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 void session_sort(struct session *se, struct reclist_sortparms *sp,
 625                   const char *mergekey, const char *rank)
 626 {
 627     struct client_list *l;
 628     const char *field = sp->name;
 629     int increasing = sp->increasing;
 630     int type  = sp->type;
 631     int clients_research = 0;
 632
 633     session_enter(se, "session_sort");
 634     session_log(se, YLOG_DEBUG, "session_sort field=%s increasing=%d type=%d",
 635                 field, increasing, type);
 636
 637     if (rank && (!se->rank || strcmp(se->rank, rank)))
 638     {
 639         /* new rank must research/reingest anyway */
 640         assert(rank);
 641         xfree(se->rank);
 642         se->rank = *rank ? xstrdup(rank) : 0;
 643         clients_research = 1;
 644         session_log(se, YLOG_DEBUG, "session_sort: new rank = %s",
 645                     rank);
 646     }
 647     if (mergekey && (!se->mergekey || strcmp(se->mergekey, mergekey)))
 648     {
 649         /* new mergekey must research/reingest anyway */
 650         assert(mergekey);
 651         xfree(se->mergekey);
 652         se->mergekey = *mergekey ? xstrdup(mergekey) : 0;
 653         clients_research = 1;
 654         session_log(se, YLOG_DEBUG, "session_sort: new mergekey = %s",
 655                     mergekey);
 656     }
 657     if (clients_research == 0)
 658     {
 659         struct reclist_sortparms *sr;
 660         for (sr = se->sorted_results; sr; sr = sr->next)
 661             if (!reclist_sortparms_cmp(sr, sp))
 662                 break;
 663         if (sr)
 664         {
 665             session_log(se, YLOG_DEBUG, "session_sort: field=%s increasing=%d type=%d already fetched",
 666                         field, increasing, type);
 667             session_leave(se, "session_sort");
 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 || se->clients_starting)
 686     {
 687         // A new sorting based on same record set
 688         struct reclist_sortparms *sr = nmem_malloc(se->nmem, sizeof(*sr));
 689         sr->name = nmem_strdup(se->nmem, field);
 690         sr->increasing = increasing;
 691         sr->type = type;
 692         sr->next = se->sorted_results;
 693         se->sorted_results = sr;
 694         session_log(se, YLOG_DEBUG, "session_sort: no research/ingesting done");
 695         session_leave(se, "session_sort");
 696     }
 697     else
 698     {
 699         se->clients_starting = 1;
 700         session_log(se, YLOG_DEBUG,
 701                     "session_sort: reset results due to %d clients researching",
 702                     clients_research);
 703         session_clear_set(se, sp);
 704         session_log(se, YLOG_DEBUG, "Re- search/ingesting for clients due to change in sort order");
 705
 706         session_leave(se, "session_sort");
 707         for (l = se->clients_active; l; l = l->next)
 708         {
 709             struct client *cl = l->client;
 710             if (client_get_state(cl) == Client_Connecting ||
 711                 client_get_state(cl) == Client_Idle ||
 712                 client_get_state(cl) == Client_Working) {
 713                 client_start_search(cl);
 714             }
 715             else
 716             {
 717                 session_log(se, YLOG_DEBUG,
 718                             "session_sort: %s: No re-start/ingest in show. "
 719                             "Wrong client state: %d",
 720                             client_get_id(cl), client_get_state(cl));
 721             }
 722         }
 723         session_enter(se, "session_sort");
 724         se->clients_starting = 0;
 725         session_leave(se, "session_sort");
 726     }
 727 }
 728
 729 enum pazpar2_error_code session_search(struct session *se,
 730                                        const char *query,
 731                                        const char *startrecs,
 732                                        const char *maxrecs,
 733                                        const char *filter,
 734                                        const char *limit,
 735                                        const char **addinfo,
 736                                        const char **addinfo2,
 737                                        struct reclist_sortparms *sp,
 738                                        const char *mergekey,
 739                                        const char *rank)
 740 {
 741     int live_channels = 0;
 742     int no_working = 0;
 743     int no_failed_query = 0;
 744     int no_failed_limit = 0;
 745     struct client_list *l;
 746
 747     session_log(se, YLOG_DEBUG, "Search");
 748
 749     *addinfo = 0;
 750
 751     session_enter(se, "session_search0");
 752     if (se->clients_starting)
 753     {
 754         session_leave(se, "session_search0");
 755         return PAZPAR2_NO_ERROR;
 756     }
 757     se->clients_starting = 1;
 758     session_leave(se, "session_search0");
 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         se->clients_starting = 0;
 788         return PAZPAR2_NO_TARGETS;
 789     }
 790
 791     facet_limits_destroy(se->facet_limits);
 792     se->facet_limits = facet_limits_create(limit);
 793     if (!se->facet_limits)
 794     {
 795         *addinfo = "limit";
 796         session_leave(se, "session_search");
 797         se->clients_starting = 0;
 798         return PAZPAR2_MALFORMED_PARAMETER_VALUE;
 799     }
 800
 801     session_leave(se, "session_search");
 802
 803     session_alert_watch(se, SESSION_WATCH_SHOW);
 804     session_alert_watch(se, SESSION_WATCH_BYTARGET);
 805     session_alert_watch(se, SESSION_WATCH_TERMLIST);
 806     session_alert_watch(se, SESSION_WATCH_SHOW_PREF);
 807
 808     for (l = se->clients_active; l; l = l->next)
 809     {
 810         int parse_ret;
 811         struct client *cl = l->client;
 812         client_parse_init(cl, 1);
 813         if (prepare_map(se, client_get_database(cl)) < 0)
 814             continue;
 815
 816         parse_ret = client_parse_query(cl, query, se->facet_limits, addinfo2);
 817         if (parse_ret == -1)
 818             no_failed_query++;
 819         else if (parse_ret == -2)
 820             no_failed_limit++;
 821         else if (parse_ret < 0)
 822             no_working++; /* other error, such as bad CCL map */
 823         else
 824         {
 825             client_parse_range(cl, startrecs, maxrecs);
 826             client_parse_sort(cl, sp);
 827             client_start_search(cl);
 828             no_working++;
 829         }
 830     }
 831     session_enter(se, "session_search2");
 832     se->clients_starting = 0;
 833     session_leave(se, "session_search2");
 834     if (no_working == 0)
 835     {
 836         if (no_failed_query > 0)
 837         {
 838             *addinfo = "query";
 839             return PAZPAR2_MALFORMED_PARAMETER_VALUE;
 840         }
 841         else if (no_failed_limit > 0)
 842         {
 843             *addinfo = "limit";
 844             return PAZPAR2_MALFORMED_PARAMETER_VALUE;
 845         }
 846         else
 847             return PAZPAR2_NO_TARGETS;
 848     }
 849     return PAZPAR2_NO_ERROR;
 850 }
 851
 852 // Creates a new session_database object for a database
 853 static void session_init_databases_fun(void *context, struct database *db)
 854 {
 855     struct session *se = (struct session *) context;
 856     struct session_database *new = nmem_malloc(se->session_nmem, sizeof(*new));
 857     int i;
 858
 859     new->database = db;
 860
 861     new->map = 0;
 862     assert(db->settings);
 863     new->settings = nmem_malloc(se->session_nmem,
 864                                 sizeof(struct settings *) * db->num_settings);
 865     new->num_settings = db->num_settings;
 866     for (i = 0; i < db->num_settings; i++)
 867     {
 868         struct setting *setting = db->settings[i];
 869         new->settings[i] = setting;
 870     }
 871     new->next = se->databases;
 872     se->databases = new;
 873 }
 874
 875 // Doesn't free memory associated with sdb -- nmem takes care of that
 876 static void session_database_destroy(struct session_database *sdb)
 877 {
 878     sdb->map = 0;
 879 }
 880
 881 // Initialize session_database list -- this represents this session's view
 882 // of the database list -- subject to modification by the settings ws command
 883 void session_init_databases(struct session *se)
 884 {
 885     se->databases = 0;
 886     predef_grep_databases(se, se->service, session_init_databases_fun);
 887 }
 888
 889 // Probably session_init_databases_fun should be refactored instead of
 890 // called here.
 891 static struct session_database *load_session_database(struct session *se,
 892                                                       const char *id)
 893 {
 894     struct database *db = new_database_inherit_settings(id, se->session_nmem, se->service->settings);
 895     session_init_databases_fun((void*) se, db);
 896
 897     // New sdb is head of se->databases list
 898     return se->databases;
 899 }
 900
 901 // Find an existing session database. If not found, load it
 902 static struct session_database *find_session_database(struct session *se,
 903                                                       const char *id)
 904 {
 905     struct session_database *sdb;
 906
 907     for (sdb = se->databases; sdb; sdb = sdb->next)
 908         if (!strcmp(sdb->database->id, id))
 909             return sdb;
 910     return load_session_database(se, id);
 911 }
 912
 913 // Apply a session override to a database
 914 void session_apply_setting(struct session *se, const char *dbname,
 915                            const char *name, const char *value)
 916 {
 917     session_enter(se, "session_apply_setting");
 918     {
 919         struct session_database *sdb = find_session_database(se, dbname);
 920         struct conf_service *service = se->service;
 921         struct setting *s;
 922         int offset = settings_create_offset(service, name);
 923
 924         expand_settings_array(&sdb->settings, &sdb->num_settings, offset,
 925                               se->session_nmem);
 926         // Force later recompute of settings-driven data structures
 927         // (happens when a search starts and client connections are prepared)
 928         if (offset == PZ_XSLT)
 929             sdb->map = 0;
 930         se->settings_modified = 1;
 931         for (s = sdb->settings[offset]; s; s = s->next)
 932             if (!strcmp(s->name, name) &&
 933                 dbname && s->target && !strcmp(dbname, s->target))
 934                 break;
 935         if (!s)
 936         {
 937             s = nmem_malloc(se->session_nmem, sizeof(*s));
 938             s->precedence = 0;
 939             s->target = nmem_strdup(se->session_nmem, dbname);
 940             s->name = nmem_strdup(se->session_nmem, name);
 941             s->next = sdb->settings[offset];
 942             sdb->settings[offset] = s;
 943         }
 944         s->value = nmem_strdup(se->session_nmem, value);
 945     }
 946     session_leave(se, "session_apply_setting");
 947 }
 948
 949 void session_destroy(struct session *se)
 950 {
 951     struct session_database *sdb;
 952     session_log(se, YLOG_LOG, "destroy");
 953     session_use(-1);
 954     session_remove_cached_clients(se);
 955
 956     for (sdb = se->databases; sdb; sdb = sdb->next)
 957         session_database_destroy(sdb);
 958     normalize_cache_destroy(se->normalize_cache);
 959     relevance_destroy(&se->relevance);
 960     reclist_destroy(se->reclist);
 961     xfree(se->mergekey);
 962     xfree(se->rank);
 963     if (nmem_total(se->nmem))
 964         session_log(se, YLOG_DEBUG, "NMEN operation usage %zd", nmem_total(se->nmem));
 965     if (nmem_total(se->session_nmem))
 966         session_log(se, YLOG_DEBUG, "NMEN session usage %zd", nmem_total(se->session_nmem));
 967     facet_limits_destroy(se->facet_limits);
 968     nmem_destroy(se->nmem);
 969     service_destroy(se->service);
 970     yaz_mutex_destroy(&se->session_mutex);
 971 }
 972
 973 size_t session_get_memory_status(struct session *session) {
 974     size_t session_nmem;
 975     if (session == 0)
 976         return 0;
 977     session_enter(session, "session_get_memory_status");
 978     session_nmem = nmem_total(session->nmem);
 979     session_leave(session, "session_get_memory_status");
 980     return session_nmem;
 981 }
 982
 983
 984 struct session *new_session(NMEM nmem, struct conf_service *service,
 985                             unsigned session_id)
 986 {
 987     int i;
 988     struct session *session = nmem_malloc(nmem, sizeof(*session));
 989
 990     char tmp_str[50];
 991
 992     sprintf(tmp_str, "session#%u", session_id);
 993
 994     session->session_id = session_id;
 995     session_log(session, YLOG_DEBUG, "New");
 996     session->service = service;
 997     session->relevance = 0;
 998     session->total_records = 0;
 999     session->number_of_warnings_unknown_elements = 0;
1000     session->number_of_warnings_unknown_metadata = 0;
1001     session->num_termlists = 0;
1002     session->reclist = reclist_create(nmem);
1003     session->clients_active = 0;
1004     session->clients_cached = 0;
1005     session->settings_modified = 0;
1006     session->session_nmem = nmem;
1007     session->nmem = nmem_create();
1008     session->databases = 0;
1009     session->sorted_results = 0;
1010     session->facet_limits = 0;
1011     session->mergekey = 0;
1012     session->rank = 0;
1013     session->clients_starting = 0;
1014
1015     for (i = 0; i <= SESSION_WATCH_MAX; i++)
1016     {
1017         session->watchlist[i].data = 0;
1018         session->watchlist[i].fun = 0;
1019     }
1020     session->normalize_cache = normalize_cache_create();
1021     session->session_mutex = 0;
1022     pazpar2_mutex_create(&session->session_mutex, tmp_str);
1023     session_log(session, YLOG_LOG, "create");
1024
1025     session_use(1);
1026     return session;
1027 }
1028
1029 const char * client_get_suggestions_xml(struct client *cl, WRBUF wrbuf);
1030
1031 static struct hitsbytarget *hitsbytarget_nb(struct session *se,
1032                                             int *count, NMEM nmem)
1033 {
1034     struct hitsbytarget *res = 0;
1035     struct client_list *l;
1036     size_t sz = 0;
1037
1038     for (l = se->clients_active; l; l = l->next)
1039         sz++;
1040
1041     res = nmem_malloc(nmem, sizeof(*res) * sz);
1042     *count = 0;
1043     for (l = se->clients_active; l; l = l->next)
1044     {
1045         struct client *cl = l->client;
1046         WRBUF w = wrbuf_alloc();
1047         const char *name = session_setting_oneval(client_get_database(cl),
1048                                                   PZ_NAME);
1049         res[*count].id = client_get_id(cl);
1050         res[*count].name = *name ? name : "Unknown";
1051         res[*count].hits = client_get_hits(cl);
1052         res[*count].approximation = client_get_approximation(cl);
1053         res[*count].records = client_get_num_records(cl,
1054                                                      &res[*count].filtered,
1055                                                      0, 0);
1056         res[*count].diagnostic =
1057             client_get_diagnostic(cl, &res[*count].message,
1058                                   &res[*count].addinfo);
1059         res[*count].state = client_get_state_str(cl);
1060         res[*count].connected  = client_get_connection(cl) ? 1 : 0;
1061         session_settings_dump(se, client_get_database(cl), w);
1062         res[*count].settings_xml = nmem_strdup(nmem, wrbuf_cstr(w));
1063         wrbuf_rewind(w);
1064         wrbuf_puts(w, "");
1065         res[*count].suggestions_xml = nmem_strdup(nmem, client_get_suggestions_xml(cl, w));
1066         wrbuf_destroy(w);
1067         (*count)++;
1068     }
1069     return res;
1070 }
1071
1072 struct hitsbytarget *get_hitsbytarget(struct session *se, int *count, NMEM nmem)
1073 {
1074     struct hitsbytarget *p;
1075     session_enter(se, "get_hitsbytarget");
1076     p = hitsbytarget_nb(se, count, nmem);
1077     session_leave(se, "get_hitsbytarget");
1078     return p;
1079 }
1080
1081 // Compares two hitsbytarget nodes by hitcount
1082 static int cmp_ht(const void *p1, const void *p2)
1083 {
1084     const struct hitsbytarget *h1 = p1;
1085     const struct hitsbytarget *h2 = p2;
1086     return h2->hits - h1->hits;
1087 }
1088
1089 // Compares two hitsbytarget nodes by hitcount
1090 static int cmp_ht_approx(const void *p1, const void *p2)
1091 {
1092     const struct hitsbytarget *h1 = p1;
1093     const struct hitsbytarget *h2 = p2;
1094     return h2->approximation - h1->approximation;
1095 }
1096
1097 static int targets_termlist_nb(WRBUF wrbuf, struct session *se, int num,
1098                                NMEM nmem, int version)
1099 {
1100     struct hitsbytarget *ht;
1101     int count, i;
1102
1103     ht = hitsbytarget_nb(se, &count, nmem);
1104     if (version >= 2)
1105         qsort(ht, count, sizeof(struct hitsbytarget), cmp_ht_approx);
1106     else
1107         qsort(ht, count, sizeof(struct hitsbytarget), cmp_ht);
1108     for (i = 0; i < count && i < num && ht[i].hits > 0; i++)
1109     {
1110
1111         // do only print terms which have display names
1112
1113         wrbuf_puts(wrbuf, "<term>\n");
1114
1115         wrbuf_puts(wrbuf, "<id>");
1116         wrbuf_xmlputs(wrbuf, ht[i].id);
1117         wrbuf_puts(wrbuf, "</id>\n");
1118
1119         wrbuf_puts(wrbuf, "<name>");
1120         if (!ht[i].name || !ht[i].name[0])
1121             wrbuf_xmlputs(wrbuf, "NO TARGET NAME");
1122         else
1123             wrbuf_xmlputs(wrbuf, ht[i].name);
1124         wrbuf_puts(wrbuf, "</name>\n");
1125
1126         wrbuf_printf(wrbuf, "<frequency>" ODR_INT_PRINTF "</frequency>\n",
1127                      ht[i].hits);
1128
1129         if (version >= 2) {
1130             // Should not print if we know it isn't a approximation.
1131             wrbuf_printf(wrbuf, "<approximation>" ODR_INT_PRINTF "</approximation>\n", ht[i].approximation);
1132             wrbuf_printf(wrbuf, "<records>%d</records>\n", ht[i].records - ht[i].filtered);
1133             wrbuf_printf(wrbuf, "<filtered>%d</filtered>\n", ht[i].filtered);
1134         }
1135
1136         wrbuf_puts(wrbuf, "<state>");
1137         wrbuf_xmlputs(wrbuf, ht[i].state);
1138         wrbuf_puts(wrbuf, "</state>\n");
1139
1140         wrbuf_printf(wrbuf, "<diagnostic>%d</diagnostic>\n",
1141                      ht[i].diagnostic);
1142         wrbuf_puts(wrbuf, "</term>\n");
1143     }
1144     return count;
1145 }
1146
1147 void perform_termlist(struct http_channel *c, struct session *se,
1148                       const char *name, int num, int version)
1149 {
1150     int i, j;
1151     NMEM nmem_tmp = nmem_create();
1152     char **names;
1153     int num_names = 0;
1154
1155     if (!name)
1156         name = "*";
1157
1158     nmem_strsplit(nmem_tmp, ",", name, &names, &num_names);
1159
1160     session_enter(se, "perform_termlist");
1161
1162     for (j = 0; j < num_names; j++)
1163     {
1164         const char *tname;
1165         int must_generate_empty = 1; /* bug 5350 */
1166
1167         for (i = 0; i < se->num_termlists; i++)
1168         {
1169             tname = se->termlists[i].name;
1170             if (!strcmp(names[j], tname) || !strcmp(names[j], "*"))
1171             {
1172                 struct termlist_score **p = 0;
1173                 int len;
1174
1175                 wrbuf_puts(c->wrbuf, "<list name=\"");
1176                 wrbuf_xmlputs(c->wrbuf, tname);
1177                 wrbuf_puts(c->wrbuf, "\">\n");
1178                 must_generate_empty = 0;
1179
1180                 p = termlist_highscore(se->termlists[i].termlist, &len,
1181                                        nmem_tmp);
1182                 if (p)
1183                 {
1184                     int i;
1185                     for (i = 0; i < len && i < num; i++)
1186                     {
1187                         // prevent sending empty term elements
1188                         if (!p[i]->display_term || !p[i]->display_term[0])
1189                             continue;
1190
1191                         wrbuf_puts(c->wrbuf, "<term>");
1192                         wrbuf_puts(c->wrbuf, "<name>");
1193                         wrbuf_xmlputs(c->wrbuf, p[i]->display_term);
1194                         wrbuf_puts(c->wrbuf, "</name>");
1195
1196                         wrbuf_printf(c->wrbuf,
1197                                      "<frequency>%d</frequency>",
1198                                      p[i]->frequency);
1199                         wrbuf_puts(c->wrbuf, "</term>\n");
1200                     }
1201                 }
1202                 wrbuf_puts(c->wrbuf, "</list>\n");
1203             }
1204         }
1205         tname = "xtargets";
1206         if (!strcmp(names[j], tname) || !strcmp(names[j], "*"))
1207         {
1208             wrbuf_puts(c->wrbuf, "<list name=\"");
1209             wrbuf_xmlputs(c->wrbuf, tname);
1210             wrbuf_puts(c->wrbuf, "\">\n");
1211
1212             targets_termlist_nb(c->wrbuf, se, num, c->nmem, version);
1213             wrbuf_puts(c->wrbuf, "</list>\n");
1214             must_generate_empty = 0;
1215         }
1216         if (must_generate_empty)
1217         {
1218             wrbuf_puts(c->wrbuf, "<list name=\"");
1219             wrbuf_xmlputs(c->wrbuf, names[j]);
1220             wrbuf_puts(c->wrbuf, "\"/>\n");
1221         }
1222     }
1223     session_leave(se, "perform_termlist");
1224     nmem_destroy(nmem_tmp);
1225 }
1226
1227 #ifdef MISSING_HEADERS
1228 void report_nmem_stats(void)
1229 {
1230     size_t in_use, is_free;
1231
1232     nmem_get_memory_in_use(&in_use);
1233     nmem_get_memory_free(&is_free);
1234
1235     yaz_log(YLOG_LOG, "nmem stat: use=%ld free=%ld",
1236             (long) in_use, (long) is_free);
1237 }
1238 #endif
1239
1240 struct record_cluster *show_single_start(struct session *se, const char *id,
1241                                          struct record_cluster **prev_r,
1242                                          struct record_cluster **next_r)
1243 {
1244     struct record_cluster *r = 0;
1245
1246     session_enter(se, "show_single_start");
1247     *prev_r = 0;
1248     *next_r = 0;
1249     reclist_limit(se->reclist, se, 1);
1250
1251     reclist_enter(se->reclist);
1252     while ((r = reclist_read_record(se->reclist)))
1253     {
1254         if (!strcmp(r->recid, id))
1255         {
1256             *next_r = reclist_read_record(se->reclist);
1257             break;
1258         }
1259         *prev_r = r;
1260     }
1261     reclist_leave(se->reclist);
1262     if (!r)
1263         session_leave(se, "show_single_start");
1264     return r;
1265 }
1266
1267 void show_single_stop(struct session *se, struct record_cluster *rec)
1268 {
1269     session_leave(se, "show_single_stop");
1270 }
1271
1272
1273 int session_fetch_more(struct session *se)
1274 {
1275     struct client_list *l;
1276     int ret = 0;
1277
1278     for (l = se->clients_active; l; l = l->next)
1279     {
1280         struct client *cl = l->client;
1281         if (client_get_state(cl) == Client_Idle)
1282         {
1283             if (client_fetch_more(cl))
1284             {
1285                 session_log(se, YLOG_LOG, "%s: more to fetch",
1286                             client_get_id(cl));
1287                 ret = 1;
1288             }
1289             else
1290             {
1291                 int filtered;
1292                 int ingest_failures;
1293                 int record_failures;
1294                 int num = client_get_num_records(
1295                     cl, &filtered, &ingest_failures, &record_failures);
1296
1297                 session_log(se, YLOG_LOG, "%s: hits=" ODR_INT_PRINTF
1298                             " fetched=%d filtered=%d",
1299                             client_get_id(cl),
1300                             client_get_hits(cl),
1301                             num, filtered);
1302                 if (ingest_failures || record_failures)
1303                 {
1304                     session_log(se, YLOG_WARN, "%s:"
1305                                 " ingest failures=%d record failures=%d",
1306                                 client_get_id(cl),
1307                                 ingest_failures, record_failures);
1308                 }
1309             }
1310         }
1311         else
1312         {
1313             session_log(se, YLOG_LOG, "%s: no fetch due to state=%s",
1314                         client_get_id(cl), client_get_state_str(cl));
1315         }
1316
1317     }
1318     return ret;
1319 }
1320
1321 struct record_cluster **show_range_start(struct session *se,
1322                                          struct reclist_sortparms *sp,
1323                                          int start, int *num, int *total,
1324                                          Odr_int *sumhits, Odr_int *approx_hits,
1325                                          void (*show_records_ready)(void *data),
1326                                          struct http_channel *chan)
1327 {
1328     struct record_cluster **recs = 0;
1329     struct reclist_sortparms *spp;
1330     struct client_list *l;
1331     int i;
1332 #if USE_TIMING
1333     yaz_timing_t t = yaz_timing_create();
1334 #endif
1335     session_enter(se, "show_range_start");
1336     *sumhits = 0;
1337     *approx_hits = 0;
1338     *total = 0;
1339     reclist_limit(se->reclist, se, 0);
1340     if (se->relevance)
1341     {
1342         for (spp = sp; spp; spp = spp->next)
1343             if (spp->type == Metadata_type_relevance)
1344             {
1345                 relevance_prepare_read(se->relevance, se->reclist);
1346                 break;
1347             }
1348         for (l = se->clients_active; l; l = l->next) {
1349             *sumhits += client_get_hits(l->client);
1350             *approx_hits += client_get_approximation(l->client);
1351         }
1352     }
1353     reclist_sort(se->reclist, sp);
1354
1355     reclist_enter(se->reclist);
1356     *total = reclist_get_num_records(se->reclist);
1357
1358     for (l = se->clients_active; l; l = l->next)
1359         client_update_show_stat(l->client, 0);
1360
1361     for (i = 0; i < start; i++)
1362     {
1363         struct record_cluster *r = reclist_read_record(se->reclist);
1364         if (!r)
1365         {
1366             *num = 0;
1367             break;
1368         }
1369         else
1370         {
1371             struct record *rec = r->records;
1372             for (;rec; rec = rec->next)
1373                 client_update_show_stat(rec->client, 1);
1374         }
1375     }
1376     recs = nmem_malloc(se->nmem, (*num > 0 ? *num : 1) * sizeof(*recs));
1377     for (i = 0; i < *num; i++)
1378     {
1379         struct record_cluster *r = reclist_read_record(se->reclist);
1380         if (!r)
1381         {
1382             *num = i;
1383             break;
1384         }
1385         else
1386         {
1387             struct record *rec = r->records;
1388             for (;rec; rec = rec->next)
1389                 client_update_show_stat(rec->client, 1);
1390             recs[i] = r;
1391         }
1392     }
1393     reclist_leave(se->reclist);
1394 #if USE_TIMING
1395     yaz_timing_stop(t);
1396     session_log(se, YLOG_LOG, "show %6.5f %3.2f %3.2f",
1397             yaz_timing_get_real(t), yaz_timing_get_user(t),
1398             yaz_timing_get_sys(t));
1399     yaz_timing_destroy(&t);
1400 #endif
1401
1402     if (!session_fetch_more(se))
1403         session_log(se, YLOG_LOG, "can not fetch more");
1404     else
1405     {
1406         show_range_stop(se, recs);
1407         session_log(se, YLOG_LOG, "fetching more in progress");
1408         if (session_set_watch(se, SESSION_WATCH_SHOW,
1409                               show_records_ready, chan, chan))
1410         {
1411             session_log(se, YLOG_WARN, "Ignoring show block");
1412             session_enter(se, "show_range_start");
1413         }
1414         else
1415         {
1416             session_log(se, YLOG_LOG, "session watch OK");
1417             return 0;
1418         }
1419     }
1420     return recs;
1421 }
1422
1423 void show_range_stop(struct session *se, struct record_cluster **recs)
1424 {
1425     session_leave(se, "show_range_stop");
1426 }
1427
1428 void statistics(struct session *se, struct statistics *stat)
1429 {
1430     struct client_list *l;
1431     int count = 0;
1432
1433     memset(stat, 0, sizeof(*stat));
1434     stat->num_hits = 0;
1435     for (l = se->clients_active; l; l = l->next)
1436     {
1437         struct client *cl = l->client;
1438         if (!client_get_connection(cl))
1439             stat->num_no_connection++;
1440         stat->num_hits += client_get_hits(cl);
1441         switch (client_get_state(cl))
1442         {
1443         case Client_Connecting: stat->num_connecting++; break;
1444         case Client_Working: stat->num_working++; break;
1445         case Client_Idle: stat->num_idle++; break;
1446         case Client_Failed: stat->num_failed++; break;
1447         case Client_Error: stat->num_error++; break;
1448         default: break;
1449         }
1450         count++;
1451     }
1452     stat->num_records = se->total_records;
1453
1454     stat->num_clients = count;
1455 }
1456
1457 static struct record_metadata *record_metadata_init(
1458     NMEM nmem, const char *value, enum conf_metadata_type type,
1459     struct _xmlAttr *attr)
1460 {
1461     struct record_metadata *rec_md = record_metadata_create(nmem);
1462     struct record_metadata_attr **attrp = &rec_md->attributes;
1463
1464     for (; attr; attr = attr->next)
1465     {
1466         if (attr->children && attr->children->content)
1467         {
1468             if (strcmp((const char *) attr->name, "type")
1469                 && strcmp((const char *) attr->name, "empty"))
1470             {  /* skip the "type" + "empty" attribute..
1471                   The "Type" is already part of the element in output
1472                   (md-%s) and so repeating it here is redundant */
1473                 *attrp = nmem_malloc(nmem, sizeof(**attrp));
1474                 (*attrp)->name =
1475                     nmem_strdup(nmem, (const char *) attr->name);
1476                 (*attrp)->value =
1477                     nmem_strdup(nmem, (const char *) attr->children->content);
1478                 attrp = &(*attrp)->next;
1479             }
1480         }
1481     }
1482     *attrp = 0;
1483
1484     switch (type)
1485     {
1486     case Metadata_type_generic:
1487     case Metadata_type_skiparticle:
1488         if (strstr(value, "://")) /* looks like a URL */
1489             rec_md->data.text.disp = nmem_strdup(nmem, value);
1490         else
1491             rec_md->data.text.disp =
1492                 normalize7bit_generic(nmem_strdup(nmem, value), " ,/.:([");
1493         rec_md->data.text.sort = 0;
1494         rec_md->data.text.snippet = 0;
1495         break;
1496     case Metadata_type_year:
1497     case Metadata_type_date:
1498     {
1499         int first, last;
1500         int longdate = 0;
1501
1502         if (type == Metadata_type_date)
1503             longdate = 1;
1504         if (extract7bit_dates((char *) value, &first, &last, longdate) < 0)
1505             return 0;
1506
1507         rec_md->data.number.min = first;
1508         rec_md->data.number.max = last;
1509     }
1510     break;
1511     case Metadata_type_float:
1512         rec_md->data.fnumber = atof(value);
1513         break;
1514     case Metadata_type_relevance:
1515     case Metadata_type_position:
1516         return 0;
1517     }
1518     return rec_md;
1519 }
1520
1521 static void mergekey_norm_wr(pp2_charset_fact_t charsets,
1522                              WRBUF norm_wr, const char *value)
1523 {
1524     const char *norm_str;
1525     pp2_charset_token_t prt =
1526         pp2_charset_token_create(charsets, "mergekey");
1527
1528     pp2_charset_token_first(prt, value, 0);
1529     while ((norm_str = pp2_charset_token_next(prt)))
1530     {
1531         if (*norm_str)
1532         {
1533             if (wrbuf_len(norm_wr))
1534                 wrbuf_puts(norm_wr, " ");
1535             wrbuf_puts(norm_wr, norm_str);
1536         }
1537     }
1538     pp2_charset_token_destroy(prt);
1539 }
1540
1541 static int get_mergekey_from_doc(xmlDoc *doc, xmlNode *root, const char *name,
1542                                  struct conf_service *service, WRBUF norm_wr)
1543 {
1544     xmlNode *n;
1545     int no_found = 0;
1546     for (n = root->children; n; n = n->next)
1547     {
1548         if (n->type != XML_ELEMENT_NODE)
1549             continue;
1550         if (!strcmp((const char *) n->name, "metadata"))
1551         {
1552             xmlChar *type = xmlGetProp(n, (xmlChar *) "type");
1553             if (type == NULL) {
1554                 yaz_log(YLOG_FATAL, "Missing type attribute on metadata element. Skipping!");
1555             }
1556             else if (!strcmp(name, (const char *) type))
1557             {
1558                 xmlChar *value = xmlNodeListGetString(doc, n->children, 1);
1559                 if (value && *value)
1560                 {
1561                     if (wrbuf_len(norm_wr) > 0)
1562                         wrbuf_puts(norm_wr, " ");
1563                     wrbuf_puts(norm_wr, name);
1564                     mergekey_norm_wr(service->charsets, norm_wr,
1565                                      (const char *) value);
1566                     no_found++;
1567                 }
1568                 if (value)
1569                     xmlFree(value);
1570             }
1571             xmlFree(type);
1572         }
1573     }
1574     return no_found;
1575 }
1576
1577 static const char *get_mergekey(xmlDoc *doc, xmlNode *root,
1578                                 struct client *cl, int record_no,
1579                                 struct conf_service *service, NMEM nmem,
1580                                 const char *session_mergekey)
1581 {
1582     char *mergekey_norm = 0;
1583     WRBUF norm_wr = wrbuf_alloc();
1584     xmlChar *mergekey;
1585
1586     if (session_mergekey)
1587     {
1588         int i, num = 0;
1589         char **values = 0;
1590         nmem_strsplit_escape2(nmem, ",", session_mergekey, &values,
1591                               &num, 1, '\\', 1);
1592
1593         for (i = 0; i < num; i++)
1594             get_mergekey_from_doc(doc, root, values[i], service, norm_wr);
1595     }
1596     else if ((mergekey = xmlGetProp(root, (xmlChar *) "mergekey")))
1597     {
1598         mergekey_norm_wr(service->charsets, norm_wr, (const char *) mergekey);
1599         xmlFree(mergekey);
1600     }
1601     else
1602     {
1603         /* no mergekey defined in XSL. Look for mergekey metadata instead */
1604         int field_id;
1605         for (field_id = 0; field_id < service->num_metadata; field_id++)
1606         {
1607             struct conf_metadata *ser_md = &service->metadata[field_id];
1608             if (ser_md->mergekey != Metadata_mergekey_no)
1609             {
1610                 int r = get_mergekey_from_doc(doc, root, ser_md->name,
1611                                               service, norm_wr);
1612                 if (r == 0 && ser_md->mergekey == Metadata_mergekey_required)
1613                 {
1614                     /* no mergekey on this one and it is required..
1615                        Generate unique key instead */
1616                     wrbuf_rewind(norm_wr);
1617                     break;
1618                 }
1619             }
1620         }
1621     }
1622
1623     /* generate unique key if none is not generated already or is empty */
1624     if (wrbuf_len(norm_wr) == 0)
1625     {
1626         wrbuf_printf(norm_wr, "position: %s-%d",
1627                      client_get_id(cl), record_no);
1628     }
1629     else
1630     {
1631         const char *lead = "content: ";
1632         wrbuf_insert(norm_wr, 0, lead, strlen(lead));
1633     }
1634     if (wrbuf_len(norm_wr) > 0)
1635         mergekey_norm = nmem_strdup(nmem, wrbuf_cstr(norm_wr));
1636     wrbuf_destroy(norm_wr);
1637     return mergekey_norm;
1638 }
1639
1640 /** \brief see if metadata for pz:recordfilter exists
1641     \param root xml root element of normalized record
1642     \param sdb session database for client
1643     \retval 0 if there is no metadata for pz:recordfilter
1644     \retval 1 if there is metadata for pz:recordfilter
1645
1646     If there is no pz:recordfilter defined, this function returns 1
1647     as well.
1648 */
1649
1650 static int check_record_filter(xmlNode *root, struct session_database *sdb)
1651 {
1652     int match = 0;
1653     xmlNode *n;
1654     const char *s;
1655     s = session_setting_oneval(sdb, PZ_RECORDFILTER);
1656
1657     if (!s || !*s)
1658         return 1;
1659
1660     for (n = root->children; n; n = n->next)
1661     {
1662         if (n->type != XML_ELEMENT_NODE)
1663             continue;
1664         if (!strcmp((const char *) n->name, "metadata"))
1665         {
1666             xmlChar *type = xmlGetProp(n, (xmlChar *) "type");
1667             if (type)
1668             {
1669                 size_t len;
1670                 int substring;
1671                 const char *eq;
1672
1673                 if ((eq = strchr(s, '=')))
1674                     substring = 0;
1675                 else if ((eq = strchr(s, '~')))
1676                     substring = 1;
1677                 if (eq)
1678                     len = eq - s;
1679                 else
1680                     len = strlen(s);
1681                 if (len == strlen((const char *)type) &&
1682                     !memcmp((const char *) type, s, len))
1683                 {
1684                     xmlChar *value = xmlNodeGetContent(n);
1685                     if (value && *value)
1686                     {
1687                         if (!eq ||
1688                             (substring && strstr((const char *) value, eq+1)) ||
1689                             (!substring && !strcmp((const char *) value, eq + 1)))
1690                             match = 1;
1691                     }
1692                     xmlFree(value);
1693                 }
1694                 xmlFree(type);
1695             }
1696         }
1697     }
1698     return match;
1699 }
1700
1701 static int ingest_to_cluster(struct client *cl,
1702                              xmlDoc *xdoc,
1703                              xmlNode *root,
1704                              int record_no,
1705                              struct record_metadata_attr *mergekey);
1706
1707 static int ingest_sub_record(struct client *cl, xmlDoc *xdoc, xmlNode *root,
1708                              int record_no, NMEM nmem,
1709                              struct session_database *sdb,
1710                              struct record_metadata_attr *mergekeys)
1711 {
1712     int ret = 0;
1713     struct session *se = client_get_session(cl);
1714
1715     if (!check_record_filter(root, sdb))
1716     {
1717         session_log(se, YLOG_LOG,
1718                     "Filtered out record no %d from %s",
1719                     record_no, sdb->database->id);
1720         return 0;
1721     }
1722     session_enter(se, "ingest_sub_record");
1723     if (client_get_session(cl) == se && se->relevance)
1724         ret = ingest_to_cluster(cl, xdoc, root, record_no, mergekeys);
1725     session_leave(se, "ingest_sub_record");
1726
1727     return ret;
1728 }
1729
1730 /** \brief ingest XML record
1731     \param cl client holds the result set for record
1732     \param rec record buffer (0 terminated)
1733     \param record_no record position (1, 2, ..)
1734     \param nmem working NMEM
1735     \retval 0 OK
1736     \retval -1 failure
1737     \retval -2 Filtered
1738 */
1739 int ingest_record(struct client *cl, const char *rec,
1740                   int record_no, NMEM nmem)
1741 {
1742     struct session *se = client_get_session(cl);
1743     struct session_database *sdb = client_get_database(cl);
1744     struct conf_service *service = se->service;
1745     xmlDoc *xdoc = normalize_record(se, sdb, service, rec, nmem);
1746     int r = ingest_xml_record(cl, xdoc, record_no, nmem, 0);
1747     client_store_xdoc(cl, record_no, xdoc);
1748     return r;
1749 }
1750
1751 int ingest_xml_record(struct client *cl, xmlDoc *xdoc,
1752                       int record_no, NMEM nmem, int cached_copy)
1753 {
1754     struct session *se = client_get_session(cl);
1755     struct session_database *sdb = client_get_database(cl);
1756     struct conf_service *service = se->service;
1757     xmlNode *root;
1758     int r = 0;
1759     if (!xdoc)
1760         return -1;
1761
1762     if (global_parameters.dump_records)
1763     {
1764         session_log(se, YLOG_LOG, "Normalized record from %s",
1765                     sdb->database->id);
1766         log_xml_doc(xdoc);
1767     }
1768
1769     root = xmlDocGetRootElement(xdoc);
1770
1771     if (!strcmp((const char *) root->name, "cluster"))
1772     {
1773         int no_merge_keys = 0;
1774         int no_merge_dups = 0;
1775         xmlNode *sroot;
1776         struct record_metadata_attr *mk = 0;
1777
1778         for (sroot = root->children; sroot; sroot = sroot->next)
1779             if (sroot->type == XML_ELEMENT_NODE &&
1780                 !strcmp((const char *) sroot->name, "record"))
1781             {
1782                 struct record_metadata_attr **mkp;
1783                 const char *mergekey_norm =
1784                     get_mergekey(xdoc, sroot, cl, record_no, service, nmem,
1785                                  se->mergekey);
1786                 if (!mergekey_norm)
1787                 {
1788                     r = -1;
1789                     break;
1790                 }
1791                 for (mkp = &mk; *mkp; mkp = &(*mkp)->next)
1792                     if (!strcmp((*mkp)->value, mergekey_norm))
1793                         break;
1794                 if (!*mkp)
1795                 {
1796                     *mkp = (struct record_metadata_attr*)
1797                         nmem_malloc(nmem, sizeof(**mkp));
1798                     (*mkp)->name = 0;
1799                     (*mkp)->value = nmem_strdup(nmem, mergekey_norm);
1800                     (*mkp)->next = 0;
1801                     no_merge_keys++;
1802                 }
1803                 else
1804                     no_merge_dups++;
1805             }
1806         if (no_merge_keys > 1 || no_merge_dups > 0)
1807         {
1808             yaz_log(YLOG_LOG, "Got %d mergekeys, %d dups for position %d",
1809                     no_merge_keys, no_merge_dups, record_no);
1810         }
1811         for (sroot = root->children; !r && sroot; sroot = sroot->next)
1812             if (sroot->type == XML_ELEMENT_NODE &&
1813                 !strcmp((const char *) sroot->name, "record"))
1814             {
1815                 if (!cached_copy)
1816                     insert_settings_values(sdb, xdoc, root, service);
1817                 r = ingest_sub_record(cl, xdoc, sroot, record_no, nmem, sdb,
1818                                       mk);
1819             }
1820     }
1821     else if (!strcmp((const char *) root->name, "record"))
1822     {
1823         const char *mergekey_norm =
1824             get_mergekey(xdoc, root, cl, record_no, service, nmem,
1825                          se->mergekey);
1826         if (mergekey_norm)
1827         {
1828             struct record_metadata_attr *mk = (struct record_metadata_attr*)
1829                 nmem_malloc(nmem, sizeof(*mk));
1830             mk->name = 0;
1831             mk->value = nmem_strdup(nmem, mergekey_norm);
1832             mk->next = 0;
1833
1834             if (!cached_copy)
1835                 insert_settings_values(sdb, xdoc, root, service);
1836             r = ingest_sub_record(cl, xdoc, root, record_no, nmem, sdb, mk);
1837         }
1838     }
1839     else
1840     {
1841         session_log(se, YLOG_WARN, "Bad pz root element: %s",
1842                     (const char *) root->name);
1843         r = -1;
1844     }
1845     return r;
1846 }
1847
1848
1849 //    struct conf_metadata *ser_md = &service->metadata[md_field_id];
1850 //    struct record_metadata *rec_md = record->metadata[md_field_id];
1851 static int match_metadata_local(struct conf_service *service,
1852                                 struct conf_metadata *ser_md,
1853                                 struct record_metadata *rec_md0,
1854                                 char **values, int num_v)
1855 {
1856     int i;
1857     struct record_metadata *rec_md = rec_md0;
1858     WRBUF val_wr = 0;
1859     WRBUF text_wr = wrbuf_alloc();
1860     for (i = 0; i < num_v; )
1861     {
1862         if (rec_md)
1863         {
1864             if (ser_md->type == Metadata_type_year
1865                 || ser_md->type == Metadata_type_date)
1866             {
1867                 int y = atoi(values[i]);
1868                 if (y >= rec_md->data.number.min
1869                     && y <= rec_md->data.number.max)
1870                     break;
1871             }
1872             else
1873             {
1874                 if (!val_wr)
1875                 {
1876                     val_wr = wrbuf_alloc();
1877                     mergekey_norm_wr(service->charsets, val_wr, values[i]);
1878                 }
1879                 wrbuf_rewind(text_wr);
1880                 mergekey_norm_wr(service->charsets, text_wr,
1881                                  rec_md->data.text.disp);
1882                 if (!strcmp(wrbuf_cstr(val_wr), wrbuf_cstr(text_wr)))
1883                     break;
1884             }
1885             rec_md = rec_md->next;
1886         }
1887         else
1888         {
1889             rec_md = rec_md0;
1890             wrbuf_destroy(val_wr);
1891             val_wr = 0;
1892             i++;
1893         }
1894     }
1895     wrbuf_destroy(val_wr);
1896     wrbuf_destroy(text_wr);
1897     return i < num_v ? 1 : 0;
1898 }
1899
1900 int session_check_cluster_limit(struct session *se, struct record_cluster *rec)
1901 {
1902     int i;
1903     struct conf_service *service = se->service;
1904     int ret = 1;
1905     const char *name;
1906     const char *value;
1907     NMEM nmem_tmp = nmem_create();
1908
1909     for (i = 0; (name = facet_limits_get(se->facet_limits, i, &value)); i++)
1910     {
1911         int j;
1912         for (j = 0; j < service->num_metadata; j++)
1913         {
1914             struct conf_metadata *md = service->metadata + j;
1915             if (!strcmp(md->name, name) && md->limitcluster)
1916             {
1917                 char **values = 0;
1918                 int num = 0;
1919                 int md_field_id =
1920                     conf_service_metadata_field_id(service,
1921                                                    md->limitcluster);
1922
1923                 if (md_field_id < 0)
1924                 {
1925                     ret = 0;
1926                     break;
1927                 }
1928
1929                 nmem_strsplit_escape2(nmem_tmp, "|", value, &values,
1930                                       &num, 1, '\\', 1);
1931
1932                 if (!match_metadata_local(service,
1933                                           &service->metadata[md_field_id],
1934                                           rec->metadata[md_field_id],
1935                                           values, num))
1936                 {
1937                     ret = 0;
1938                     break;
1939                 }
1940             }
1941         }
1942     }
1943     nmem_destroy(nmem_tmp);
1944     return ret;
1945 }
1946
1947 // Skip record on non-zero
1948 static int check_limit_local(struct client *cl,
1949                              struct record *record,
1950                              int record_no)
1951 {
1952     int skip_record = 0;
1953     struct session *se = client_get_session(cl);
1954     struct conf_service *service = se->service;
1955     NMEM nmem_tmp = nmem_create();
1956     struct session_database *sdb = client_get_database(cl);
1957     int l = 0;
1958     while (!skip_record)
1959     {
1960         int md_field_id;
1961         char **values = 0;
1962         int num_v = 0;
1963         const char *name =
1964             client_get_facet_limit_local(cl, sdb, &l, nmem_tmp,
1965                                          &num_v, &values);
1966         if (!name)
1967             break;
1968
1969         if (!strcmp(name, "*"))
1970         {
1971             for (md_field_id = 0; md_field_id < service->num_metadata;
1972                  md_field_id++)
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                     break;
1980             }
1981             if (md_field_id == service->num_metadata)
1982                 skip_record = 1;
1983         }
1984         else
1985         {
1986             md_field_id = conf_service_metadata_field_id(service, name);
1987             if (md_field_id < 0)
1988             {
1989                 skip_record = 1;
1990                 break;
1991             }
1992             if (!match_metadata_local(
1993                     service,
1994                     &service->metadata[md_field_id],
1995                     record->metadata[md_field_id],
1996                     values, num_v))
1997             {
1998                 skip_record = 1;
1999             }
2000         }
2001     }
2002     nmem_destroy(nmem_tmp);
2003     return skip_record;
2004 }
2005
2006 static int ingest_to_cluster(struct client *cl,
2007                              xmlDoc *xdoc,
2008                              xmlNode *root,
2009                              int record_no,
2010                              struct record_metadata_attr *merge_keys)
2011 {
2012     xmlNode *n;
2013     xmlChar *type = 0;
2014     xmlChar *value = 0;
2015     struct session *se = client_get_session(cl);
2016     struct conf_service *service = se->service;
2017     int term_factor = 1;
2018     struct record_cluster *cluster;
2019     struct record_metadata **metadata0;
2020     struct session_database *sdb = client_get_database(cl);
2021     NMEM ingest_nmem = 0;
2022     char **rank_values = 0;
2023     int rank_num = 0;
2024     struct record *record = record_create(se->nmem,
2025                                           service->num_metadata,
2026                                           service->num_sortkeys, cl,
2027                                           record_no);
2028
2029     for (n = root->children; n; n = n->next)
2030     {
2031         if (type)
2032             xmlFree(type);
2033         if (value)
2034             xmlFree(value);
2035         type = value = 0;
2036
2037         if (n->type != XML_ELEMENT_NODE)
2038             continue;
2039         if (!strcmp((const char *) n->name, "metadata"))
2040         {
2041             struct conf_metadata *ser_md = 0;
2042             struct record_metadata **wheretoput = 0;
2043             struct record_metadata *rec_md = 0;
2044             int md_field_id = -1;
2045
2046             type = xmlGetProp(n, (xmlChar *) "type");
2047             value = xmlNodeListGetString(xdoc, n->children, 1);
2048             if (!type)
2049                 continue;
2050             if (!value || !*value)
2051             {
2052                 xmlChar *empty = xmlGetProp(n, (xmlChar *) "empty");
2053                 if (!empty)
2054                     continue;
2055                 if (value)
2056                     xmlFree(value);
2057                 value = empty;
2058             }
2059             md_field_id
2060                 = conf_service_metadata_field_id(service, (const char *) type);
2061             if (md_field_id < 0)
2062             {
2063                 if (se->number_of_warnings_unknown_metadata == 0)
2064                 {
2065                     session_log(se, YLOG_WARN,
2066                             "Ignoring unknown metadata element: %s", type);
2067                 }
2068                 se->number_of_warnings_unknown_metadata++;
2069                 continue;
2070             }
2071
2072             ser_md = &service->metadata[md_field_id];
2073
2074             // non-merged metadata
2075             rec_md = record_metadata_init(se->nmem, (const char *) value,
2076                                           ser_md->type, n->properties);
2077             if (!rec_md)
2078             {
2079                 session_log(se, YLOG_WARN, "bad metadata data '%s' "
2080                             "for element '%s'", value, type);
2081                 continue;
2082             }
2083
2084             if (ser_md->type == Metadata_type_generic)
2085             {
2086                 WRBUF w = wrbuf_alloc();
2087                 if (relevance_snippet(se->relevance,
2088                                       (char*) value, ser_md->name, w))
2089                     rec_md->data.text.snippet = nmem_strdup(se->nmem,
2090                                                             wrbuf_cstr(w));
2091                 wrbuf_destroy(w);
2092             }
2093
2094
2095             wheretoput = &record->metadata[md_field_id];
2096             while (*wheretoput)
2097                 wheretoput = &(*wheretoput)->next;
2098             *wheretoput = rec_md;
2099         }
2100     }
2101
2102     if (check_limit_local(cl, record, record_no))
2103     {
2104         if (type)
2105             xmlFree(type);
2106         if (value)
2107             xmlFree(value);
2108         return -2;
2109     }
2110     cluster = reclist_insert(se->reclist, se->relevance, service, record,
2111                              merge_keys, &se->total_merged);
2112     if (!cluster)
2113     {
2114         if (type)
2115             xmlFree(type);
2116         if (value)
2117             xmlFree(value);
2118         return 0; // complete match with existing record
2119     }
2120
2121     {
2122         const char *use_term_factor_str =
2123             session_setting_oneval(sdb, PZ_TERMLIST_TERM_FACTOR);
2124         if (use_term_factor_str && use_term_factor_str[0] == '1')
2125         {
2126             int maxrecs = client_get_maxrecs(cl);
2127             int hits = (int) client_get_hits(cl);
2128             term_factor = MAX(hits, maxrecs) /  MAX(1, maxrecs);
2129             assert(term_factor >= 1);
2130             session_log(se, YLOG_DEBUG, "Using term factor: %d (%d / %d)",
2131                         term_factor, MAX(hits, maxrecs), MAX(1, maxrecs));
2132         }
2133     }
2134
2135     if (global_parameters.dump_records)
2136         session_log(se, YLOG_LOG, "Cluster id %s from %s (#%d)", cluster->recid,
2137                     sdb->database->id, record_no);
2138
2139     // original metadata, to check if first existence of a field
2140     metadata0 = xmalloc(sizeof(*metadata0) * service->num_metadata);
2141     memcpy(metadata0, cluster->metadata,
2142            sizeof(*metadata0) * service->num_metadata);
2143
2144     ingest_nmem = nmem_create();
2145     if (se->rank)
2146     {
2147         yaz_log(YLOG_LOG, "local in sort : %s", se->rank);
2148         nmem_strsplit_escape2(ingest_nmem, ",", se->rank, &rank_values,
2149                               &rank_num, 1, '\\', 1);
2150     }
2151
2152     // now parsing XML record and adding data to cluster or record metadata
2153     for (n = root->children; n; n = n->next)
2154     {
2155         pp2_charset_token_t prt;
2156         if (type)
2157             xmlFree(type);
2158         if (value)
2159             xmlFree(value);
2160         type = value = 0;
2161
2162         if (n->type != XML_ELEMENT_NODE)
2163             continue;
2164         if (!strcmp((const char *) n->name, "metadata"))
2165         {
2166             struct conf_metadata *ser_md = 0;
2167             struct conf_sortkey *ser_sk = 0;
2168             struct record_metadata **wheretoput = 0;
2169             struct record_metadata *rec_md = 0;
2170             int md_field_id = -1;
2171             int sk_field_id = -1;
2172             const char *rank = 0;
2173             xmlChar *xml_rank = 0;
2174
2175             type = xmlGetProp(n, (xmlChar *) "type");
2176             value = xmlNodeListGetString(xdoc, n->children, 1);
2177
2178             if (!type || !value || !*value)
2179                 continue;
2180
2181             md_field_id
2182                 = conf_service_metadata_field_id(service, (const char *) type);
2183             if (md_field_id < 0)
2184                 continue;
2185
2186             ser_md = &service->metadata[md_field_id];
2187
2188             if (ser_md->sortkey_offset >= 0)
2189             {
2190                 sk_field_id = ser_md->sortkey_offset;
2191                 ser_sk = &service->sortkeys[sk_field_id];
2192             }
2193
2194             // merged metadata
2195             rec_md = record_metadata_init(se->nmem, (const char *) value,
2196                                           ser_md->type, 0);
2197
2198             // see if the field was not in cluster already (from beginning)
2199
2200             if (!rec_md)
2201                 continue;
2202
2203             if (rank_num)
2204             {
2205                 int i;
2206                 for (i = 0; i < rank_num; i++)
2207                 {
2208                     const char *val = rank_values[i];
2209                     const char *cp = strchr(val, '=');
2210                     if (!cp)
2211                         continue;
2212                     if ((cp - val) == strlen((const char *) type)
2213                         && !memcmp(val, type, cp - val))
2214                     {
2215                         rank = cp + 1;
2216                         break;
2217                     }
2218                 }
2219             }
2220             else
2221             {
2222                 xml_rank = xmlGetProp(n, (xmlChar *) "rank");
2223                 rank = xml_rank ? (const char *) xml_rank : ser_md->rank;
2224             }
2225
2226             wheretoput = &cluster->metadata[md_field_id];
2227
2228             if (ser_md->merge == Metadata_merge_first)
2229             {
2230                 if (!metadata0[md_field_id])
2231                 {
2232                     while (*wheretoput)
2233                         wheretoput = &(*wheretoput)->next;
2234                     *wheretoput = rec_md;
2235                 }
2236             }
2237             else if (ser_md->merge == Metadata_merge_unique)
2238             {
2239                 while (*wheretoput)
2240                 {
2241                     if (!strcmp((const char *) (*wheretoput)->data.text.disp,
2242                                 rec_md->data.text.disp))
2243                         break;
2244                     wheretoput = &(*wheretoput)->next;
2245                 }
2246                 if (!*wheretoput)
2247                     *wheretoput = rec_md;
2248             }
2249             else if (ser_md->merge == Metadata_merge_longest)
2250             {
2251                 if (!*wheretoput
2252                     || strlen(rec_md->data.text.disp)
2253                     > strlen((*wheretoput)->data.text.disp))
2254                 {
2255                     *wheretoput = rec_md;
2256                     if (ser_sk)
2257                     {
2258                         const char *sort_str = 0;
2259                         int skip_article =
2260                             ser_sk->type == Metadata_type_skiparticle;
2261
2262                         if (!cluster->sortkeys[sk_field_id])
2263                             cluster->sortkeys[sk_field_id] =
2264                                 nmem_malloc(se->nmem,
2265                                             sizeof(union data_types));
2266
2267                         prt =
2268                             pp2_charset_token_create(service->charsets, "sort");
2269
2270                         pp2_charset_token_first(prt, rec_md->data.text.disp,
2271                                                 skip_article);
2272
2273                         pp2_charset_token_next(prt);
2274
2275                         sort_str = pp2_get_sort(prt);
2276
2277                         cluster->sortkeys[sk_field_id]->text.disp =
2278                             rec_md->data.text.disp;
2279                         if (!sort_str)
2280                         {
2281                             sort_str = rec_md->data.text.disp;
2282                             session_log(se, YLOG_WARN,
2283                                     "Could not make sortkey. Bug #1858");
2284                         }
2285                         cluster->sortkeys[sk_field_id]->text.sort =
2286                             nmem_strdup(se->nmem, sort_str);
2287                         pp2_charset_token_destroy(prt);
2288                     }
2289                 }
2290             }
2291             else if (ser_md->merge == Metadata_merge_all)
2292             {
2293                 while (*wheretoput)
2294                     wheretoput = &(*wheretoput)->next;
2295                 *wheretoput = rec_md;
2296             }
2297             else if (ser_md->merge == Metadata_merge_range)
2298             {
2299                 if (!*wheretoput)
2300                 {
2301                     *wheretoput = rec_md;
2302                     if (ser_sk)
2303                         cluster->sortkeys[sk_field_id]
2304                             = &rec_md->data;
2305                 }
2306                 else
2307                 {
2308                     int this_min = rec_md->data.number.min;
2309                     int this_max = rec_md->data.number.max;
2310                     if (this_min < (*wheretoput)->data.number.min)
2311                         (*wheretoput)->data.number.min = this_min;
2312                     if (this_max > (*wheretoput)->data.number.max)
2313                         (*wheretoput)->data.number.max = this_max;
2314                 }
2315             }
2316
2317             // ranking of _all_ fields enabled ...
2318             if (rank)
2319             {
2320                 relevance_countwords(se->relevance, cluster,
2321                                      (char *) value, rank, ser_md->name);
2322             }
2323             // construct facets ... unless the client already has reported them
2324             if (ser_md->termlist && !client_has_facet(cl, (char *) type))
2325             {
2326                 if (ser_md->type == Metadata_type_year)
2327                 {
2328                     char year[64];
2329                     sprintf(year, "%d", rec_md->data.number.max);
2330
2331                     add_facet(se, (char *) type, year, term_factor);
2332                     if (rec_md->data.number.max != rec_md->data.number.min)
2333                     {
2334                         sprintf(year, "%d", rec_md->data.number.min);
2335                         add_facet(se, (char *) type, year, term_factor);
2336                     }
2337                 }
2338                 else
2339                     add_facet(se, (char *) type, (char *) value, term_factor);
2340             }
2341
2342             // cleaning up
2343             if (xml_rank)
2344                 xmlFree(xml_rank);
2345             xmlFree(type);
2346             xmlFree(value);
2347             type = value = 0;
2348         }
2349         else
2350         {
2351             if (se->number_of_warnings_unknown_elements == 0)
2352                 session_log(se, YLOG_WARN,
2353                         "Unexpected element in internal record: %s", n->name);
2354             se->number_of_warnings_unknown_elements++;
2355         }
2356     }
2357     if (type)
2358         xmlFree(type);
2359     if (value)
2360         xmlFree(value);
2361
2362     nmem_destroy(ingest_nmem);
2363     xfree(metadata0);
2364     relevance_donerecord(se->relevance, cluster);
2365     se->total_records++;
2366
2367     return 0;
2368 }
2369
2370 void session_log(struct session *s, int level, const char *fmt, ...)
2371 {
2372     char buf[1024];
2373     va_list ap;
2374     va_start(ap, fmt);
2375
2376     yaz_vsnprintf(buf, sizeof(buf)-30, fmt, ap);
2377     yaz_log(level, "Session %u: %s", s ? s->session_id : 0, buf);
2378
2379     va_end(ap);
2380 }
2381
2382 /*
2383  * Local variables:
2384  * c-basic-offset: 4
2385  * c-file-style: "Stroustrup"
2386  * indent-tabs-mode: nil
2387  * End:
2388  * vim: shiftwidth=4 tabstop=8 expandtab
2389  */
2390