src/session.c

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