if (type)
1672             {
1673                 size_t len;
1674                 int substring;
1675                 const char *eq;
1676
1677                 if ((eq = strchr(s, '=')))
1678                     substring = 0;
1679                 else if ((eq = strchr(s, '~')))
1680                     substring = 1;
1681                 if (eq)
1682                     len = eq - s;
1683                 else
1684                     len = strlen(s);
1685                 if (len == strlen((const char *)type) &&
1686                     !memcmp((const char *) type, s, len))
1687                 {
1688                     xmlChar *value = xmlNodeGetContent(n);
1689                     if (value && *value)
1690                     {
1691                         if (!eq ||
1692                             (substring && strstr((const char *) value, eq+1)) ||
1693                             (!substring && !strcmp((const char *) value, eq + 1)))
1694                             match = 1;
1695                     }
1696                     xmlFree(value);
1697                 }
1698                 xmlFree(type);
1699             }
1700         }
1701     }
1702     return match;
1703 }
1704
1705 static int ingest_to_cluster(struct client *cl,
1706                              xmlDoc *xdoc,
1707                              xmlNode *root,
1708                              int record_no,
1709                              struct record_metadata_attr *mergekey);
1710
1711 static int ingest_sub_record(struct client *cl, xmlDoc *xdoc, xmlNode *root,
1712                              int record_no, NMEM nmem,
1713                              struct session_database *sdb,
1714                              struct record_metadata_attr *mergekeys)
1715 {
1716     int ret = 0;
1717     struct session *se = client_get_session(cl);
1718
1719     if (!check_record_filter(root, sdb))
1720     {
1721         session_log(se, YLOG_LOG,
1722                     "Filtered out record no %d from %s",
1723                     record_no, sdb->database->id);
1724         return 0;
1725     }
1726     session_enter(se, "ingest_sub_record");
1727     if (client_get_session(cl) == se && se->relevance)
1728         ret = ingest_to_cluster(cl, xdoc, root, record_no, mergekeys);
1729     session_leave(se, "ingest_sub_record");
1730
1731     return ret;
1732 }
1733
1734 /** \brief ingest XML record
1735     \param cl client holds the result set for record
1736     \param rec record buffer (0 terminated)
1737     \param record_no record position (1, 2, ..)
1738     \param nmem working NMEM
1739     \retval 0 OK
1740     \retval -1 failure
1741     \retval -2 Filtered
1742 */
1743 int ingest_record(struct client *cl, const char *rec,
1744                   int record_no, NMEM nmem)
1745 {
1746     struct session *se = client_get_session(cl);
1747     struct session_database *sdb = client_get_database(cl);
1748     struct conf_service *service = se->service;
1749     xmlDoc *xdoc = normalize_record(se, sdb, service, rec, nmem);
1750     int r = ingest_xml_record(cl, xdoc, record_no, nmem, 0);
1751     client_store_xdoc(cl, record_no, xdoc);
1752     return r;
1753 }
1754
1755 int ingest_xml_record(struct client *cl, xmlDoc *xdoc,
1756                       int record_no, NMEM nmem, int cached_copy)
1757 {
1758     struct session *se = client_get_session(cl);
1759     struct session_database *sdb = client_get_database(cl);
1760     struct conf_service *service = se->service;
1761     xmlNode *root;
1762     int r = 0;
1763     if (!xdoc)
1764         return -1;
1765
1766     if (global_parameters.dump_records)
1767     {
1768         session_log(se, YLOG_LOG, "Normalized record from %s",
1769                     sdb->database->id);
1770         log_xml_doc(xdoc);
1771     }
1772
1773     root = xmlDocGetRootElement(xdoc);
1774
1775     if (!strcmp((const char *) root->name, "cluster"))
1776     {
1777         int no_merge_keys = 0;
1778         int no_merge_dups = 0;
1779         xmlNode *sroot;
1780         struct record_metadata_attr *mk = 0;
1781
1782         for (sroot = root->children; sroot; sroot = sroot->next)
1783             if (sroot->type == XML_ELEMENT_NODE &&
1784                 !strcmp((const char *) sroot->name, "record"))
1785             {
1786                 struct record_metadata_attr **mkp;
1787                 const char *mergekey_norm =
1788                     get_mergekey(xdoc, sroot, cl, record_no, service, nmem,
1789                                  se->mergekey);
1790                 if (!mergekey_norm)
1791                 {
1792                     r = -1;
1793                     break;
1794                 }
1795                 for (mkp = &mk; *mkp; mkp = &(*mkp)->next)
1796                     if (!strcmp((*mkp)->value, mergekey_norm))
1797                         break;
1798                 if (!*mkp)
1799                 {
1800                     *mkp = (struct record_metadata_attr*)
1801                         nmem_malloc(nmem, sizeof(**mkp));
1802                     (*mkp)->name = 0;
1803                     (*mkp)->value = nmem_strdup(nmem, mergekey_norm);
1804                     (*mkp)->next = 0;
1805                     no_merge_keys++;
1806                 }
1807                 else
1808                     no_merge_dups++;
1809             }
1810         if (no_merge_keys > 1 || no_merge_dups > 0)
1811         {
1812             yaz_log(YLOG_LOG, "Got %d mergekeys, %d dups for position %d",
1813                     no_merge_keys, no_merge_dups, record_no);
1814         }
1815         for (sroot = root->children; !r && sroot; sroot = sroot->next)
1816             if (sroot->type == XML_ELEMENT_NODE &&
1817                 !strcmp((const char *) sroot->name, "record"))
1818             {
1819                 if (!cached_copy)
1820                     insert_settings_values(sdb, xdoc, root, service);
1821                 r = ingest_sub_record(cl, xdoc, sroot, record_no, nmem, sdb,
1822                                       mk);
1823             }
1824     }
1825     else if (!strcmp((const char *) root->name, "record"))
1826     {
1827         const char *mergekey_norm =
1828             get_mergekey(xdoc, root, cl, record_no, service, nmem,
1829                          se->mergekey);
1830         if (mergekey_norm)
1831         {
1832             struct record_metadata_attr *mk = (struct record_metadata_attr*)
1833                 nmem_malloc(nmem, sizeof(*mk));
1834             mk->name = 0;
1835             mk->value = nmem_strdup(nmem, mergekey_norm);
1836             mk->next = 0;
1837
1838             if (!cached_copy)
1839                 insert_settings_values(sdb, xdoc, root, service);
1840             r = ingest_sub_record(cl, xdoc, root, record_no, nmem, sdb, mk);
1841         }
1842     }
1843     else
1844     {
1845         session_log(se, YLOG_WARN, "Bad pz root element: %s",
1846                     (const char *) root->name);
1847         r = -1;
1848     }
1849     return r;
1850 }
1851
1852
1853 //    struct conf_metadata *ser_md = &service->metadata[md_field_id];
1854 //    struct record_metadata *rec_md = record->metadata[md_field_id];
1855 static int match_metadata_local(struct conf_service *service,
1856                                 struct conf_metadata *ser_md,
1857                                 struct record_metadata *rec_md0,
1858                                 char **values, int num_v)
1859 {
1860     int i;
1861     struct record_metadata *rec_md = rec_md0;
1862     WRBUF val_wr = 0;
1863     WRBUF text_wr = wrbuf_alloc();
1864     for (i = 0; i < num_v; )
1865     {
1866         if (rec_md)
1867         {
1868             if (ser_md->type == Metadata_type_year
1869                 || ser_md->type == Metadata_type_date)
1870             {
1871                 int y = atoi(values[i]);
1872                 if (y >= rec_md->data.number.min
1873                     && y <= rec_md->data.number.max)
1874                     break;
1875             }
1876             else
1877             {
1878                 if (!val_wr)
1879                 {
1880                     val_wr = wrbuf_alloc();
1881                     mergekey_norm_wr(service->charsets, val_wr, values[i]);
1882                 }
1883                 wrbuf_rewind(text_wr);
1884                 mergekey_norm_wr(service->charsets, text_wr,
1885                                  rec_md->data.text.disp);
1886                 if (!strcmp(wrbuf_cstr(val_wr), wrbuf_cstr(text_wr)))
1887                     break;
1888             }
1889             rec_md = rec_md->next;
1890         }
1891         else
1892         {
1893             rec_md = rec_md0;
1894             wrbuf_destroy(val_wr);
1895             val_wr = 0;
1896             i++;
1897         }
1898     }
1899     wrbuf_destroy(val_wr);
1900     wrbuf_destroy(text_wr);
1901     return i < num_v ? 1 : 0;
1902 }
1903
1904 int session_check_cluster_limit(struct session *se, struct record_cluster *rec)
1905 {
1906     int i;
1907     struct conf_service *service = se->service;
1908     int ret = 1;
1909     const char *name;
1910     const char *value;
1911     NMEM nmem_tmp = nmem_create();
1912
1913     for (i = 0; (name = facet_limits_get(se->facet_limits, i, &value)); i++)
1914     {
1915         int j;
1916         for (j = 0; j < service->num_metadata; j++)
1917         {
1918             struct conf_metadata *md = service->metadata + j;
1919             if (!strcmp(md->name, name) && md->limitcluster)
1920             {
1921                 char **values = 0;
1922                 int num = 0;
1923                 int md_field_id =
1924                     conf_service_metadata_field_id(service,
1925                                                    md->limitcluster);
1926
1927                 if (md_field_id < 0)
1928                 {
1929                     ret = 0;
1930                     break;
1931                 }
1932
1933                 nmem_strsplit_escape2(nmem_tmp, "|", value, &values,
1934                                       &num, 1, '\\', 1);
1935
1936                 if (!match_metadata_local(service,
1937                                           &service->metadata[md_field_id],
1938                                           rec->metadata[md_field_id],
1939                                           values, num))
1940                 {
1941                     ret = 0;
1942                     break;
1943                 }
1944             }
1945         }
1946     }
1947     nmem_destroy(nmem_tmp);
1948     return ret;
1949 }
1950
1951 // Skip record on non-zero
1952 static int check_limit_local(struct client *cl,
1953                              struct record *record,
1954                              int record_no)
1955 {
1956     int skip_record = 0;
1957     struct session *se = client_get_session(cl);
1958     struct conf_service *service = se->service;
1959     NMEM nmem_tmp = nmem_create();
1960     struct session_database *sdb = client_get_database(cl);
1961     int l = 0;
1962     while (!skip_record)
1963     {
1964         int md_field_id;
1965         char **values = 0;
1966         int num_v = 0;
1967         const char *name =
1968             client_get_facet_limit_local(cl, sdb, &l, nmem_tmp,
1969                                          &num_v, &values);
1970         if (!name)
1971             break;
1972
1973         if (!strcmp(name, "*"))
1974         {
1975             for (md_field_id = 0; md_field_id < service->num_metadata;
1976                  md_field_id++)
1977             {
1978                 if (match_metadata_local(
1979                         service,
1980                         &service->metadata[md_field_id],
1981                         record->metadata[md_field_id],
1982                         values, num_v))
1983                     break;
1984             }
1985             if (md_field_id == service->num_metadata)
1986                 skip_record = 1;
1987         }
1988         else
1989         {
1990             md_field_id = conf_service_metadata_field_id(service, name);
1991             if (md_field_id < 0)
1992             {
1993                 skip_record = 1;
1994                 break;
1995             }
1996             if (!match_metadata_local(
1997                     service,
1998                     &service->metadata[md_field_id],
1999                     record->metadata[md_field_id],
2000                     values, num_v))
2001             {
2002                 skip_record = 1;
2003             }
2004         }
2005     }
2006     nmem_destroy(nmem_tmp);
2007     return skip_record;
2008 }
2009
2010 static int ingest_to_cluster(struct client *cl,
2011                              xmlDoc *xdoc,
2012                              xmlNode *root,
2013                              int record_no,
2014                              struct record_metadata_attr *merge_keys)
2015 {
2016     xmlNode *n;
2017     xmlChar *type = 0;
2018     xmlChar *value = 0;
2019     struct session *se = client_get_session(cl);
2020     struct conf_service *service = se->service;
2021     int term_factor = 1;
2022     struct record_cluster *cluster;
2023     struct record_metadata **metadata0;
2024     struct session_database *sdb = client_get_database(cl);
2025     NMEM ingest_nmem = 0;
2026     char **rank_values = 0;
2027     int rank_num = 0;
2028     struct record *record = record_create(se->nmem,
2029                                           service->num_metadata,
2030                                           service->num_sortkeys, cl,
2031                                           record_no);
2032
2033     for (n = root->children; n; n = n->next)
2034     {
2035         if (type)
2036             xmlFree(type);
2037         if (value)
2038             xmlFree(value);
2039         type = value = 0;
2040
2041         if (n->type != XML_ELEMENT_NODE)
2042             continue;
2043         if (!strcmp((const char *) n->name, "metadata"))
2044         {
2045             struct conf_metadata *ser_md = 0;
2046             struct record_metadata **wheretoput = 0;
2047             struct record_metadata *rec_md = 0;
2048             int md_field_id = -1;
2049
2050             type = xmlGetProp(n, (xmlChar *) "type");
2051             value = xmlNodeListGetString(xdoc, n->children, 1);
2052             if (!type)
2053                 continue;
2054             if (!value || !*value)
2055             {
2056                 xmlChar *empty = xmlGetProp(n, (xmlChar *) "empty");
2057                 if (!empty)
2058                     continue;
2059                 if (value)
2060                     xmlFree(value);
2061                 value = empty;
2062             }
2063             md_field_id
2064                 = conf_service_metadata_field_id(service, (const char *) type);
2065             if (md_field_id < 0)
2066             {
2067                 if (se->number_of_warnings_unknown_metadata == 0)
2068                 {
2069                     session_log(se, YLOG_WARN,
2070                             "Ignoring unknown metadata element: %s", type);
2071                 }
2072                 se->number_of_warnings_unknown_metadata++;
2073                 continue;
2074             }
2075
2076             ser_md = &service->metadata[md_field_id];
2077
2078             // non-merged metadata
2079             rec_md = record_metadata_init(se->nmem, (const char *) value,
2080                                           ser_md->type, n->properties);
2081             if (!rec_md)
2082             {
2083                 session_log(se, YLOG_WARN, "bad metadata data '%s' "
2084                             "for element '%s'", value, type);
2085                 continue;
2086             }
2087
2088             if (ser_md->type == Metadata_type_generic)
2089             {
2090                 WRBUF w = wrbuf_alloc();
2091                 if (relevance_snippet(se->relevance,
2092                                       (char*) value, ser_md->name, w))
2093                     rec_md->data.text.snippet = nmem_strdup(se->nmem,
2094                                                             wrbuf_cstr(w));
2095                 wrbuf_destroy(w);
2096             }
2097
2098
2099             wheretoput = &record->metadata[md_field_id];
2100             while (*wheretoput)
2101                 wheretoput = &(*wheretoput)->next;
2102             *wheretoput = rec_md;
2103         }
2104     }
2105
2106     if (check_limit_local(cl, record, record_no))
2107     {
2108         if (type)
2109             xmlFree(type);
2110         if (value)
2111             xmlFree(value);
2112         return -2;
2113     }
2114     cluster = reclist_insert(se->reclist, se->relevance, service, record,
2115                              merge_keys, &se->total_merged);
2116     if (!cluster)
2117     {
2118         if (type)
2119             xmlFree(type);
2120         if (value)
2121             xmlFree(value);
2122         return 0; // complete match with existing record
2123     }
2124
2125     {
2126         const char *use_term_factor_str =
2127             session_setting_oneval(sdb, PZ_TERMLIST_TERM_FACTOR);
2128         if (use_term_factor_str && use_term_factor_str[0] == '1')
2129         {
2130             int maxrecs = client_get_maxrecs(cl);
2131             int hits = (int) client_get_hits(cl);
2132             term_factor = MAX(hits, maxrecs) /  MAX(1, maxrecs);
2133             assert(term_factor >= 1);
2134             session_log(se, YLOG_DEBUG, "Using term factor: %d (%d / %d)",
2135                         term_factor, MAX(hits, maxrecs), MAX(1, maxrecs));
2136         }
2137     }
2138
2139     if (global_parameters.dump_records)
2140         session_log(se, YLOG_LOG, "Cluster id %s from %s (#%d)", cluster->recid,
2141                     sdb->database->id, record_no);
2142
2143     // original metadata, to check if first existence of a field
2144     metadata0 = xmalloc(sizeof(*metadata0) * service->num_metadata);
2145     memcpy(metadata0, cluster->metadata,
2146            sizeof(*metadata0) * service->num_metadata);
2147
2148     ingest_nmem = nmem_create();
2149     if (se->rank)
2150     {
2151         yaz_log(YLOG_LOG, "local in sort : %s", se->rank);
2152         nmem_strsplit_escape2(ingest_nmem, ",", se->rank, &rank_values,
2153                               &rank_num, 1, '\\', 1);
2154     }
2155
2156     // now parsing XML record and adding data to cluster or record metadata
2157     for (n = root->children; n; n = n->next)
2158     {
2159         pp2_charset_token_t prt;
2160         if (type)
2161             xmlFree(type);
2162         if (value)
2163             xmlFree(value);
2164         type = value = 0;
2165
2166         if (n->type != XML_ELEMENT_NODE)
2167             continue;
2168         if (!strcmp((const char *) n->name, "metadata"))
2169         {
2170             struct conf_metadata *ser_md = 0;
2171             struct conf_sortkey *ser_sk = 0;
2172             struct record_metadata **wheretoput = 0;
2173             struct record_metadata *rec_md = 0;
2174             int md_field_id = -1;
2175             int sk_field_id = -1;
2176             const char *rank = 0;
2177             xmlChar *xml_rank = 0;
2178
2179             type = xmlGetProp(n, (xmlChar *) "type");
2180             value = xmlNodeListGetString(xdoc, n->children, 1);
2181
2182             if (!type || !value || !*value)
2183                 continue;
2184
2185             md_field_id
2186                 = conf_service_metadata_field_id(service, (const char *) type);
2187             if (md_field_id < 0)
2188                 continue;
2189
2190             ser_md = &service->metadata[md_field_id];
2191
2192             if (ser_md->sortkey_offset >= 0)
2193             {
2194                 sk_field_id = ser_md->sortkey_offset;
2195                 ser_sk = &service->sortkeys[sk_field_id];
2196             }
2197
2198             // merged metadata
2199             rec_md = record_metadata_init(se->nmem, (const char *) value,
2200                                           ser_md->type, 0);
2201
2202             // see if the field was not in cluster already (from beginning)
2203
2204             if (!rec_md)
2205                 continue;
2206
2207             if (rank_num)
2208             {
2209                 int i;
2210                 for (i = 0; i < rank_num; i++)
2211                 {
2212                     const char *val = rank_values[i];
2213                     const char *cp = strchr(val, '=');
2214                     if (!cp)
2215                         continue;
2216                     if ((cp - val) == strlen((const char *) type)
2217                         && !memcmp(val, type, cp - val))
2218                     {
2219                         rank = cp + 1;
2220                         break;
2221                     }
2222                 }
2223             }
2224             else
2225             {
2226                 xml_rank = xmlGetProp(n, (xmlChar *) "rank");
2227                 rank = xml_rank ? (const char *) xml_rank : ser_md->rank;
2228             }
2229
2230             wheretoput = &cluster->metadata[md_field_id];
2231
2232             if (ser_md->merge == Metadata_merge_first)
2233             {
2234                 if (!metadata0[md_field_id])
2235                 {
2236                     while (*wheretoput)
2237                         wheretoput = &(*wheretoput)->next;
2238                     *wheretoput = rec_md;
2239                 }
2240             }
2241             else if (ser_md->merge == Metadata_merge_unique)
2242             {
2243                 while (*wheretoput)
2244                 {
2245                     if (!strcmp((const char *) (*wheretoput)->data.text.disp,
2246                                 rec_md->data.text.disp))
2247                         break;
2248                     wheretoput = &(*wheretoput)->next;
2249                 }
2250                 if (!*wheretoput)
2251                     *wheretoput = rec_md;
2252             }
2253             else if (ser_md->merge == Metadata_merge_longest)
2254             {
2255                 if (!*wheretoput
2256                     || strlen(rec_md->data.text.disp)
2257                     > strlen((*wheretoput)->data.text.disp))
2258                 {
2259                     *wheretoput = rec_md;
2260                     if (ser_sk)
2261                     {
2262                         const char *sort_str = 0;
2263                         int skip_article =
2264                             ser_sk->type == Metadata_type_skiparticle;
2265
2266                         if (!cluster->sortkeys[sk_field_id])
2267                             cluster->sortkeys[sk_field_id] =
2268                                 nmem_malloc(se->nmem,
2269                                             sizeof(union data_types));
2270
2271                         prt =
2272                             pp2_charset_token_create(service->charsets, "sort");
2273
2274                         pp2_charset_token_first(prt, rec_md->data.text.disp,
2275                                                 skip_article);
2276
2277                         pp2_charset_token_next(prt);
2278
2279                         sort_str = pp2_get_sort(prt);
2280
2281                         cluster->sortkeys[sk_field_id]->text.disp =
2282                             rec_md->data.text.disp;
2283                         if (!sort_str)
2284                         {
2285                             sort_str = rec_md->data.text.disp;
2286                             session_log(se, YLOG_WARN,
2287                                     "Could not make sortkey. Bug #1858");
2288                         }
2289                         cluster->sortkeys[sk_field_id]->text.sort =
2290                             nmem_strdup(se->nmem, sort_str);
2291                         pp2_charset_token_destroy(prt);
2292                     }
2293                 }
2294             }
2295             else if (ser_md->merge == Metadata_merge_all)
2296             {
2297                 while (*wheretoput)
2298                     wheretoput = &(*wheretoput)->next;
2299                 *wheretoput = rec_md;
2300             }
2301             else if (ser_md->merge == Metadata_merge_range)
2302             {
2303                 if (!*wheretoput)
2304                 {
2305                     *wheretoput = rec_md;
2306                     if (ser_sk)
2307                         cluster->sortkeys[sk_field_id]
2308                             = &rec_md->data;
2309                 }
2310                 else
2311                 {
2312                     int this_min = rec_md->data.number.min;
2313                     int this_max = rec_md->data.number.max;
2314                     if (this_min < (*wheretoput)->data.number.min)
2315                         (*wheretoput)->data.number.min = this_min;
2316                     if (this_max > (*wheretoput)->data.number.max)
2317                         (*wheretoput)->data.number.max = this_max;
2318                 }
2319             }
2320
2321             // ranking of _all_ fields enabled ...
2322             if (rank)
2323             {
2324                 relevance_countwords(se->relevance, cluster,
2325                                      (char *) value, rank, ser_md->name);
2326             }
2327             // construct facets ... unless the client already has reported them
2328             if (ser_md->termlist && !client_has_facet(cl, (char *) type))
2329             {
2330                 if (ser_md->type == Metadata_type_year)
2331                 {
2332                     char year[64];
2333                     sprintf(year, "%d", rec_md->data.number.max);
2334
2335                     add_facet(se, (char *) type, year, term_factor);
2336                     if (rec_md->data.number.max != rec_md->data.number.min)
2337                     {
2338                         sprintf(year, "%d", rec_md->data.number.min);
2339                         add_facet(se, (char *) type, year, term_factor);
2340                     }
2341                 }
2342                 else
2343                     add_facet(se, (char *) type, (char *) value, term_factor);
2344             }
2345
2346             // cleaning up
2347             if (xml_rank)
2348                 xmlFree(xml_rank);
2349             xmlFree(type);
2350             xmlFree(value);
2351             type = value = 0;
2352         }
2353         else
2354         {
2355             if (se->number_of_warnings_unknown_elements == 0)
2356                 session_log(se, YLOG_WARN,
2357                         "Unexpected element in internal record: %s", n->name);
2358             se->number_of_warnings_unknown_elements++;
2359         }
2360     }
2361     if (type)
2362         xmlFree(type);
2363     if (value)
2364         xmlFree(value);
2365
2366     nmem_destroy(ingest_nmem);
2367     xfree(metadata0);
2368     relevance_donerecord(se->relevance, cluster);
2369     se->total_records++;
2370
2371     return 0;
2372 }
2373
2374 void session_log(struct session *s, int level, const char *fmt, ...)
2375 {
2376     char buf[1024];
2377     va_list ap;
2378     va_start(ap, fmt);
2379
2380     yaz_vsnprintf(buf, sizeof(buf)-30, fmt, ap);
2381     yaz_log(level, "Session %u: %s", s ? s->session_id : 0, buf);
2382
2383     va_end(ap);
2384 }
2385
2386 /*
2387  * Local variables:
2388  * c-basic-offset: 4
2389  * c-file-style: "Stroustrup"
2390  * indent-tabs-mode: nil
2391  * End:
2392  * vim: shiftwidth=4 tabstop=8 expandtab
2393  */
2394