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, *l0;
 750
 751     session_alert_watch(se, SESSION_WATCH_SHOW);
 752     session_alert_watch(se, SESSION_WATCH_BYTARGET);
 753     session_alert_watch(se, SESSION_WATCH_TERMLIST);
 754     session_alert_watch(se, SESSION_WATCH_SHOW_PREF);
 755
 756     session_log(se, YLOG_DEBUG, "Search");
 757
 758     *addinfo = 0;
 759
 760     if (se->settings_modified) {
 761         session_remove_cached_clients(se);
 762     }
 763     else
 764         session_reset_active_clients(se, 0);
 765
 766     session_enter(se, "session_search");
 767     se->settings_modified = 0;
 768
 769     if (mergekey)
 770     {
 771         xfree(se->mergekey);
 772         se->mergekey = *mergekey ? xstrdup(mergekey) : 0;
 773     }
 774     if (rank)
 775     {
 776         xfree(se->rank);
 777         se->rank = *rank ? xstrdup(rank) : 0;
 778     }
 779
 780     session_clear_set(se, sp);
 781     relevance_destroy(&se->relevance);
 782
 783     live_channels = select_targets(se, filter);
 784     if (!live_channels)
 785     {
 786         session_leave(se, "session_search");
 787         return PAZPAR2_NO_TARGETS;
 788     }
 789
 790     facet_limits_destroy(se->facet_limits);
 791     se->facet_limits = facet_limits_create(limit);
 792     if (!se->facet_limits)
 793     {
 794         *addinfo = "limit";
 795         session_leave(se, "session_search");
 796         return PAZPAR2_MALFORMED_PARAMETER_VALUE;
 797     }
 798
 799     l0 = se->clients_active;
 800     se->clients_active = 0;
 801     session_leave(se, "session_search");
 802
 803     for (l = l0; l; l = l->next)
 804     {
 805         int parse_ret;
 806         struct client *cl = l->client;
 807         client_parse_init(cl, 1);
 808         if (prepare_map(se, client_get_database(cl)) < 0)
 809             continue;
 810
 811         parse_ret = client_parse_query(cl, query, se->facet_limits, addinfo2);
 812         if (parse_ret == -1)
 813             no_failed_query++;
 814         else if (parse_ret == -2)
 815             no_failed_limit++;
 816         else if (parse_ret < 0)
 817             no_working++; /* other error, such as bad CCL map */
 818         else
 819         {
 820             client_parse_range(cl, startrecs, maxrecs);
 821             client_parse_sort(cl, sp);
 822             client_start_search(cl);
 823             no_working++;
 824         }
 825     }
 826     session_reset_active_clients(se, l0);
 827
 828     if (no_working == 0)
 829     {
 830         if (no_failed_query > 0)
 831         {
 832             *addinfo = "query";
 833             return PAZPAR2_MALFORMED_PARAMETER_VALUE;
 834         }
 835         else if (no_failed_limit > 0)
 836         {
 837             *addinfo = "limit";
 838             return PAZPAR2_MALFORMED_PARAMETER_VALUE;
 839         }
 840         else
 841             return PAZPAR2_NO_TARGETS;
 842     }
 843     return PAZPAR2_NO_ERROR;
 844 }
 845
 846 // Creates a new session_database object for a database
 847 static void session_init_databases_fun(void *context, struct database *db)
 848 {
 849     struct session *se = (struct session *) context;
 850     struct session_database *new = nmem_malloc(se->session_nmem, sizeof(*new));
 851     int i;
 852
 853     new->database = db;
 854
 855     new->map = 0;
 856     assert(db->settings);
 857     new->settings = nmem_malloc(se->session_nmem,
 858                                 sizeof(struct settings *) * db->num_settings);
 859     new->num_settings = db->num_settings;
 860     for (i = 0; i < db->num_settings; i++)
 861     {
 862         struct setting *setting = db->settings[i];
 863         new->settings[i] = setting;
 864     }
 865     new->next = se->databases;
 866     se->databases = new;
 867 }
 868
 869 // Doesn't free memory associated with sdb -- nmem takes care of that
 870 static void session_database_destroy(struct session_database *sdb)
 871 {
 872     sdb->map = 0;
 873 }
 874
 875 // Initialize session_database list -- this represents this session's view
 876 // of the database list -- subject to modification by the settings ws command
 877 void session_init_databases(struct session *se)
 878 {
 879     se->databases = 0;
 880     predef_grep_databases(se, se->service, session_init_databases_fun);
 881 }
 882
 883 // Probably session_init_databases_fun should be refactored instead of
 884 // called here.
 885 static struct session_database *load_session_database(struct session *se,
 886                                                       char *id)
 887 {
 888     struct database *db = new_database_inherit_settings(id, se->session_nmem, se->service->settings);
 889     session_init_databases_fun((void*) se, db);
 890
 891     // New sdb is head of se->databases list
 892     return se->databases;
 893 }
 894
 895 // Find an existing session database. If not found, load it
 896 static struct session_database *find_session_database(struct session *se,
 897                                                       char *id)
 898 {
 899     struct session_database *sdb;
 900
 901     for (sdb = se->databases; sdb; sdb = sdb->next)
 902         if (!strcmp(sdb->database->id, id))
 903             return sdb;
 904     return load_session_database(se, id);
 905 }
 906
 907 // Apply a session override to a database
 908 void session_apply_setting(struct session *se, char *dbname, char *name,
 909                            char *value)
 910 {
 911     struct session_database *sdb = find_session_database(se, dbname);
 912     struct conf_service *service = se->service;
 913     struct setting *s;
 914     int offset = settings_create_offset(service, name);
 915
 916     expand_settings_array(&sdb->settings, &sdb->num_settings, offset,
 917                           se->session_nmem);
 918
 919     // Force later recompute of settings-driven data structures
 920     // (happens when a search starts and client connections are prepared)
 921     if (offset == PZ_XSLT)
 922         sdb->map = 0;
 923
 924     se->settings_modified = 1;
 925     for (s = sdb->settings[offset]; s; s = s->next)
 926         if (!strcmp(s->name, name) &&
 927             dbname && s->target && !strcmp(dbname, s->target))
 928         {
 929             s->value = value;
 930             return;
 931         }
 932     s = nmem_malloc(se->session_nmem, sizeof(*s));
 933     s->precedence = 0;
 934     s->target = dbname;
 935     s->name = name;
 936     s->value = value;
 937     s->next = sdb->settings[offset];
 938     sdb->settings[offset] = s;
 939
 940 }
 941
 942 void session_destroy(struct session *se)
 943 {
 944     struct session_database *sdb;
 945     session_log(se, YLOG_LOG, "destroy");
 946     session_use(-1);
 947     session_remove_cached_clients(se);
 948
 949     for (sdb = se->databases; sdb; sdb = sdb->next)
 950         session_database_destroy(sdb);
 951     normalize_cache_destroy(se->normalize_cache);
 952     relevance_destroy(&se->relevance);
 953     reclist_destroy(se->reclist);
 954     xfree(se->mergekey);
 955     xfree(se->rank);
 956     if (nmem_total(se->nmem))
 957         session_log(se, YLOG_DEBUG, "NMEN operation usage %zd", nmem_total(se->nmem));
 958     if (nmem_total(se->session_nmem))
 959         session_log(se, YLOG_DEBUG, "NMEN session usage %zd", nmem_total(se->session_nmem));
 960     facet_limits_destroy(se->facet_limits);
 961     nmem_destroy(se->nmem);
 962     service_destroy(se->service);
 963     yaz_mutex_destroy(&se->session_mutex);
 964 }
 965
 966 size_t session_get_memory_status(struct session *session) {
 967     size_t session_nmem;
 968     if (session == 0)
 969         return 0;
 970     session_enter(session, "session_get_memory_status");
 971     session_nmem = nmem_total(session->nmem);
 972     session_leave(session, "session_get_memory_status");
 973     return session_nmem;
 974 }
 975
 976
 977 struct session *new_session(NMEM nmem, struct conf_service *service,
 978                             unsigned session_id)
 979 {
 980     int i;
 981     struct session *session = nmem_malloc(nmem, sizeof(*session));
 982
 983     char tmp_str[50];
 984
 985     sprintf(tmp_str, "session#%u", session_id);
 986
 987     session->session_id = session_id;
 988     session_log(session, YLOG_DEBUG, "New");
 989     session->service = service;
 990     session->relevance = 0;
 991     session->total_records = 0;
 992     session->number_of_warnings_unknown_elements = 0;
 993     session->number_of_warnings_unknown_metadata = 0;
 994     session->num_termlists = 0;
 995     session->reclist = reclist_create(nmem);
 996     session->clients_active = 0;
 997     session->clients_cached = 0;
 998     session->settings_modified = 0;
 999     session->session_nmem = nmem;
1000     session->nmem = nmem_create();
1001     session->databases = 0;
1002     session->sorted_results = 0;
1003     session->facet_limits = 0;
1004     session->mergekey = 0;
1005     session->rank = 0;
1006
1007     for (i = 0; i <= SESSION_WATCH_MAX; i++)
1008     {
1009         session->watchlist[i].data = 0;
1010         session->watchlist[i].fun = 0;
1011     }
1012     session->normalize_cache = normalize_cache_create();
1013     session->session_mutex = 0;
1014     pazpar2_mutex_create(&session->session_mutex, tmp_str);
1015     session_log(session, YLOG_LOG, "create");
1016
1017     session_use(1);
1018     return session;
1019 }
1020
1021 const char * client_get_suggestions_xml(struct client *cl, WRBUF wrbuf);
1022
1023 static struct hitsbytarget *hitsbytarget_nb(struct session *se,
1024                                             int *count, NMEM nmem)
1025 {
1026     struct hitsbytarget *res = 0;
1027     struct client_list *l;
1028     size_t sz = 0;
1029
1030     for (l = se->clients_active; l; l = l->next)
1031         sz++;
1032
1033     res = nmem_malloc(nmem, sizeof(*res) * sz);
1034     *count = 0;
1035     for (l = se->clients_active; l; l = l->next)
1036     {
1037         struct client *cl = l->client;
1038         WRBUF w = wrbuf_alloc();
1039         const char *name = session_setting_oneval(client_get_database(cl),
1040                                                   PZ_NAME);
1041
1042         res[*count].id = client_get_id(cl);
1043         res[*count].name = *name ? name : "Unknown";
1044         res[*count].hits = client_get_hits(cl);
1045         res[*count].approximation = client_get_approximation(cl);
1046         res[*count].records = client_get_num_records(cl);
1047         res[*count].filtered = client_get_num_records_filtered(cl);
1048         res[*count].diagnostic =
1049             client_get_diagnostic(cl, &res[*count].message,
1050                                   &res[*count].addinfo);
1051         res[*count].state = client_get_state_str(cl);
1052         res[*count].connected  = client_get_connection(cl) ? 1 : 0;
1053         session_settings_dump(se, client_get_database(cl), w);
1054         res[*count].settings_xml = nmem_strdup(nmem, wrbuf_cstr(w));
1055         wrbuf_rewind(w);
1056         wrbuf_puts(w, "");
1057         res[*count].suggestions_xml = nmem_strdup(nmem, client_get_suggestions_xml(cl, w));
1058         wrbuf_destroy(w);
1059         (*count)++;
1060     }
1061     return res;
1062 }
1063
1064 struct hitsbytarget *get_hitsbytarget(struct session *se, int *count, NMEM nmem)
1065 {
1066     struct hitsbytarget *p;
1067     session_enter(se, "get_hitsbytarget");
1068     p = hitsbytarget_nb(se, count, nmem);
1069     session_leave(se, "get_hitsbytarget");
1070     return p;
1071 }
1072
1073 // Compares two hitsbytarget nodes by hitcount
1074 static int cmp_ht(const void *p1, const void *p2)
1075 {
1076     const struct hitsbytarget *h1 = p1;
1077     const struct hitsbytarget *h2 = p2;
1078     return h2->hits - h1->hits;
1079 }
1080
1081 // Compares two hitsbytarget nodes by hitcount
1082 static int cmp_ht_approx(const void *p1, const void *p2)
1083 {
1084     const struct hitsbytarget *h1 = p1;
1085     const struct hitsbytarget *h2 = p2;
1086     return h2->approximation - h1->approximation;
1087 }
1088
1089 static int targets_termlist_nb(WRBUF wrbuf, struct session *se, int num,
1090                                NMEM nmem, int version)
1091 {
1092     struct hitsbytarget *ht;
1093     int count, i;
1094
1095     ht = hitsbytarget_nb(se, &count, nmem);
1096     if (version >= 2)
1097         qsort(ht, count, sizeof(struct hitsbytarget), cmp_ht_approx);
1098     else
1099         qsort(ht, count, sizeof(struct hitsbytarget), cmp_ht);
1100     for (i = 0; i < count && i < num && ht[i].hits > 0; i++)
1101     {
1102
1103         // do only print terms which have display names
1104
1105         wrbuf_puts(wrbuf, "<term>\n");
1106
1107         wrbuf_puts(wrbuf, "<id>");
1108         wrbuf_xmlputs(wrbuf, ht[i].id);
1109         wrbuf_puts(wrbuf, "</id>\n");
1110
1111         wrbuf_puts(wrbuf, "<name>");
1112         if (!ht[i].name || !ht[i].name[0])
1113             wrbuf_xmlputs(wrbuf, "NO TARGET NAME");
1114         else
1115             wrbuf_xmlputs(wrbuf, ht[i].name);
1116         wrbuf_puts(wrbuf, "</name>\n");
1117
1118         wrbuf_printf(wrbuf, "<frequency>" ODR_INT_PRINTF "</frequency>\n",
1119                      ht[i].hits);
1120
1121         if (version >= 2) {
1122             // Should not print if we know it isn't a approximation.
1123             wrbuf_printf(wrbuf, "<approximation>" ODR_INT_PRINTF "</approximation>\n", ht[i].approximation);
1124             wrbuf_printf(wrbuf, "<records>%d</records>\n", ht[i].records - ht[i].filtered);
1125             wrbuf_printf(wrbuf, "<filtered>%d</filtered>\n", ht[i].filtered);
1126         }
1127
1128         wrbuf_puts(wrbuf, "<state>");
1129         wrbuf_xmlputs(wrbuf, ht[i].state);
1130         wrbuf_puts(wrbuf, "</state>\n");
1131
1132         wrbuf_printf(wrbuf, "<diagnostic>%d</diagnostic>\n",
1133                      ht[i].diagnostic);
1134         wrbuf_puts(wrbuf, "</term>\n");
1135     }
1136     return count;
1137 }
1138
1139 void perform_termlist(struct http_channel *c, struct session *se,
1140                       const char *name, int num, int version)
1141 {
1142     int i, j;
1143     NMEM nmem_tmp = nmem_create();
1144     char **names;
1145     int num_names = 0;
1146
1147     if (!name)
1148         name = "*";
1149
1150     nmem_strsplit(nmem_tmp, ",", name, &names, &num_names);
1151
1152     session_enter(se, "perform_termlist");
1153
1154     for (j = 0; j < num_names; j++)
1155     {
1156         const char *tname;
1157         int must_generate_empty = 1; /* bug 5350 */
1158
1159         for (i = 0; i < se->num_termlists; i++)
1160         {
1161             tname = se->termlists[i].name;
1162             if (!strcmp(names[j], tname) || !strcmp(names[j], "*"))
1163             {
1164                 struct termlist_score **p = 0;
1165                 int len;
1166
1167                 wrbuf_puts(c->wrbuf, "<list name=\"");
1168                 wrbuf_xmlputs(c->wrbuf, tname);
1169                 wrbuf_puts(c->wrbuf, "\">\n");
1170                 must_generate_empty = 0;
1171
1172                 p = termlist_highscore(se->termlists[i].termlist, &len,
1173                                        nmem_tmp);
1174                 if (p)
1175                 {
1176                     int i;
1177                     for (i = 0; i < len && i < num; i++)
1178                     {
1179                         // prevent sending empty term elements
1180                         if (!p[i]->display_term || !p[i]->display_term[0])
1181                             continue;
1182
1183                         wrbuf_puts(c->wrbuf, "<term>");
1184                         wrbuf_puts(c->wrbuf, "<name>");
1185                         wrbuf_xmlputs(c->wrbuf, p[i]->display_term);
1186                         wrbuf_puts(c->wrbuf, "</name>");
1187
1188                         wrbuf_printf(c->wrbuf,
1189                                      "<frequency>%d</frequency>",
1190                                      p[i]->frequency);
1191                         wrbuf_puts(c->wrbuf, "</term>\n");
1192                     }
1193                 }
1194                 wrbuf_puts(c->wrbuf, "</list>\n");
1195             }
1196         }
1197         tname = "xtargets";
1198         if (!strcmp(names[j], tname) || !strcmp(names[j], "*"))
1199         {
1200             wrbuf_puts(c->wrbuf, "<list name=\"");
1201             wrbuf_xmlputs(c->wrbuf, tname);
1202             wrbuf_puts(c->wrbuf, "\">\n");
1203
1204             targets_termlist_nb(c->wrbuf, se, num, c->nmem, version);
1205             wrbuf_puts(c->wrbuf, "</list>\n");
1206             must_generate_empty = 0;
1207         }
1208         if (must_generate_empty)
1209         {
1210             wrbuf_puts(c->wrbuf, "<list name=\"");
1211             wrbuf_xmlputs(c->wrbuf, names[j]);
1212             wrbuf_puts(c->wrbuf, "\"/>\n");
1213         }
1214     }
1215     session_leave(se, "perform_termlist");
1216     nmem_destroy(nmem_tmp);
1217 }
1218
1219 #ifdef MISSING_HEADERS
1220 void report_nmem_stats(void)
1221 {
1222     size_t in_use, is_free;
1223
1224     nmem_get_memory_in_use(&in_use);
1225     nmem_get_memory_free(&is_free);
1226
1227     yaz_log(YLOG_LOG, "nmem stat: use=%ld free=%ld",
1228             (long) in_use, (long) is_free);
1229 }
1230 #endif
1231
1232 struct record_cluster *show_single_start(struct session *se, const char *id,
1233                                          struct record_cluster **prev_r,
1234                                          struct record_cluster **next_r)
1235 {
1236     struct record_cluster *r = 0;
1237
1238     session_enter(se, "show_single_start");
1239     *prev_r = 0;
1240     *next_r = 0;
1241     reclist_limit(se->reclist, se, 1);
1242
1243     reclist_enter(se->reclist);
1244     while ((r = reclist_read_record(se->reclist)))
1245     {
1246         if (!strcmp(r->recid, id))
1247         {
1248             *next_r = reclist_read_record(se->reclist);
1249             break;
1250         }
1251         *prev_r = r;
1252     }
1253     reclist_leave(se->reclist);
1254     if (!r)
1255         session_leave(se, "show_single_start");
1256     return r;
1257 }
1258
1259 void show_single_stop(struct session *se, struct record_cluster *rec)
1260 {
1261     session_leave(se, "show_single_stop");
1262 }
1263
1264
1265 int session_fetch_more(struct session *se)
1266 {
1267     struct client_list *l;
1268     int ret = 0;
1269
1270     for (l = se->clients_active; l; l = l->next)
1271     {
1272         struct client *cl = l->client;
1273         if (client_get_state(cl) == Client_Idle)
1274         {
1275             if (client_fetch_more(cl))
1276             {
1277                 session_log(se, YLOG_LOG, "%s: more to fetch",
1278                             client_get_id(cl));
1279                 ret = 1;
1280             }
1281             else
1282             {
1283                 session_log(se, YLOG_LOG, "%s: no more to fetch",
1284                             client_get_id(cl));
1285                 session_log(se, YLOG_LOG, "%s: hits=" ODR_INT_PRINTF
1286                             " records=%d filtered=%d",
1287                             client_get_id(cl),
1288                             client_get_hits(cl),
1289                             client_get_num_records(cl),
1290                             client_get_num_records_filtered(cl));
1291             }
1292         }
1293         else
1294         {
1295             session_log(se, YLOG_LOG, "%s: no fetch due to state=%s",
1296                         client_get_id(cl), client_get_state_str(cl));
1297         }
1298
1299     }
1300     return ret;
1301 }
1302
1303 struct record_cluster **show_range_start(struct session *se,
1304                                          struct reclist_sortparms *sp,
1305                                          int start, int *num, int *total,
1306                                          Odr_int *sumhits, Odr_int *approx_hits,
1307                                          void (*show_records_ready)(void *data),
1308                                          struct http_channel *chan)
1309 {
1310     struct record_cluster **recs = 0;
1311     struct reclist_sortparms *spp;
1312     struct client_list *l;
1313     int i;
1314 #if USE_TIMING
1315     yaz_timing_t t = yaz_timing_create();
1316 #endif
1317     session_enter(se, "show_range_start");
1318     *sumhits = 0;
1319     *approx_hits = 0;
1320     *total = 0;
1321     reclist_limit(se->reclist, se, 0);
1322     if (se->relevance)
1323     {
1324         for (spp = sp; spp; spp = spp->next)
1325             if (spp->type == Metadata_type_relevance)
1326             {
1327                 relevance_prepare_read(se->relevance, se->reclist);
1328                 break;
1329             }
1330         for (l = se->clients_active; l; l = l->next) {
1331             *sumhits += client_get_hits(l->client);
1332             *approx_hits += client_get_approximation(l->client);
1333         }
1334     }
1335     reclist_sort(se->reclist, sp);
1336
1337     reclist_enter(se->reclist);
1338     *total = reclist_get_num_records(se->reclist);
1339
1340     for (l = se->clients_active; l; l = l->next)
1341         client_update_show_stat(l->client, 0);
1342
1343     for (i = 0; i < start; i++)
1344     {
1345         struct record_cluster *r = reclist_read_record(se->reclist);
1346         if (!r)
1347         {
1348             *num = 0;
1349             break;
1350         }
1351         else
1352         {
1353             struct record *rec = r->records;
1354             for (;rec; rec = rec->next)
1355                 client_update_show_stat(rec->client, 1);
1356         }
1357     }
1358     recs = nmem_malloc(se->nmem, (*num > 0 ? *num : 1) * sizeof(*recs));
1359     for (i = 0; i < *num; i++)
1360     {
1361         struct record_cluster *r = reclist_read_record(se->reclist);
1362         if (!r)
1363         {
1364             *num = i;
1365             break;
1366         }
1367         else
1368         {
1369             struct record *rec = r->records;
1370             for (;rec; rec = rec->next)
1371                 client_update_show_stat(rec->client, 1);
1372             recs[i] = r;
1373         }
1374     }
1375     reclist_leave(se->reclist);
1376 #if USE_TIMING
1377     yaz_timing_stop(t);
1378     session_log(se, YLOG_LOG, "show %6.5f %3.2f %3.2f",
1379             yaz_timing_get_real(t), yaz_timing_get_user(t),
1380             yaz_timing_get_sys(t));
1381     yaz_timing_destroy(&t);
1382 #endif
1383
1384     if (!session_fetch_more(se))
1385         session_log(se, YLOG_LOG, "can not fetch more");
1386     else
1387     {
1388         show_range_stop(se, recs);
1389         session_log(se, YLOG_LOG, "fetching more in progress");
1390         if (session_set_watch(se, SESSION_WATCH_SHOW,
1391                               show_records_ready, chan, chan))
1392         {
1393             session_log(se, YLOG_WARN, "Ignoring show block");
1394             session_enter(se, "show_range_start");
1395         }
1396         else
1397         {
1398             session_log(se, YLOG_LOG, "session watch OK");
1399             return 0;
1400         }
1401     }
1402     return recs;
1403 }
1404
1405 void show_range_stop(struct session *se, struct record_cluster **recs)
1406 {
1407     session_leave(se, "show_range_stop");
1408 }
1409
1410 void statistics(struct session *se, struct statistics *stat)
1411 {
1412     struct client_list *l;
1413     int count = 0;
1414
1415     memset(stat, 0, sizeof(*stat));
1416     stat->num_hits = 0;
1417     for (l = se->clients_active; l; l = l->next)
1418     {
1419         struct client *cl = l->client;
1420         if (!client_get_connection(cl))
1421             stat->num_no_connection++;
1422         stat->num_hits += client_get_hits(cl);
1423         switch (client_get_state(cl))
1424         {
1425         case Client_Connecting: stat->num_connecting++; break;
1426         case Client_Working: stat->num_working++; break;
1427         case Client_Idle: stat->num_idle++; break;
1428         case Client_Failed: stat->num_failed++; break;
1429         case Client_Error: stat->num_error++; break;
1430         default: break;
1431         }
1432         count++;
1433     }
1434     stat->num_records = se->total_records;
1435
1436     stat->num_clients = count;
1437 }
1438
1439 static struct record_metadata *record_metadata_init(
1440     NMEM nmem, const char *value, enum conf_metadata_type type,
1441     struct _xmlAttr *attr)
1442 {
1443     struct record_metadata *rec_md = record_metadata_create(nmem);
1444     struct record_metadata_attr **attrp = &rec_md->attributes;
1445
1446     for (; attr; attr = attr->next)
1447     {
1448         if (attr->children && attr->children->content)
1449         {
1450             if (strcmp((const char *) attr->name, "type")
1451                 && strcmp((const char *) attr->name, "empty"))
1452             {  /* skip the "type" + "empty" attribute..
1453                   The "Type" is already part of the element in output
1454                   (md-%s) and so repeating it here is redundant */
1455                 *attrp = nmem_malloc(nmem, sizeof(**attrp));
1456                 (*attrp)->name =
1457                     nmem_strdup(nmem, (const char *) attr->name);
1458                 (*attrp)->value =
1459                     nmem_strdup(nmem, (const char *) attr->children->content);
1460                 attrp = &(*attrp)->next;
1461             }
1462         }
1463     }
1464     *attrp = 0;
1465
1466     switch (type)
1467     {
1468     case Metadata_type_generic:
1469     case Metadata_type_skiparticle:
1470         if (strstr(value, "://")) /* looks like a URL */
1471             rec_md->data.text.disp = nmem_strdup(nmem, value);
1472         else
1473             rec_md->data.text.disp =
1474                 normalize7bit_generic(nmem_strdup(nmem, value), " ,/.:([");
1475         rec_md->data.text.sort = 0;
1476         rec_md->data.text.snippet = 0;
1477         break;
1478     case Metadata_type_year:
1479     case Metadata_type_date:
1480     {
1481         int first, last;
1482         int longdate = 0;
1483
1484         if (type == Metadata_type_date)
1485             longdate = 1;
1486         if (extract7bit_dates((char *) value, &first, &last, longdate) < 0)
1487             return 0;
1488
1489         rec_md->data.number.min = first;
1490         rec_md->data.number.max = last;
1491     }
1492     break;
1493     case Metadata_type_float:
1494         rec_md->data.fnumber = atof(value);
1495         break;
1496     case Metadata_type_relevance:
1497     case Metadata_type_position:
1498         return 0;
1499     }
1500     return rec_md;
1501 }
1502
1503 static void mergekey_norm_wr(pp2_charset_fact_t charsets,
1504                              WRBUF norm_wr, const char *value)
1505 {
1506     const char *norm_str;
1507     pp2_charset_token_t prt =
1508         pp2_charset_token_create(charsets, "mergekey");
1509
1510     pp2_charset_token_first(prt, value, 0);
1511     while ((norm_str = pp2_charset_token_next(prt)))
1512     {
1513         if (*norm_str)
1514         {
1515             if (wrbuf_len(norm_wr))
1516                 wrbuf_puts(norm_wr, " ");
1517             wrbuf_puts(norm_wr, norm_str);
1518         }
1519     }
1520     pp2_charset_token_destroy(prt);
1521 }
1522
1523 static int get_mergekey_from_doc(xmlDoc *doc, xmlNode *root, const char *name,
1524                                  struct conf_service *service, WRBUF norm_wr)
1525 {
1526     xmlNode *n;
1527     int no_found = 0;
1528     for (n = root->children; n; n = n->next)
1529     {
1530         if (n->type != XML_ELEMENT_NODE)
1531             continue;
1532         if (!strcmp((const char *) n->name, "metadata"))
1533         {
1534             xmlChar *type = xmlGetProp(n, (xmlChar *) "type");
1535             if (type == NULL) {
1536                 yaz_log(YLOG_FATAL, "Missing type attribute on metadata element. Skipping!");
1537             }
1538             else if (!strcmp(name, (const char *) type))
1539             {
1540                 xmlChar *value = xmlNodeListGetString(doc, n->children, 1);
1541                 if (value && *value)
1542                 {
1543                     if (wrbuf_len(norm_wr) > 0)
1544                         wrbuf_puts(norm_wr, " ");
1545                     wrbuf_puts(norm_wr, name);
1546                     mergekey_norm_wr(service->charsets, norm_wr,
1547                                      (const char *) value);
1548                     no_found++;
1549                 }
1550                 if (value)
1551                     xmlFree(value);
1552             }
1553             xmlFree(type);
1554         }
1555     }
1556     return no_found;
1557 }
1558
1559 static const char *get_mergekey(xmlDoc *doc, xmlNode *root,
1560                                 struct client *cl, int record_no,
1561                                 struct conf_service *service, NMEM nmem,
1562                                 const char *session_mergekey)
1563 {
1564     char *mergekey_norm = 0;
1565     WRBUF norm_wr = wrbuf_alloc();
1566     xmlChar *mergekey;
1567
1568     if (session_mergekey)
1569     {
1570         int i, num = 0;
1571         char **values = 0;
1572         nmem_strsplit_escape2(nmem, ",", session_mergekey, &values,
1573                               &num, 1, '\\', 1);
1574
1575         for (i = 0; i < num; i++)
1576             get_mergekey_from_doc(doc, root, values[i], service, norm_wr);
1577     }
1578     else if ((mergekey = xmlGetProp(root, (xmlChar *) "mergekey")))
1579     {
1580         mergekey_norm_wr(service->charsets, norm_wr, (const char *) mergekey);
1581         xmlFree(mergekey);
1582     }
1583     else
1584     {
1585         /* no mergekey defined in XSL. Look for mergekey metadata instead */
1586         int field_id;
1587         for (field_id = 0; field_id < service->num_metadata; field_id++)
1588         {
1589             struct conf_metadata *ser_md = &service->metadata[field_id];
1590             if (ser_md->mergekey != Metadata_mergekey_no)
1591             {
1592                 int r = get_mergekey_from_doc(doc, root, ser_md->name,
1593                                               service, norm_wr);
1594                 if (r == 0 && ser_md->mergekey == Metadata_mergekey_required)
1595                 {
1596                     /* no mergekey on this one and it is required..
1597                        Generate unique key instead */
1598                     wrbuf_rewind(norm_wr);
1599                     break;
1600                 }
1601             }
1602         }
1603     }
1604
1605     /* generate unique key if none is not generated already or is empty */
1606     if (wrbuf_len(norm_wr) == 0)
1607     {
1608         wrbuf_printf(norm_wr, "position: %s-%d",
1609                      client_get_id(cl), record_no);
1610     }
1611     else
1612     {
1613         const char *lead = "content: ";
1614         wrbuf_insert(norm_wr, 0, lead, strlen(lead));
1615     }
1616     if (wrbuf_len(norm_wr) > 0)
1617         mergekey_norm = nmem_strdup(nmem, wrbuf_cstr(norm_wr));
1618     wrbuf_destroy(norm_wr);
1619     return mergekey_norm;
1620 }
1621
1622 /** \brief see if metadata for pz:recordfilter exists
1623     \param root xml root element of normalized record
1624     \param sdb session database for client
1625     \retval 0 if there is no metadata for pz:recordfilter
1626     \retval 1 if there is metadata for pz:recordfilter
1627
1628     If there is no pz:recordfilter defined, this function returns 1
1629     as well.
1630 */
1631
1632 static int check_record_filter(xmlNode *root, struct session_database *sdb)
1633 {
1634     int match = 0;
1635     xmlNode *n;
1636     const char *s;
1637     s = session_setting_oneval(sdb, PZ_RECORDFILTER);
1638
1639     if (!s || !*s)
1640         return 1;
1641
1642     for (n = root->children; n; n = n->next)
1643     {
1644         if (n->type != XML_ELEMENT_NODE)
1645             continue;
1646         if (!strcmp((const char *) n->name, "metadata"))
1647         {
1648             xmlChar *type = xmlGetProp(n, (xmlChar *) "type");
1649             if (type)
1650             {
1651                 size_t len;
1652                 int substring;
1653                 const char *eq;
1654
1655                 if ((eq = strchr(s, '=')))
1656                     substring = 0;
1657                 else if ((eq = strchr(s, '~')))
1658                     substring = 1;
1659                 if (eq)
1660                     len = eq - s;
1661                 else
1662                     len = strlen(s);
1663                 if (len == strlen((const char *)type) &&
1664                     !memcmp((const char *) type, s, len))
1665                 {
1666                     xmlChar *value = xmlNodeGetContent(n);
1667                     if (value && *value)
1668                     {
1669                         if (!eq ||
1670                             (substring && strstr((const char *) value, eq+1)) ||
1671                             (!substring && !strcmp((const char *) value, eq + 1)))
1672                             match = 1;
1673                     }
1674                     xmlFree(value);
1675                 }
1676                 xmlFree(type);
1677             }
1678         }
1679     }
1680     return match;
1681 }
1682
1683 static int ingest_to_cluster(struct client *cl,
1684                              xmlDoc *xdoc,
1685                              xmlNode *root,
1686                              int record_no,
1687                              struct record_metadata_attr *mergekey);
1688
1689 static int ingest_sub_record(struct client *cl, xmlDoc *xdoc, xmlNode *root,
1690                              int record_no, NMEM nmem,
1691                              struct session_database *sdb,
1692                              struct record_metadata_attr *mergekeys)
1693 {
1694     int ret = 0;
1695     struct session *se = client_get_session(cl);
1696
1697     if (!check_record_filter(root, sdb))
1698     {
1699         session_log(se, YLOG_LOG,
1700                     "Filtered out record no %d from %s",
1701                     record_no, sdb->database->id);
1702         return 0;
1703     }
1704     session_enter(se, "ingest_sub_record");
1705     if (client_get_session(cl) == se && se->relevance)
1706         ret = ingest_to_cluster(cl, xdoc, root, record_no, mergekeys);
1707     session_leave(se, "ingest_sub_record");
1708
1709     return ret;
1710 }
1711
1712 /** \brief ingest XML record
1713     \param cl client holds the result set for record
1714     \param rec record buffer (0 terminated)
1715     \param record_no record position (1, 2, ..)
1716     \param nmem working NMEM
1717     \retval 0 OK
1718     \retval -1 failure
1719     \retval -2 Filtered
1720 */
1721 int ingest_record(struct client *cl, const char *rec,
1722                   int record_no, NMEM nmem)
1723 {
1724     struct session *se = client_get_session(cl);
1725     struct session_database *sdb = client_get_database(cl);
1726     struct conf_service *service = se->service;
1727     xmlDoc *xdoc = normalize_record(se, sdb, service, rec, nmem);
1728     int r = ingest_xml_record(cl, xdoc, record_no, nmem, 0);
1729     client_store_xdoc(cl, record_no, xdoc);
1730     return r;
1731 }
1732
1733 int ingest_xml_record(struct client *cl, xmlDoc *xdoc,
1734                       int record_no, NMEM nmem, int cached_copy)
1735 {
1736     struct session *se = client_get_session(cl);
1737     struct session_database *sdb = client_get_database(cl);
1738     struct conf_service *service = se->service;
1739     xmlNode *root;
1740     int r = 0;
1741     if (!xdoc)
1742         return -1;
1743
1744     if (global_parameters.dump_records)
1745     {
1746         session_log(se, YLOG_LOG, "Normalized record from %s",
1747                     sdb->database->id);
1748         log_xml_doc(xdoc);
1749     }
1750
1751     root = xmlDocGetRootElement(xdoc);
1752
1753     if (!strcmp((const char *) root->name, "cluster"))
1754     {
1755         int no_merge_keys = 0;
1756         int no_merge_dups = 0;
1757         xmlNode *sroot;
1758         struct record_metadata_attr *mk = 0;
1759
1760         for (sroot = root->children; sroot; sroot = sroot->next)
1761             if (sroot->type == XML_ELEMENT_NODE &&
1762                 !strcmp((const char *) sroot->name, "record"))
1763             {
1764                 struct record_metadata_attr **mkp;
1765                 const char *mergekey_norm =
1766                     get_mergekey(xdoc, sroot, cl, record_no, service, nmem,
1767                                  se->mergekey);
1768                 if (!mergekey_norm)
1769                 {
1770                     r = -1;
1771                     break;
1772                 }
1773                 for (mkp = &mk; *mkp; mkp = &(*mkp)->next)
1774                     if (!strcmp((*mkp)->value, mergekey_norm))
1775                         break;
1776                 if (!*mkp)
1777                 {
1778                     *mkp = (struct record_metadata_attr*)
1779                         nmem_malloc(nmem, sizeof(**mkp));
1780                     (*mkp)->name = 0;
1781                     (*mkp)->value = nmem_strdup(nmem, mergekey_norm);
1782                     (*mkp)->next = 0;
1783                     no_merge_keys++;
1784                 }
1785                 else
1786                     no_merge_dups++;
1787             }
1788         if (no_merge_keys > 1 || no_merge_dups > 0)
1789         {
1790             yaz_log(YLOG_LOG, "Got %d mergekeys, %d dups for position %d",
1791                     no_merge_keys, no_merge_dups, record_no);
1792         }
1793         for (sroot = root->children; !r && sroot; sroot = sroot->next)
1794             if (sroot->type == XML_ELEMENT_NODE &&
1795                 !strcmp((const char *) sroot->name, "record"))
1796             {
1797                 if (!cached_copy)
1798                     insert_settings_values(sdb, xdoc, root, service);
1799                 r = ingest_sub_record(cl, xdoc, sroot, record_no, nmem, sdb,
1800                                       mk);
1801             }
1802     }
1803     else if (!strcmp((const char *) root->name, "record"))
1804     {
1805         const char *mergekey_norm =
1806             get_mergekey(xdoc, root, cl, record_no, service, nmem,
1807                          se->mergekey);
1808         if (mergekey_norm)
1809         {
1810             struct record_metadata_attr *mk = (struct record_metadata_attr*)
1811                 nmem_malloc(nmem, sizeof(*mk));
1812             mk->name = 0;
1813             mk->value = nmem_strdup(nmem, mergekey_norm);
1814             mk->next = 0;
1815
1816             if (!cached_copy)
1817                 insert_settings_values(sdb, xdoc, root, service);
1818             r = ingest_sub_record(cl, xdoc, root, record_no, nmem, sdb, mk);
1819         }
1820     }
1821     else
1822     {
1823         session_log(se, YLOG_WARN, "Bad pz root element: %s",
1824                     (const char *) root->name);
1825         r = -1;
1826     }
1827     return r;
1828 }
1829
1830
1831 //    struct conf_metadata *ser_md = &service->metadata[md_field_id];
1832 //    struct record_metadata *rec_md = record->metadata[md_field_id];
1833 static int match_metadata_local(struct conf_service *service,
1834                                 struct conf_metadata *ser_md,
1835                                 struct record_metadata *rec_md0,
1836                                 char **values, int num_v)
1837 {
1838     int i;
1839     struct record_metadata *rec_md = rec_md0;
1840     WRBUF val_wr = 0;
1841     WRBUF text_wr = wrbuf_alloc();
1842     for (i = 0; i < num_v; )
1843     {
1844         if (rec_md)
1845         {
1846             if (ser_md->type == Metadata_type_year
1847                 || ser_md->type == Metadata_type_date)
1848             {
1849                 int y = atoi(values[i]);
1850                 if (y >= rec_md->data.number.min
1851                     && y <= rec_md->data.number.max)
1852                     break;
1853             }
1854             else
1855             {
1856                 if (!val_wr)
1857                 {
1858                     val_wr = wrbuf_alloc();
1859                     mergekey_norm_wr(service->charsets, val_wr, values[i]);
1860                 }
1861                 wrbuf_rewind(text_wr);
1862                 mergekey_norm_wr(service->charsets, text_wr,
1863                                  rec_md->data.text.disp);
1864                 if (!strcmp(wrbuf_cstr(val_wr), wrbuf_cstr(text_wr)))
1865                     break;
1866             }
1867             rec_md = rec_md->next;
1868         }
1869         else
1870         {
1871             rec_md = rec_md0;
1872             wrbuf_destroy(val_wr);
1873             val_wr = 0;
1874             i++;
1875         }
1876     }
1877     wrbuf_destroy(val_wr);
1878     wrbuf_destroy(text_wr);
1879     return i < num_v ? 1 : 0;
1880 }
1881
1882 int session_check_cluster_limit(struct session *se, struct record_cluster *rec)
1883 {
1884     int i;
1885     struct conf_service *service = se->service;
1886     int ret = 1;
1887     const char *name;
1888     const char *value;
1889     NMEM nmem_tmp = nmem_create();
1890
1891     for (i = 0; (name = facet_limits_get(se->facet_limits, i, &value)); i++)
1892     {
1893         int j;
1894         for (j = 0; j < service->num_metadata; j++)
1895         {
1896             struct conf_metadata *md = service->metadata + j;
1897             if (!strcmp(md->name, name) && md->limitcluster)
1898             {
1899                 char **values = 0;
1900                 int num = 0;
1901                 int md_field_id =
1902                     conf_service_metadata_field_id(service,
1903                                                    md->limitcluster);
1904
1905                 if (md_field_id < 0)
1906                 {
1907                     ret = 0;
1908                     break;
1909                 }
1910
1911                 nmem_strsplit_escape2(nmem_tmp, "|", value, &values,
1912                                       &num, 1, '\\', 1);
1913
1914                 if (!match_metadata_local(service,
1915                                           &service->metadata[md_field_id],
1916                                           rec->metadata[md_field_id],
1917                                           values, num))
1918                 {
1919                     ret = 0;
1920                     break;
1921                 }
1922             }
1923         }
1924     }
1925     nmem_destroy(nmem_tmp);
1926     return ret;
1927 }
1928
1929 // Skip record on non-zero
1930 static int check_limit_local(struct client *cl,
1931                              struct record *record,
1932                              int record_no)
1933 {
1934     int skip_record = 0;
1935     struct session *se = client_get_session(cl);
1936     struct conf_service *service = se->service;
1937     NMEM nmem_tmp = nmem_create();
1938     struct session_database *sdb = client_get_database(cl);
1939     int l = 0;
1940     while (!skip_record)
1941     {
1942         int md_field_id;
1943         char **values = 0;
1944         int num_v = 0;
1945         const char *name =
1946             client_get_facet_limit_local(cl, sdb, &l, nmem_tmp,
1947                                          &num_v, &values);
1948         if (!name)
1949             break;
1950
1951         if (!strcmp(name, "*"))
1952         {
1953             for (md_field_id = 0; md_field_id < service->num_metadata;
1954                  md_field_id++)
1955             {
1956                 if (match_metadata_local(
1957                         service,
1958                         &service->metadata[md_field_id],
1959                         record->metadata[md_field_id],
1960                         values, num_v))
1961                     break;
1962             }
1963             if (md_field_id == service->num_metadata)
1964                 skip_record = 1;
1965         }
1966         else
1967         {
1968             md_field_id = conf_service_metadata_field_id(service, name);
1969             if (md_field_id < 0)
1970             {
1971                 skip_record = 1;
1972                 break;
1973             }
1974             if (!match_metadata_local(
1975                     service,
1976                     &service->metadata[md_field_id],
1977                     record->metadata[md_field_id],
1978                     values, num_v))
1979             {
1980                 skip_record = 1;
1981             }
1982         }
1983     }
1984     nmem_destroy(nmem_tmp);
1985     return skip_record;
1986 }
1987
1988 static int ingest_to_cluster(struct client *cl,
1989                              xmlDoc *xdoc,
1990                              xmlNode *root,
1991                              int record_no,
1992                              struct record_metadata_attr *merge_keys)
1993 {
1994     xmlNode *n;
1995     xmlChar *type = 0;
1996     xmlChar *value = 0;
1997     struct session *se = client_get_session(cl);
1998     struct conf_service *service = se->service;
1999     int term_factor = 1;
2000     struct record_cluster *cluster;
2001     struct record_metadata **metadata0;
2002     struct session_database *sdb = client_get_database(cl);
2003     NMEM ingest_nmem = 0;
2004     char **rank_values = 0;
2005     int rank_num = 0;
2006     struct record *record = record_create(se->nmem,
2007                                           service->num_metadata,
2008                                           service->num_sortkeys, cl,
2009                                           record_no);
2010
2011     for (n = root->children; n; n = n->next)
2012     {
2013         if (type)
2014             xmlFree(type);
2015         if (value)
2016             xmlFree(value);
2017         type = value = 0;
2018
2019         if (n->type != XML_ELEMENT_NODE)
2020             continue;
2021         if (!strcmp((const char *) n->name, "metadata"))
2022         {
2023             struct conf_metadata *ser_md = 0;
2024             struct record_metadata **wheretoput = 0;
2025             struct record_metadata *rec_md = 0;
2026             int md_field_id = -1;
2027
2028             type = xmlGetProp(n, (xmlChar *) "type");
2029             value = xmlNodeListGetString(xdoc, n->children, 1);
2030             if (!type)
2031                 continue;
2032             if (!value || !*value)
2033             {
2034                 xmlChar *empty = xmlGetProp(n, (xmlChar *) "empty");
2035                 if (!empty)
2036                     continue;
2037                 if (value)
2038                     xmlFree(value);
2039                 value = empty;
2040             }
2041             md_field_id
2042                 = conf_service_metadata_field_id(service, (const char *) type);
2043             if (md_field_id < 0)
2044             {
2045                 if (se->number_of_warnings_unknown_metadata == 0)
2046                 {
2047                     session_log(se, YLOG_WARN,
2048                             "Ignoring unknown metadata element: %s", type);
2049                 }
2050                 se->number_of_warnings_unknown_metadata++;
2051                 continue;
2052             }
2053
2054             ser_md = &service->metadata[md_field_id];
2055
2056             // non-merged metadata
2057             rec_md = record_metadata_init(se->nmem, (const char *) value,
2058                                           ser_md->type, n->properties);
2059             if (!rec_md)
2060             {
2061                 session_log(se, YLOG_WARN, "bad metadata data '%s' "
2062                             "for element '%s'", value, type);
2063                 continue;
2064             }
2065
2066             if (ser_md->type == Metadata_type_generic)
2067             {
2068                 WRBUF w = wrbuf_alloc();
2069                 if (relevance_snippet(se->relevance,
2070                                       (char*) value, ser_md->name, w))
2071                     rec_md->data.text.snippet = nmem_strdup(se->nmem,
2072                                                             wrbuf_cstr(w));
2073                 wrbuf_destroy(w);
2074             }
2075
2076
2077             wheretoput = &record->metadata[md_field_id];
2078             while (*wheretoput)
2079                 wheretoput = &(*wheretoput)->next;
2080             *wheretoput = rec_md;
2081         }
2082     }
2083
2084     if (check_limit_local(cl, record, record_no))
2085     {
2086         if (type)
2087             xmlFree(type);
2088         if (value)
2089             xmlFree(value);
2090         return -2;
2091     }
2092     cluster = reclist_insert(se->reclist, se->relevance, service, record,
2093                              merge_keys, &se->total_merged);
2094     if (!cluster)
2095     {
2096         if (type)
2097             xmlFree(type);
2098         if (value)
2099             xmlFree(value);
2100         return 0; // complete match with existing record
2101     }
2102
2103     {
2104         const char *use_term_factor_str =
2105             session_setting_oneval(sdb, PZ_TERMLIST_TERM_FACTOR);
2106         if (use_term_factor_str && use_term_factor_str[0] == '1')
2107         {
2108             int maxrecs = client_get_maxrecs(cl);
2109             int hits = (int) client_get_hits(cl);
2110             term_factor = MAX(hits, maxrecs) /  MAX(1, maxrecs);
2111             assert(term_factor >= 1);
2112             session_log(se, YLOG_DEBUG, "Using term factor: %d (%d / %d)",
2113                         term_factor, MAX(hits, maxrecs), MAX(1, maxrecs));
2114         }
2115     }
2116
2117     if (global_parameters.dump_records)
2118         session_log(se, YLOG_LOG, "Cluster id %s from %s (#%d)", cluster->recid,
2119                     sdb->database->id, record_no);
2120
2121     // original metadata, to check if first existence of a field
2122     metadata0 = xmalloc(sizeof(*metadata0) * service->num_metadata);
2123     memcpy(metadata0, cluster->metadata,
2124            sizeof(*metadata0) * service->num_metadata);
2125
2126     ingest_nmem = nmem_create();
2127     if (se->rank)
2128     {
2129         yaz_log(YLOG_LOG, "local in sort : %s", se->rank);
2130         nmem_strsplit_escape2(ingest_nmem, ",", se->rank, &rank_values,
2131                               &rank_num, 1, '\\', 1);
2132     }
2133
2134     // now parsing XML record and adding data to cluster or record metadata
2135     for (n = root->children; n; n = n->next)
2136     {
2137         pp2_charset_token_t prt;
2138         if (type)
2139             xmlFree(type);
2140         if (value)
2141             xmlFree(value);
2142         type = value = 0;
2143
2144         if (n->type != XML_ELEMENT_NODE)
2145             continue;
2146         if (!strcmp((const char *) n->name, "metadata"))
2147         {
2148             struct conf_metadata *ser_md = 0;
2149             struct conf_sortkey *ser_sk = 0;
2150             struct record_metadata **wheretoput = 0;
2151             struct record_metadata *rec_md = 0;
2152             int md_field_id = -1;
2153             int sk_field_id = -1;
2154             const char *rank = 0;
2155             xmlChar *xml_rank = 0;
2156
2157             type = xmlGetProp(n, (xmlChar *) "type");
2158             value = xmlNodeListGetString(xdoc, n->children, 1);
2159
2160             if (!type || !value || !*value)
2161                 continue;
2162
2163             md_field_id
2164                 = conf_service_metadata_field_id(service, (const char *) type);
2165             if (md_field_id < 0)
2166                 continue;
2167
2168             ser_md = &service->metadata[md_field_id];
2169
2170             if (ser_md->sortkey_offset >= 0)
2171             {
2172                 sk_field_id = ser_md->sortkey_offset;
2173                 ser_sk = &service->sortkeys[sk_field_id];
2174             }
2175
2176             // merged metadata
2177             rec_md = record_metadata_init(se->nmem, (const char *) value,
2178                                           ser_md->type, 0);
2179
2180             // see if the field was not in cluster already (from beginning)
2181
2182             if (!rec_md)
2183                 continue;
2184
2185             if (rank_num)
2186             {
2187                 int i;
2188                 for (i = 0; i < rank_num; i++)
2189                 {
2190                     const char *val = rank_values[i];
2191                     const char *cp = strchr(val, '=');
2192                     if (!cp)
2193                         continue;
2194                     if ((cp - val) == strlen((const char *) type)
2195                         && !memcmp(val, type, cp - val))
2196                     {
2197                         rank = cp + 1;
2198                         break;
2199                     }
2200                 }
2201             }
2202             else
2203             {
2204                 xml_rank = xmlGetProp(n, (xmlChar *) "rank");
2205                 rank = xml_rank ? (const char *) xml_rank : ser_md->rank;
2206             }
2207
2208             wheretoput = &cluster->metadata[md_field_id];
2209
2210             if (ser_md->merge == Metadata_merge_first)
2211             {
2212                 if (!metadata0[md_field_id])
2213                 {
2214                     while (*wheretoput)
2215                         wheretoput = &(*wheretoput)->next;
2216                     *wheretoput = rec_md;
2217                 }
2218             }
2219             else if (ser_md->merge == Metadata_merge_unique)
2220             {
2221                 while (*wheretoput)
2222                 {
2223                     if (!strcmp((const char *) (*wheretoput)->data.text.disp,
2224                                 rec_md->data.text.disp))
2225                         break;
2226                     wheretoput = &(*wheretoput)->next;
2227                 }
2228                 if (!*wheretoput)
2229                     *wheretoput = rec_md;
2230             }
2231             else if (ser_md->merge == Metadata_merge_longest)
2232             {
2233                 if (!*wheretoput
2234                     || strlen(rec_md->data.text.disp)
2235                     > strlen((*wheretoput)->data.text.disp))
2236                 {
2237                     *wheretoput = rec_md;
2238                     if (ser_sk)
2239                     {
2240                         const char *sort_str = 0;
2241                         int skip_article =
2242                             ser_sk->type == Metadata_type_skiparticle;
2243
2244                         if (!cluster->sortkeys[sk_field_id])
2245                             cluster->sortkeys[sk_field_id] =
2246                                 nmem_malloc(se->nmem,
2247                                             sizeof(union data_types));
2248
2249                         prt =
2250                             pp2_charset_token_create(service->charsets, "sort");
2251
2252                         pp2_charset_token_first(prt, rec_md->data.text.disp,
2253                                                 skip_article);
2254
2255                         pp2_charset_token_next(prt);
2256
2257                         sort_str = pp2_get_sort(prt);
2258
2259                         cluster->sortkeys[sk_field_id]->text.disp =
2260                             rec_md->data.text.disp;
2261                         if (!sort_str)
2262                         {
2263                             sort_str = rec_md->data.text.disp;
2264                             session_log(se, YLOG_WARN,
2265                                     "Could not make sortkey. Bug #1858");
2266                         }
2267                         cluster->sortkeys[sk_field_id]->text.sort =
2268                             nmem_strdup(se->nmem, sort_str);
2269                         pp2_charset_token_destroy(prt);
2270                     }
2271                 }
2272             }
2273             else if (ser_md->merge == Metadata_merge_all)
2274             {
2275                 while (*wheretoput)
2276                     wheretoput = &(*wheretoput)->next;
2277                 *wheretoput = rec_md;
2278             }
2279             else if (ser_md->merge == Metadata_merge_range)
2280             {
2281                 if (!*wheretoput)
2282                 {
2283                     *wheretoput = rec_md;
2284                     if (ser_sk)
2285                         cluster->sortkeys[sk_field_id]
2286                             = &rec_md->data;
2287                 }
2288                 else
2289                 {
2290                     int this_min = rec_md->data.number.min;
2291                     int this_max = rec_md->data.number.max;
2292                     if (this_min < (*wheretoput)->data.number.min)
2293                         (*wheretoput)->data.number.min = this_min;
2294                     if (this_max > (*wheretoput)->data.number.max)
2295                         (*wheretoput)->data.number.max = this_max;
2296                 }
2297             }
2298
2299             // ranking of _all_ fields enabled ...
2300             if (rank)
2301             {
2302                 relevance_countwords(se->relevance, cluster,
2303                                      (char *) value, rank, ser_md->name);
2304             }
2305             // construct facets ... unless the client already has reported them
2306             if (ser_md->termlist && !client_has_facet(cl, (char *) type))
2307             {
2308                 if (ser_md->type == Metadata_type_year)
2309                 {
2310                     char year[64];
2311                     sprintf(year, "%d", rec_md->data.number.max);
2312
2313                     add_facet(se, (char *) type, year, term_factor);
2314                     if (rec_md->data.number.max != rec_md->data.number.min)
2315                     {
2316                         sprintf(year, "%d", rec_md->data.number.min);
2317                         add_facet(se, (char *) type, year, term_factor);
2318                     }
2319                 }
2320                 else
2321                     add_facet(se, (char *) type, (char *) value, term_factor);
2322             }
2323
2324             // cleaning up
2325             if (xml_rank)
2326                 xmlFree(xml_rank);
2327             xmlFree(type);
2328             xmlFree(value);
2329             type = value = 0;
2330         }
2331         else
2332         {
2333             if (se->number_of_warnings_unknown_elements == 0)
2334                 session_log(se, YLOG_WARN,
2335                         "Unexpected element in internal record: %s", n->name);
2336             se->number_of_warnings_unknown_elements++;
2337         }
2338     }
2339     if (type)
2340         xmlFree(type);
2341     if (value)
2342         xmlFree(value);
2343
2344     nmem_destroy(ingest_nmem);
2345     xfree(metadata0);
2346     relevance_donerecord(se->relevance, cluster);
2347     se->total_records++;
2348
2349     return 0;
2350 }
2351
2352 void session_log(struct session *s, int level, const char *fmt, ...)
2353 {
2354     char buf[1024];
2355     va_list ap;
2356     va_start(ap, fmt);
2357
2358     yaz_vsnprintf(buf, sizeof(buf)-30, fmt, ap);
2359     yaz_log(level, "Session %u: %s", s ? s->session_id : 0, buf);
2360
2361     va_end(ap);
2362 }
2363
2364 /*
2365  * Local variables:
2366  * c-basic-offset: 4
2367  * c-file-style: "Stroustrup"
2368  * indent-tabs-mode: nil
2369  * End:
2370  * vim: shiftwidth=4 tabstop=8 expandtab
2371  */
2372