src/session.c

   1 /* This file is part of Pazpar2.
   2    Copyright (C) Index Data
   3
   4 Pazpar2 is free software; you can redistribute it and/or modify it under
   5 the terms of the GNU General Public License as published by the Free
   6 Software Foundation; either version 2, or (at your option) any later
   7 version.
   8
   9 Pazpar2 is distributed in the hope that it will be useful, but WITHOUT ANY
  10 WARRANTY; without even the implied warranty of MERCHANTABILITY or
  11 FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License
  12 for more details.
  13
  14 You should have received a copy of the GNU General Public License
  15 along with this program; if not, write to the Free Software
  16 Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
  17
  18 */
  19
  20 /** \file session.c
  21     \brief high-level logic; mostly user sessions and settings
  22 */
  23
  24 #if HAVE_CONFIG_H
  25 #include <config.h>
  26 #endif
  27
  28 #include <time.h>
  29 #include <stdlib.h>
  30 #include <stdio.h>
  31 #include <string.h>
  32 #if HAVE_SYS_TIME_H
  33 #include <sys/time.h>
  34 #endif
  35 #if HAVE_UNISTD_H
  36 #include <unistd.h>
  37 #endif
  38 #ifdef WIN32
  39 #include <windows.h>
  40 #endif
  41 #include <signal.h>
  42 #include <ctype.h>
  43 #include <assert.h>
  44 #include <math.h>
  45
  46 #include <yaz/marcdisp.h>
  47 #include <yaz/comstack.h>
  48 #include <yaz/tcpip.h>
  49 #include <yaz/proto.h>
  50 #include <yaz/readconf.h>
  51 #include <yaz/pquery.h>
  52 #include <yaz/otherinfo.h>
  53 #include <yaz/yaz-util.h>
  54 #include <yaz/nmem.h>
  55 #include <yaz/query-charset.h>
  56 #include <yaz/querytowrbuf.h>
  57 #include <yaz/oid_db.h>
  58 #include <yaz/snprintf.h>
  59 #include <yaz/xml_get.h>
  60
  61 #define USE_TIMING 0
  62 #if USE_TIMING
  63 #include <yaz/timing.h>
  64 #endif
  65
  66 #include "ppmutex.h"
  67 #include "parameters.h"
  68 #include "session.h"
  69 #include "eventl.h"
  70 #include "http.h"
  71 #include "termlists.h"
  72 #include "reclists.h"
  73 #include "relevance.h"
  74 #include "database.h"
  75 #include "client.h"
  76 #include "settings.h"
  77 #include "normalize7bit.h"
  78
  79 #include <libxml/tree.h>
  80
  81 #define MAX_CHUNK 15
  82
  83 #define MAX(a,b) ((a)>(b)?(a):(b))
  84
  85 // Note: Some things in this structure will eventually move to configuration
  86 struct parameters global_parameters =
  87 {
  88     0,   // dump_records
  89     0,   // debug_mode
  90     0,   // predictable sessions
  91 };
  92
  93 struct client_list {
  94     struct client *client;
  95     struct client_list *next;
  96 };
  97
  98 /* session counting (1) , disable client counting (0) */
  99 static YAZ_MUTEX g_session_mutex = 0;
 100 static int no_sessions = 0;
 101
 102 static int session_use(int delta)
 103 {
 104     int sessions;
 105     if (!g_session_mutex)
 106         yaz_mutex_create(&g_session_mutex);
 107     yaz_mutex_enter(g_session_mutex);
 108     no_sessions += delta;
 109     sessions = no_sessions;
 110     yaz_mutex_leave(g_session_mutex);
 111     yaz_log(YLOG_DEBUG, "%s sessions=%d", delta == 0 ? "" :
 112             (delta > 0 ? "INC" : "DEC"), no_sessions);
 113     return sessions;
 114 }
 115
 116 int sessions_count(void)
 117 {
 118     return session_use(0);
 119 }
 120
 121 static void log_xml_doc(xmlDoc *doc)
 122 {
 123     FILE *lf = yaz_log_file();
 124     xmlChar *result = 0;
 125     int len = 0;
 126 #if LIBXML_VERSION >= 20600
 127     xmlDocDumpFormatMemory(doc, &result, &len, 1);
 128 #else
 129     xmlDocDumpMemory(doc, &result, &len);
 130 #endif
 131     if (lf && len)
 132     {
 133         (void) fwrite(result, 1, len, lf);
 134         fprintf(lf, "\n");
 135     }
 136     xmlFree(result);
 137 }
 138
 139 static void session_enter(struct session *s, const char *caller)
 140 {
 141     if (caller)
 142         session_log(s, YLOG_DEBUG, "Session lock by %s", caller);
 143     yaz_mutex_enter(s->session_mutex);
 144 }
 145
 146 static void session_leave(struct session *s, const char *caller)
 147 {
 148     yaz_mutex_leave(s->session_mutex);
 149     if (caller)
 150         session_log(s, YLOG_DEBUG, "Session unlock by %s", caller);
 151 }
 152
 153 static int run_icu(struct session *s, const char *icu_chain_id,
 154                    const char *value,
 155                    WRBUF norm_wr, WRBUF disp_wr)
 156 {
 157     const char *facet_component;
 158     struct conf_service *service = s->service;
 159     pp2_charset_token_t prt =
 160         pp2_charset_token_create(service->charsets, icu_chain_id);
 161     if (!prt)
 162     {
 163         session_log(s, YLOG_FATAL,
 164                     "Unknown ICU chain '%s'", icu_chain_id);
 165         return 0;
 166     }
 167     pp2_charset_token_first(prt, value, 0);
 168     while ((facet_component = pp2_charset_token_next(prt)))
 169     {
 170         const char *display_component;
 171         if (*facet_component)
 172         {
 173             if (wrbuf_len(norm_wr))
 174                 wrbuf_puts(norm_wr, " ");
 175             wrbuf_puts(norm_wr, facet_component);
 176         }
 177         display_component = pp2_get_display(prt);
 178         if (display_component)
 179         {
 180             if (wrbuf_len(disp_wr))
 181                 wrbuf_puts(disp_wr, " ");
 182             wrbuf_puts(disp_wr, display_component);
 183         }
 184     }
 185     pp2_charset_token_destroy(prt);
 186     return 1;
 187 }
 188
 189 static void session_normalize_facet(struct session *s,
 190                                     const char *type, const char *value,
 191                                     WRBUF display_wrbuf, WRBUF facet_wrbuf)
 192 {
 193     struct conf_service *service = s->service;
 194     int i;
 195     const char *icu_chain_id = 0;
 196
 197     for (i = 0; i < service->num_metadata; i++)
 198         if (!strcmp((service->metadata + i)->name, type))
 199             icu_chain_id = (service->metadata + i)->facetrule;
 200     if (!icu_chain_id)
 201         icu_chain_id = "facet";
 202
 203     run_icu(s, icu_chain_id, value, facet_wrbuf, display_wrbuf);
 204 }
 205
 206 struct facet_id {
 207     char *client_id;
 208     char *type;
 209     char *id;
 210     char *term;
 211     struct facet_id *next;
 212 };
 213
 214 static void session_add_id_facet(struct session *s, struct client *cl,
 215                                  const char *type,
 216                                  const char *id,
 217                                  size_t id_len,
 218                                  const char *term)
 219 {
 220     struct facet_id *t = nmem_malloc(s->session_nmem, sizeof(*t));
 221
 222     t->client_id = nmem_strdup(s->session_nmem, client_get_id(cl));
 223     t->type = nmem_strdup(s->session_nmem, type);
 224     t->id = nmem_strdupn(s->session_nmem, id, id_len);
 225     t->term = nmem_strdup(s->session_nmem, term);
 226     t->next = s->facet_id_list;
 227     s->facet_id_list = t;
 228 }
 229
 230
 231 // Look up a facet term, and return matching id
 232 // If facet type not found, returns 0
 233 // If facet type found, but no matching term, returns ""
 234 const char *session_lookup_id_facet(struct session *s, struct client *cl,
 235                                     const char *type,
 236                                     const char *term)
 237 {
 238     char *retval = 0;
 239     struct facet_id *t = s->facet_id_list;
 240     for (; t; t = t->next)
 241     {
 242         if (!strcmp(client_get_id(cl), t->client_id) &&  !strcmp(t->type, type) )
 243         {
 244             retval = "";
 245             if ( !strcmp(t->term, term))
 246             {
 247                 return t->id;
 248             }
 249         }
 250     }
 251     return retval;
 252 }
 253
 254 void add_facet(struct session *s, const char *type, const char *value, int count, struct client *cl)
 255 {
 256     WRBUF facet_wrbuf = wrbuf_alloc();
 257     WRBUF display_wrbuf = wrbuf_alloc();
 258     const char *id = 0;
 259     size_t id_len = 0;
 260
 261     /* inspect pz:facetmap:split:name ?? */
 262     if (!strncmp(type, "split:", 6))
 263     {
 264         const char *cp = strchr(value, ':');
 265         if (cp)
 266         {
 267             id = value;
 268             id_len = cp - value;
 269             value = cp + 1;
 270         }
 271         type += 6;
 272     }
 273
 274     session_normalize_facet(s, type, value, display_wrbuf, facet_wrbuf);
 275     if (wrbuf_len(facet_wrbuf))
 276     {
 277         struct named_termlist **tp = &s->termlists;
 278         for (; (*tp); tp = &(*tp)->next)
 279             if (!strcmp((*tp)->name, type))
 280                 break;
 281         if (!*tp)
 282         {
 283             *tp = nmem_malloc(s->nmem, sizeof(**tp));
 284             (*tp)->name = nmem_strdup(s->nmem, type);
 285             (*tp)->termlist = termlist_create(s->nmem);
 286             (*tp)->next = 0;
 287         }
 288         termlist_insert((*tp)->termlist, wrbuf_cstr(display_wrbuf),
 289                         wrbuf_cstr(facet_wrbuf), id, id_len, count);
 290         if (id)
 291             session_add_id_facet(s, cl, type, id, id_len,
 292                                  wrbuf_cstr(display_wrbuf));
 293     }
 294     wrbuf_destroy(facet_wrbuf);
 295     wrbuf_destroy(display_wrbuf);
 296 }
 297
 298 static xmlDoc *record_to_xml(struct session *se,
 299                              struct session_database *sdb, const char *rec)
 300 {
 301     struct database *db = sdb->database;
 302     xmlDoc *rdoc = 0;
 303
 304     rdoc = xmlParseMemory(rec, strlen(rec));
 305
 306     if (!rdoc)
 307     {
 308         session_log(se, YLOG_WARN, "Non-wellformed XML");
 309         return 0;
 310     }
 311
 312     if (global_parameters.dump_records)
 313     {
 314         session_log(se, YLOG_LOG, "Un-normalized record from %s", db->id);
 315         log_xml_doc(rdoc);
 316     }
 317
 318     return rdoc;
 319 }
 320
 321 #define MAX_XSLT_ARGS 16
 322
 323 // Add static values from session database settings if applicable
 324 static void insert_settings_parameters(struct session_database *sdb,
 325                                        struct conf_service *service,
 326                                        char **parms,
 327                                        NMEM nmem)
 328 {
 329     int i;
 330     int nparms = 0;
 331     int offset = 0;
 332
 333     for (i = 0; i < service->num_metadata; i++)
 334     {
 335         struct conf_metadata *md = &service->metadata[i];
 336         int setting;
 337
 338         if (md->setting == Metadata_setting_parameter &&
 339             (setting = settings_lookup_offset(service, md->name)) >= 0)
 340         {
 341             const char *val = session_setting_oneval(sdb, setting);
 342             if (val && nparms < MAX_XSLT_ARGS)
 343             {
 344                 char *buf;
 345                 int len = strlen(val);
 346                 buf = nmem_malloc(nmem, len + 3);
 347                 buf[0] = '\'';
 348                 strcpy(buf + 1, val);
 349                 buf[len+1] = '\'';
 350                 buf[len+2] = '\0';
 351                 parms[offset++] = md->name;
 352                 parms[offset++] = buf;
 353                 nparms++;
 354             }
 355         }
 356     }
 357     parms[offset] = 0;
 358 }
 359
 360 // Add static values from session database settings if applicable
 361 static void insert_settings_values(struct session_database *sdb, xmlDoc *doc,
 362                                    xmlNode *root,
 363                                    struct conf_service *service)
 364 {
 365     int i;
 366
 367     for (i = 0; i < service->num_metadata; i++)
 368     {
 369         struct conf_metadata *md = &service->metadata[i];
 370         int offset;
 371
 372         if (md->setting == Metadata_setting_postproc &&
 373             (offset = settings_lookup_offset(service, md->name)) >= 0)
 374         {
 375             const char *val = session_setting_oneval(sdb, offset);
 376             if (val)
 377             {
 378                 xmlNode *n = xmlNewTextChild(root, 0, (xmlChar *) "metadata",
 379                                              (xmlChar *) val);
 380                 xmlSetProp(n, (xmlChar *) "type", (xmlChar *) md->name);
 381             }
 382         }
 383     }
 384 }
 385
 386 static xmlDoc *normalize_record(struct session *se,
 387                                 struct session_database *sdb,
 388                                 struct conf_service *service,
 389                                 const char *rec, NMEM nmem)
 390 {
 391     xmlDoc *rdoc = record_to_xml(se, sdb, rec);
 392
 393     if (rdoc)
 394     {
 395         char *parms[MAX_XSLT_ARGS*2+1];
 396
 397         insert_settings_parameters(sdb, service, parms, nmem);
 398
 399         if (normalize_record_transform(sdb->map, &rdoc, (const char **)parms))
 400         {
 401             session_log(se, YLOG_WARN, "Normalize failed");
 402         }
 403     }
 404     return rdoc;
 405 }
 406
 407 void session_settings_dump(struct session *se,
 408                            struct session_database *db,
 409                            WRBUF w)
 410 {
 411     if (db->settings)
 412     {
 413         int i, num = db->num_settings;
 414         for (i = 0; i < num; i++)
 415         {
 416             struct setting *s = db->settings[i];
 417             for (;s ; s = s->next)
 418             {
 419                 wrbuf_puts(w, "<set name=\"");
 420                 wrbuf_xmlputs(w, s->name);
 421                 wrbuf_puts(w, "\" value=\"");
 422                 wrbuf_xmlputs(w, s->value);
 423                 wrbuf_puts(w, "\"/>");
 424             }
 425             if (db->settings[i])
 426                 wrbuf_puts(w, "\n");
 427         }
 428     }
 429 }
 430
 431 // Retrieve first defined value for 'name' for given database.
 432 // Will be extended to take into account user associated with session
 433 const char *session_setting_oneval(struct session_database *db, int offset)
 434 {
 435     if (offset >= db->num_settings || !db->settings[offset])
 436         return "";
 437     return db->settings[offset]->value;
 438 }
 439
 440 // Prepare XSLT stylesheets for record normalization
 441 // Structures are allocated on the session_wide nmem to avoid having
 442 // to recompute this for every search. This would lead
 443 // to leaking if a single session was to repeatedly change the PZ_XSLT
 444 // setting. However, this is not a realistic use scenario.
 445 static int prepare_map(struct session *se, struct session_database *sdb)
 446 {
 447     if (sdb->settings && !sdb->map)
 448     {
 449         const char *s;
 450
 451         if (sdb->settings[PZ_XSLT] &&
 452             (s = session_setting_oneval(sdb, PZ_XSLT)))
 453         {
 454             char auto_stylesheet[256];
 455
 456             if (!strcmp(s, "auto"))
 457             {
 458                 const char *request_syntax = session_setting_oneval(
 459                     sdb, PZ_REQUESTSYNTAX);
 460                 if (request_syntax)
 461                 {
 462                     char *cp;
 463                     yaz_snprintf(auto_stylesheet, sizeof(auto_stylesheet),
 464                                  "%s.xsl", request_syntax);
 465                     for (cp = auto_stylesheet; *cp; cp++)
 466                     {
 467                         /* deliberately only consider ASCII */
 468                         if (*cp > 32 && *cp < 127)
 469                             *cp = tolower(*cp);
 470                     }
 471                     s = auto_stylesheet;
 472                 }
 473                 else
 474                 {
 475                     session_log(se, YLOG_WARN,
 476                                 "No pz:requestsyntax for auto stylesheet");
 477                 }
 478             }
 479             sdb->map = normalize_cache_get(se->normalize_cache,
 480                                            se->service, s);
 481             if (!sdb->map)
 482                 return -1;
 483         }
 484     }
 485     return 0;
 486 }
 487
 488 // called if watch should be removed because http_channel is to be destroyed
 489 static void session_watch_cancel(void *data, struct http_channel *c,
 490                                  void *data2)
 491 {
 492     struct session_watchentry *ent = data;
 493
 494     ent->fun = 0;
 495     ent->data = 0;
 496     ent->obs = 0;
 497 }
 498
 499 // set watch. Returns 0=OK, -1 if watch is already set
 500 int session_set_watch(struct session *s, int what,
 501                       session_watchfun fun, void *data,
 502                       struct http_channel *chan)
 503 {
 504     int ret;
 505     session_enter(s, "session_set_watch");
 506     if (s->watchlist[what].fun)
 507         ret = -1;
 508     else
 509     {
 510
 511         s->watchlist[what].fun = fun;
 512         s->watchlist[what].data = data;
 513         s->watchlist[what].obs = http_add_observer(chan, &s->watchlist[what],
 514                                                    session_watch_cancel);
 515         ret = 0;
 516     }
 517     session_leave(s, "session_set_watch");
 518     return ret;
 519 }
 520
 521 void session_alert_watch(struct session *s, int what)
 522 {
 523     assert(s);
 524     session_enter(s, "session_alert_watch");
 525     if (s->watchlist[what].fun)
 526     {
 527         /* our watch is no longer associated with http_channel */
 528         void *data;
 529         session_watchfun fun;
 530
 531         http_remove_observer(s->watchlist[what].obs);
 532         fun  = s->watchlist[what].fun;
 533         data = s->watchlist[what].data;
 534
 535         /* reset watch before fun is invoked - in case fun wants to set
 536            it again */
 537         s->watchlist[what].fun = 0;
 538         s->watchlist[what].data = 0;
 539         s->watchlist[what].obs = 0;
 540
 541         session_leave(s, "session_alert_watch");
 542         session_log(s, YLOG_DEBUG,
 543                     "Alert Watch: %d calling function: %p", what, fun);
 544         fun(data);
 545     }
 546     else
 547         session_leave(s,"session_alert_watch");
 548 }
 549
 550 //callback for grep_databases
 551 static void select_targets_callback(struct session *se,
 552                                     struct session_database *db)
 553 {
 554     struct client *cl;
 555     struct client_list *l;
 556
 557     for (l = se->clients_cached; l; l = l->next)
 558         if (client_get_database(l->client) == db)
 559             break;
 560
 561     if (l)
 562         cl = l->client;
 563     else
 564     {
 565         cl = client_create(db->database->id);
 566         client_set_database(cl, db);
 567
 568         l = xmalloc(sizeof(*l));
 569         l->client = cl;
 570         l->next = se->clients_cached;
 571         se->clients_cached = l;
 572     }
 573     client_set_session(cl, se);
 574
 575     l = xmalloc(sizeof(*l));
 576     l->client = cl;
 577     l->next = se->clients_active;
 578     se->clients_active = l;
 579 }
 580
 581 static void session_reset_active_clients(struct session *se,
 582                                          struct client_list *new_list)
 583 {
 584     struct client_list *l;
 585
 586     session_enter(se, "session_reset_active_clients");
 587     l = se->clients_active;
 588     se->clients_active = new_list;
 589     session_leave(se, "session_reset_active_clients");
 590
 591     while (l)
 592     {
 593         struct client_list *l_next = l->next;
 594
 595         client_lock(l->client);
 596         client_set_session(l->client, 0); /* mark client inactive */
 597         client_unlock(l->client);
 598
 599         xfree(l);
 600         l = l_next;
 601     }
 602 }
 603
 604 static void session_remove_cached_clients(struct session *se)
 605 {
 606     struct client_list *l;
 607
 608     session_reset_active_clients(se, 0);
 609
 610     session_enter(se, "session_remove_cached_clients");
 611     l = se->clients_cached;
 612     se->clients_cached = 0;
 613     session_leave(se, "session_remove_cached_clients");
 614
 615     while (l)
 616     {
 617         struct client_list *l_next = l->next;
 618         client_lock(l->client);
 619         client_set_session(l->client, 0);
 620         client_set_database(l->client, 0);
 621         client_mark_dead(l->client);
 622         client_unlock(l->client);
 623         client_destroy(l->client);
 624         xfree(l);
 625         l = l_next;
 626     }
 627 }
 628
 629 // Associates a set of clients with a session;
 630 // Note: Session-databases represent databases with per-session
 631 // setting overrides
 632 static int select_targets(struct session *se, const char *filter)
 633 {
 634     return session_grep_databases(se, filter, select_targets_callback);
 635 }
 636
 637 int session_active_clients(struct session *s)
 638 {
 639     struct client_list *l;
 640     int res = 0;
 641
 642     for (l = s->clients_active; l; l = l->next)
 643         if (client_is_active(l->client))
 644             res++;
 645
 646     return res;
 647 }
 648
 649 int session_is_preferred_clients_ready(struct session *s)
 650 {
 651     struct client_list *l;
 652     int res = 0;
 653
 654     for (l = s->clients_active; l; l = l->next)
 655         if (client_is_active_preferred(l->client))
 656             res++;
 657     session_log(s, YLOG_DEBUG, "Has %d active preferred clients.", res);
 658     return res == 0;
 659 }
 660
 661 static void session_clear_set(struct session *se, struct reclist_sortparms *sp)
 662 {
 663     reclist_destroy(se->reclist);
 664     if (nmem_total(se->nmem))
 665         session_log(se, YLOG_DEBUG, "NMEN operation usage %zd",
 666                     nmem_total(se->nmem));
 667     nmem_reset(se->nmem);
 668     se->total_records = se->total_merged = 0;
 669     se->termlists = 0;
 670     relevance_clear(se->relevance);
 671
 672     /* reset list of sorted results and clear to relevance search */
 673     se->sorted_results = nmem_malloc(se->nmem, sizeof(*se->sorted_results));
 674     se->sorted_results->name = nmem_strdup(se->nmem, sp->name);
 675     se->sorted_results->increasing = sp->increasing;
 676     se->sorted_results->type = sp->type;
 677     se->sorted_results->next = 0;
 678
 679     session_log(se, YLOG_DEBUG, "clear_set session_sort: field=%s increasing=%d type=%d configured",
 680             sp->name, sp->increasing, sp->type);
 681
 682     se->reclist = reclist_create(se->nmem);
 683 }
 684
 685 void session_sort(struct session *se, struct reclist_sortparms *sp,
 686                   const char *mergekey, const char *rank)
 687 {
 688     struct client_list *l;
 689     const char *field = sp->name;
 690     int increasing = sp->increasing;
 691     int type  = sp->type;
 692     int clients_research = 0;
 693
 694     session_enter(se, "session_sort");
 695     session_log(se, YLOG_DEBUG, "session_sort field=%s increasing=%d type=%d",
 696                 field, increasing, type);
 697
 698     if (rank && (!se->rank || strcmp(se->rank, rank)))
 699     {
 700         /* new rank must research/reingest anyway */
 701         assert(rank);
 702         xfree(se->rank);
 703         se->rank = *rank ? xstrdup(rank) : 0;
 704         clients_research = 1;
 705         session_log(se, YLOG_DEBUG, "session_sort: new rank = %s",
 706                     rank);
 707     }
 708     if (mergekey && (!se->mergekey || strcmp(se->mergekey, mergekey)))
 709     {
 710         /* new mergekey must research/reingest anyway */
 711         assert(mergekey);
 712         xfree(se->mergekey);
 713         se->mergekey = *mergekey ? xstrdup(mergekey) : 0;
 714         clients_research = 1;
 715         session_log(se, YLOG_DEBUG, "session_sort: new mergekey = %s",
 716                     mergekey);
 717     }
 718     if (clients_research == 0)
 719     {
 720         struct reclist_sortparms *sr;
 721         for (sr = se->sorted_results; sr; sr = sr->next)
 722             if (!reclist_sortparms_cmp(sr, sp))
 723                 break;
 724         if (sr)
 725         {
 726             session_log(se, YLOG_LOG, "session_sort: field=%s increasing=%d type=%d already fetched",
 727                         field, increasing, type);
 728             session_leave(se, "session_sort");
 729             return;
 730         }
 731     }
 732     session_log(se, YLOG_DEBUG, "session_sort: field=%s increasing=%d type=%d must fetch",
 733                 field, increasing, type);
 734
 735     // We need to reset reclist on every sort that changes the records, not just for position
 736     // So if just one client requires new searching, we need to clear set.
 737     // Ask each of the client if sorting requires re-search due to native sort
 738     // If it does it will require us to
 739     for (l = se->clients_active; l; l = l->next)
 740     {
 741         struct client *cl = l->client;
 742         // Assume no re-search is required.
 743         client_parse_init(cl, 1);
 744         clients_research += client_parse_sort(cl, sp, 0);
 745     }
 746     if (!clients_research || se->clients_starting)
 747     {
 748         // A new sorting based on same record set
 749         struct reclist_sortparms *sr = nmem_malloc(se->nmem, sizeof(*sr));
 750         sr->name = nmem_strdup(se->nmem, field);
 751         sr->increasing = increasing;
 752         sr->type = type;
 753         sr->next = se->sorted_results;
 754         se->sorted_results = sr;
 755         session_log(se, YLOG_DEBUG, "session_sort: no research/ingesting done");
 756         session_leave(se, "session_sort");
 757     }
 758     else
 759     {
 760         se->clients_starting = 1;
 761         session_log(se, YLOG_DEBUG,
 762                     "session_sort: reset results due to %d clients researching",
 763                     clients_research);
 764         session_clear_set(se, sp);
 765         session_log(se, YLOG_DEBUG, "Re- search/ingesting for clients due to change in sort order");
 766
 767         session_leave(se, "session_sort");
 768         for (l = se->clients_active; l; l = l->next)
 769         {
 770             struct client *cl = l->client;
 771             if (client_get_state(cl) == Client_Connecting ||
 772                 client_get_state(cl) == Client_Idle ||
 773                 client_get_state(cl) == Client_Working) {
 774                 client_start_search(cl);
 775             }
 776             else
 777             {
 778                 session_log(se, YLOG_DEBUG,
 779                             "session_sort: %s: No re-start/ingest in show. "
 780                             "Wrong client state: %d",
 781                             client_get_id(cl), client_get_state(cl));
 782             }
 783         }
 784         session_enter(se, "session_sort");
 785         se->clients_starting = 0;
 786         se->force_position = 0;
 787         session_leave(se, "session_sort");
 788     }
 789 }
 790
 791 void session_stop(struct session *se)
 792 {
 793     struct client_list *l;
 794     session_enter(se, "session_stop1");
 795     if (se->clients_starting)
 796     {
 797         session_leave(se, "session_stop1");
 798         return;
 799     }
 800     se->clients_starting = 1;
 801     session_leave(se, "session_stop1");
 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         struct client *cl = l->client;
 811         client_stop(cl);
 812     }
 813     session_enter(se, "session_stop2");
 814     se->clients_starting = 0;
 815     session_leave(se, "session_stop2");
 816 }
 817
 818 enum pazpar2_error_code session_search(struct session *se,
 819                                        const char *query,
 820                                        const char *startrecs,
 821                                        const char *maxrecs,
 822                                        const char *filter,
 823                                        const char *limit,
 824                                        const char **addinfo,
 825                                        const char **addinfo2,
 826                                        struct reclist_sortparms *sp,
 827                                        const char *mergekey,
 828                                        const char *rank)
 829 {
 830     int live_channels = 0;
 831     int no_working = 0;
 832     int no_failed_query = 0;
 833     int no_failed_limit = 0;
 834     int no_sortmap = 0;
 835     struct client_list *l;
 836
 837     session_log(se, YLOG_DEBUG, "Search");
 838
 839     *addinfo = 0;
 840
 841     session_enter(se, "session_search0");
 842     if (se->clients_starting)
 843     {
 844         session_leave(se, "session_search0");
 845         return PAZPAR2_NO_ERROR;
 846     }
 847     se->clients_starting = 1;
 848     se->force_position = 0;
 849     session_leave(se, "session_search0");
 850
 851     if (se->settings_modified) {
 852         session_remove_cached_clients(se);
 853     }
 854     else
 855         session_reset_active_clients(se, 0);
 856
 857     session_enter(se, "session_search");
 858     se->settings_modified = 0;
 859
 860     if (mergekey)
 861     {
 862         xfree(se->mergekey);
 863         se->mergekey = *mergekey ? xstrdup(mergekey) : 0;
 864     }
 865     if (rank)
 866     {
 867         xfree(se->rank);
 868         se->rank = *rank ? xstrdup(rank) : 0;
 869     }
 870
 871     session_clear_set(se, sp);
 872     relevance_destroy(&se->relevance);
 873
 874     live_channels = select_targets(se, filter);
 875     if (!live_channels)
 876     {
 877         session_leave(se, "session_search");
 878         se->clients_starting = 0;
 879         return PAZPAR2_NO_TARGETS;
 880     }
 881
 882     facet_limits_destroy(se->facet_limits);
 883     se->facet_limits = facet_limits_create(limit);
 884     if (!se->facet_limits)
 885     {
 886         *addinfo = "limit";
 887         session_leave(se, "session_search");
 888         se->clients_starting = 0;
 889         session_reset_active_clients(se, 0);
 890         return PAZPAR2_MALFORMED_PARAMETER_VALUE;
 891     }
 892
 893     session_leave(se, "session_search");
 894
 895     session_alert_watch(se, SESSION_WATCH_SHOW);
 896     session_alert_watch(se, SESSION_WATCH_BYTARGET);
 897     session_alert_watch(se, SESSION_WATCH_TERMLIST);
 898     session_alert_watch(se, SESSION_WATCH_SHOW_PREF);
 899
 900     for (l = se->clients_active; l; l = l->next)
 901     {
 902         int parse_ret;
 903         struct client *cl = l->client;
 904         client_parse_init(cl, 1);
 905         if (prepare_map(se, client_get_database(cl)) < 0)
 906             continue;
 907
 908         parse_ret = client_parse_query(cl, query, se->facet_limits, addinfo2);
 909         if (parse_ret == -1)
 910             no_failed_query++;
 911         else if (parse_ret == -2)
 912             no_failed_limit++;
 913         else if (parse_ret < 0)
 914             no_working++; /* other error, such as bad CCL map */
 915         else
 916         {
 917             client_parse_range(cl, startrecs, maxrecs);
 918             client_parse_sort(cl, sp, &no_sortmap);
 919             client_start_search(cl);
 920             no_working++;
 921         }
 922     }
 923     yaz_log(YLOG_LOG, "session_search: no_working=%d no_sortmap=%d",
 924             no_working, no_sortmap);
 925     session_enter(se, "session_search2");
 926     if (no_working == 1 && no_sortmap == 1)
 927     {
 928         se->force_position = 1;
 929         yaz_log(YLOG_LOG, "force_position=1");
 930     }
 931     se->clients_starting = 0;
 932     session_leave(se, "session_search2");
 933     if (no_working == 0)
 934     {
 935         if (no_failed_query > 0)
 936         {
 937             *addinfo = "query";
 938             return PAZPAR2_MALFORMED_PARAMETER_VALUE;
 939         }
 940         else if (no_failed_limit > 0)
 941         {
 942             *addinfo = "limit";
 943             return PAZPAR2_MALFORMED_PARAMETER_VALUE;
 944         }
 945         else
 946             return PAZPAR2_NO_TARGETS;
 947     }
 948     return PAZPAR2_NO_ERROR;
 949 }
 950
 951 // Creates a new session_database object for a database
 952 static void session_init_databases_fun(void *context, struct database *db)
 953 {
 954     struct session *se = (struct session *) context;
 955     struct session_database *new = nmem_malloc(se->session_nmem, sizeof(*new));
 956     int i;
 957
 958     new->database = db;
 959
 960     new->map = 0;
 961     assert(db->settings);
 962     new->settings = nmem_malloc(se->session_nmem,
 963                                 sizeof(struct settings *) * db->num_settings);
 964     new->num_settings = db->num_settings;
 965     for (i = 0; i < db->num_settings; i++)
 966     {
 967         struct setting *setting = db->settings[i];
 968         new->settings[i] = setting;
 969     }
 970     new->next = se->databases;
 971     se->databases = new;
 972 }
 973
 974 // Doesn't free memory associated with sdb -- nmem takes care of that
 975 static void session_database_destroy(struct session_database *sdb)
 976 {
 977     sdb->map = 0;
 978 }
 979
 980 // Initialize session_database list -- this represents this session's view
 981 // of the database list -- subject to modification by the settings ws command
 982 void session_init_databases(struct session *se)
 983 {
 984     se->databases = 0;
 985     predef_grep_databases(se, se->service, session_init_databases_fun);
 986 }
 987
 988 // Probably session_init_databases_fun should be refactored instead of
 989 // called here.
 990 static struct session_database *load_session_database(struct session *se,
 991                                                       const char *id)
 992 {
 993     struct database *db = new_database_inherit_settings(id, se->session_nmem, se->service->settings);
 994     session_init_databases_fun((void*) se, db);
 995
 996     // New sdb is head of se->databases list
 997     return se->databases;
 998 }
 999
1000 // Find an existing session database. If not found, load it
1001 static struct session_database *find_session_database(struct session *se,
1002                                                       const char *id)
1003 {
1004     struct session_database *sdb;
1005
1006     for (sdb = se->databases; sdb; sdb = sdb->next)
1007         if (!strcmp(sdb->database->id, id))
1008             return sdb;
1009     return load_session_database(se, id);
1010 }
1011
1012 // Apply a session override to a database
1013 void session_apply_setting(struct session *se, const char *dbname,
1014                            const char *name, const char *value)
1015 {
1016     session_enter(se, "session_apply_setting");
1017     {
1018         struct session_database *sdb = find_session_database(se, dbname);
1019         struct conf_service *service = se->service;
1020         struct setting *s;
1021         int offset = settings_create_offset(service, name);
1022
1023         expand_settings_array(&sdb->settings, &sdb->num_settings, offset,
1024                               se->session_nmem);
1025         // Force later recompute of settings-driven data structures
1026         // (happens when a search starts and client connections are prepared)
1027         if (offset == PZ_XSLT)
1028             sdb->map = 0;
1029         se->settings_modified = 1;
1030         for (s = sdb->settings[offset]; s; s = s->next)
1031             if (!strcmp(s->name, name) &&
1032                 dbname && s->target && !strcmp(dbname, s->target))
1033                 break;
1034         if (!s)
1035         {
1036             s = nmem_malloc(se->session_nmem, sizeof(*s));
1037             s->precedence = 0;
1038             s->target = nmem_strdup(se->session_nmem, dbname);
1039             s->name = nmem_strdup(se->session_nmem, name);
1040             s->next = sdb->settings[offset];
1041             sdb->settings[offset] = s;
1042         }
1043         s->value = nmem_strdup(se->session_nmem, value);
1044     }
1045     session_leave(se, "session_apply_setting");
1046 }
1047
1048 void session_destroy(struct session *se)
1049 {
1050     struct session_database *sdb;
1051     struct facet_id *t;
1052     int sessions_total = session_use(-1);
1053     int no_facet_ids = 0;
1054
1055     for (t = se->facet_id_list; t; t = t->next)
1056         no_facet_ids++;
1057     session_log(se, YLOG_LOG, "destroy "
1058                 "session-total %d nmem-op %zd nmem-ses %zd facets-ids %d",
1059                 sessions_total,
1060                 nmem_total(se->nmem), nmem_total(se->session_nmem),
1061                 no_facet_ids);
1062     session_remove_cached_clients(se);
1063
1064     for (sdb = se->databases; sdb; sdb = sdb->next)
1065         session_database_destroy(sdb);
1066     normalize_cache_destroy(se->normalize_cache);
1067     relevance_destroy(&se->relevance);
1068     reclist_destroy(se->reclist);
1069     xfree(se->mergekey);
1070     xfree(se->rank);
1071     facet_limits_destroy(se->facet_limits);
1072     nmem_destroy(se->nmem);
1073     service_destroy(se->service);
1074     yaz_mutex_destroy(&se->session_mutex);
1075 }
1076
1077 size_t session_get_memory_status(struct session *session) {
1078     size_t session_nmem;
1079     if (session == 0)
1080         return 0;
1081     session_enter(session, "session_get_memory_status");
1082     session_nmem = nmem_total(session->nmem);
1083     session_leave(session, "session_get_memory_status");
1084     return session_nmem;
1085 }
1086
1087
1088 struct session *session_create(NMEM nmem, struct conf_service *service,
1089                                unsigned session_id)
1090 {
1091     int i;
1092     struct session *session = nmem_malloc(nmem, sizeof(*session));
1093
1094     char tmp_str[50];
1095
1096     sprintf(tmp_str, "session#%u", session_id);
1097
1098     session->session_id = session_id;
1099     session_log(session, YLOG_DEBUG, "New");
1100     session->service = service;
1101     session->relevance = 0;
1102     session->total_records = 0;
1103     session->number_of_warnings_unknown_elements = 0;
1104     session->number_of_warnings_unknown_metadata = 0;
1105     session->termlists = 0;
1106     session->reclist = reclist_create(nmem);
1107     session->clients_active = 0;
1108     session->clients_cached = 0;
1109     session->settings_modified = 0;
1110     session->session_nmem = nmem;
1111     session->facet_id_list = 0;
1112     session->nmem = nmem_create();
1113     session->databases = 0;
1114     session->sorted_results = 0;
1115     session->facet_limits = 0;
1116     session->mergekey = 0;
1117     session->rank = 0;
1118     session->clients_starting = 0;
1119     session->force_position = 0;
1120
1121     for (i = 0; i <= SESSION_WATCH_MAX; i++)
1122     {
1123         session->watchlist[i].data = 0;
1124         session->watchlist[i].fun = 0;
1125     }
1126     session->normalize_cache = normalize_cache_create();
1127     session->session_mutex = 0;
1128     pazpar2_mutex_create(&session->session_mutex, tmp_str);
1129
1130     i = session_use(1);
1131     session_log(session, YLOG_LOG, "create session-total %d", i);
1132     return session;
1133 }
1134
1135 static struct hitsbytarget *hitsbytarget_nb(struct session *se,
1136                                             int *count, NMEM nmem)
1137 {
1138     struct hitsbytarget *res = 0;
1139     struct client_list *l;
1140     size_t sz = 0;
1141
1142     for (l = se->clients_active; l; l = l->next)
1143         sz++;
1144
1145     res = nmem_malloc(nmem, sizeof(*res) * sz);
1146     *count = 0;
1147     for (l = se->clients_active; l; l = l->next)
1148     {
1149         struct client *cl = l->client;
1150         WRBUF w = wrbuf_alloc();
1151         const char *name = session_setting_oneval(client_get_database(cl),
1152                                                   PZ_NAME);
1153         res[*count].id = client_get_id(cl);
1154         res[*count].name = *name ? name : "Unknown";
1155         res[*count].hits = client_get_hits(cl);
1156         res[*count].approximation = client_get_approximation(cl);
1157         res[*count].records = client_get_num_records(cl,
1158                                                      &res[*count].filtered,
1159                                                      0, 0);
1160         res[*count].diagnostic =
1161             client_get_diagnostic(cl, &res[*count].message,
1162                                   &res[*count].addinfo);
1163         res[*count].state = client_get_state_str(cl);
1164         res[*count].connected  = client_get_connection(cl) ? 1 : 0;
1165         session_settings_dump(se, client_get_database(cl), w);
1166         res[*count].settings_xml = nmem_strdup(nmem, wrbuf_cstr(w));
1167         wrbuf_rewind(w);
1168         res[*count].suggestions_xml =
1169             nmem_strdup(nmem, client_get_suggestions_xml(cl, w));
1170
1171         res[*count].query_data =
1172             client_get_query(cl, &res[*count].query_type, nmem);
1173         wrbuf_destroy(w);
1174         (*count)++;
1175     }
1176     return res;
1177 }
1178
1179 struct hitsbytarget *get_hitsbytarget(struct session *se, int *count, NMEM nmem)
1180 {
1181     struct hitsbytarget *p;
1182     session_enter(se, "get_hitsbytarget");
1183     p = hitsbytarget_nb(se, count, nmem);
1184     session_leave(se, "get_hitsbytarget");
1185     return p;
1186 }
1187
1188 // Compares two hitsbytarget nodes by hitcount
1189 static int cmp_ht(const void *p1, const void *p2)
1190 {
1191     const struct hitsbytarget *h1 = p1;
1192     const struct hitsbytarget *h2 = p2;
1193     return h2->hits - h1->hits;
1194 }
1195
1196 // Compares two hitsbytarget nodes by hitcount
1197 static int cmp_ht_approx(const void *p1, const void *p2)
1198 {
1199     const struct hitsbytarget *h1 = p1;
1200     const struct hitsbytarget *h2 = p2;
1201     return h2->approximation - h1->approximation;
1202 }
1203
1204 static int targets_termlist_nb(WRBUF wrbuf, struct session *se, int num,
1205                                NMEM nmem, int version)
1206 {
1207     struct hitsbytarget *ht;
1208     int count, i;
1209
1210     ht = hitsbytarget_nb(se, &count, nmem);
1211     if (version >= 2)
1212         qsort(ht, count, sizeof(struct hitsbytarget), cmp_ht_approx);
1213     else
1214         qsort(ht, count, sizeof(struct hitsbytarget), cmp_ht);
1215     for (i = 0; i < count && i < num && ht[i].hits > 0; i++)
1216     {
1217
1218         // do only print terms which have display names
1219
1220         wrbuf_puts(wrbuf, "<term>\n");
1221
1222         wrbuf_puts(wrbuf, "<id>");
1223         wrbuf_xmlputs(wrbuf, ht[i].id);
1224         wrbuf_puts(wrbuf, "</id>\n");
1225
1226         wrbuf_puts(wrbuf, "<name>");
1227         if (!ht[i].name || !ht[i].name[0])
1228             wrbuf_xmlputs(wrbuf, "NO TARGET NAME");
1229         else
1230             wrbuf_xmlputs(wrbuf, ht[i].name);
1231         wrbuf_puts(wrbuf, "</name>\n");
1232
1233         wrbuf_printf(wrbuf, "<frequency>" ODR_INT_PRINTF "</frequency>\n",
1234                      ht[i].hits);
1235
1236         if (version >= 2) {
1237             // Should not print if we know it isn't a approximation.
1238             wrbuf_printf(wrbuf, "<approximation>" ODR_INT_PRINTF "</approximation>\n", ht[i].approximation);
1239             wrbuf_printf(wrbuf, "<records>%d</records>\n", ht[i].records - ht[i].filtered);
1240             wrbuf_printf(wrbuf, "<filtered>%d</filtered>\n", ht[i].filtered);
1241         }
1242
1243         wrbuf_puts(wrbuf, "<state>");
1244         wrbuf_xmlputs(wrbuf, ht[i].state);
1245         wrbuf_puts(wrbuf, "</state>\n");
1246
1247         wrbuf_printf(wrbuf, "<diagnostic>%d</diagnostic>\n",
1248                      ht[i].diagnostic);
1249         wrbuf_puts(wrbuf, "</term>\n");
1250     }
1251     return count;
1252 }
1253
1254 void perform_termlist(struct http_channel *c, struct session *se,
1255                       const char *name, int num, int version)
1256 {
1257     int j;
1258     NMEM nmem_tmp = nmem_create();
1259     char **names;
1260     int num_names = 0;
1261
1262     if (!name)
1263         name = "*";
1264
1265     nmem_strsplit(nmem_tmp, ",", name, &names, &num_names);
1266
1267     session_enter(se, "perform_termlist");
1268
1269     for (j = 0; j < num_names; j++)
1270     {
1271         const char *tname;
1272         int must_generate_empty = 1; /* bug 5350 */
1273
1274         struct named_termlist *t = se->termlists;
1275         for (; t; t = t->next)
1276         {
1277             tname = t->name;
1278             if (!strcmp(names[j], tname) || !strcmp(names[j], "*"))
1279             {
1280                 struct termlist_score **p = 0;
1281                 int len;
1282
1283                 wrbuf_puts(c->wrbuf, "<list name=\"");
1284                 wrbuf_xmlputs(c->wrbuf, tname);
1285                 wrbuf_puts(c->wrbuf, "\">\n");
1286                 must_generate_empty = 0;
1287
1288                 p = termlist_highscore(t->termlist, &len, nmem_tmp);
1289                 if (p)
1290                 {
1291                     int i;
1292                     for (i = 0; i < len && i < num; i++)
1293                     {
1294                         // prevent sending empty term elements
1295                         if (!p[i]->display_term || !p[i]->display_term[0])
1296                             continue;
1297
1298                         wrbuf_puts(c->wrbuf, "<term>");
1299                         wrbuf_puts(c->wrbuf, "<name>");
1300                         wrbuf_xmlputs(c->wrbuf, p[i]->display_term);
1301                         wrbuf_puts(c->wrbuf, "</name>");
1302                         wrbuf_printf(c->wrbuf,
1303                                      "<frequency>%d</frequency>",
1304                                      p[i]->frequency);
1305                         wrbuf_puts(c->wrbuf, "</term>\n");
1306                     }
1307                 }
1308                 wrbuf_puts(c->wrbuf, "</list>\n");
1309             }
1310         }
1311         tname = "xtargets";
1312         if (!strcmp(names[j], tname) || !strcmp(names[j], "*"))
1313         {
1314             wrbuf_puts(c->wrbuf, "<list name=\"");
1315             wrbuf_xmlputs(c->wrbuf, tname);
1316             wrbuf_puts(c->wrbuf, "\">\n");
1317
1318             targets_termlist_nb(c->wrbuf, se, num, c->nmem, version);
1319             wrbuf_puts(c->wrbuf, "</list>\n");
1320             must_generate_empty = 0;
1321         }
1322         if (must_generate_empty)
1323         {
1324             wrbuf_puts(c->wrbuf, "<list name=\"");
1325             wrbuf_xmlputs(c->wrbuf, names[j]);
1326             wrbuf_puts(c->wrbuf, "\"/>\n");
1327         }
1328     }
1329     session_leave(se, "perform_termlist");
1330     nmem_destroy(nmem_tmp);
1331 }
1332
1333 #ifdef MISSING_HEADERS
1334 void report_nmem_stats(void)
1335 {
1336     size_t in_use, is_free;
1337
1338     nmem_get_memory_in_use(&in_use);
1339     nmem_get_memory_free(&is_free);
1340
1341     yaz_log(YLOG_LOG, "nmem stat: use=%ld free=%ld",
1342             (long) in_use, (long) is_free);
1343 }
1344 #endif
1345
1346 struct record_cluster *show_single_start(struct session *se, const char *id,
1347                                          struct record_cluster **prev_r,
1348                                          struct record_cluster **next_r)
1349 {
1350     struct record_cluster *r = 0;
1351
1352     session_enter(se, "show_single_start");
1353     *prev_r = 0;
1354     *next_r = 0;
1355     reclist_limit(se->reclist, se, 1);
1356
1357     reclist_enter(se->reclist);
1358     while ((r = reclist_read_record(se->reclist)))
1359     {
1360         if (!strcmp(r->recid, id))
1361         {
1362             *next_r = reclist_read_record(se->reclist);
1363             break;
1364         }
1365         *prev_r = r;
1366     }
1367     reclist_leave(se->reclist);
1368     if (!r)
1369         session_leave(se, "show_single_start");
1370     return r;
1371 }
1372
1373 void show_single_stop(struct session *se, struct record_cluster *rec)
1374 {
1375     session_leave(se, "show_single_stop");
1376 }
1377
1378
1379 int session_fetch_more(struct session *se)
1380 {
1381     struct client_list *l;
1382     int ret = 0;
1383
1384     for (l = se->clients_active; l; l = l->next)
1385     {
1386         struct client *cl = l->client;
1387         if (client_get_state(cl) == Client_Idle)
1388         {
1389             if (client_fetch_more(cl))
1390             {
1391                 session_log(se, YLOG_LOG, "%s: more to fetch",
1392                             client_get_id(cl));
1393                 ret = 1;
1394             }
1395             else
1396             {
1397                 int filtered;
1398                 int ingest_failures;
1399                 int record_failures;
1400                 int num = client_get_num_records(
1401                     cl, &filtered, &ingest_failures, &record_failures);
1402
1403                 session_log(se, YLOG_LOG, "%s: hits=" ODR_INT_PRINTF
1404                             " fetched=%d filtered=%d",
1405                             client_get_id(cl),
1406                             client_get_hits(cl),
1407                             num, filtered);
1408                 if (ingest_failures || record_failures)
1409                 {
1410                     session_log(se, YLOG_WARN, "%s:"
1411                                 " ingest failures=%d record failures=%d",
1412                                 client_get_id(cl),
1413                                 ingest_failures, record_failures);
1414                 }
1415             }
1416         }
1417         else
1418         {
1419             session_log(se, YLOG_LOG, "%s: no fetch due to state=%s",
1420                         client_get_id(cl), client_get_state_str(cl));
1421         }
1422
1423     }
1424     return ret;
1425 }
1426
1427 struct record_cluster **show_range_start(struct session *se,
1428                                          struct reclist_sortparms *sp,
1429                                          int start, int *num, int *total,
1430                                          Odr_int *sumhits, Odr_int *approx_hits,
1431                                          void (*show_records_ready)(void *data),
1432                                          struct http_channel *chan)
1433 {
1434     struct record_cluster **recs = 0;
1435     struct reclist_sortparms *spp;
1436     struct client_list *l;
1437     int i;
1438     NMEM nmem_tmp = 0;
1439 #if USE_TIMING
1440     yaz_timing_t t = yaz_timing_create();
1441 #endif
1442     session_enter(se, "show_range_start");
1443     *sumhits = 0;
1444     *approx_hits = 0;
1445     *total = 0;
1446     reclist_limit(se->reclist, se, 0);
1447     if (se->relevance)
1448     {
1449         for (spp = sp; spp; spp = spp->next)
1450             if (spp->type == Metadata_type_relevance)
1451             {
1452                 relevance_prepare_read(se->relevance, se->reclist);
1453                 break;
1454             }
1455         for (l = se->clients_active; l; l = l->next) {
1456             *sumhits += client_get_hits(l->client);
1457             *approx_hits += client_get_approximation(l->client);
1458         }
1459     }
1460     if (se->force_position)
1461     {
1462         nmem_tmp = nmem_create();
1463         sp = reclist_parse_sortparms(nmem_tmp, "position:1", 0);
1464         assert(sp);
1465     }
1466     reclist_sort(se->reclist, sp);
1467     if (nmem_tmp)
1468         nmem_destroy(nmem_tmp);
1469
1470     reclist_enter(se->reclist);
1471     *total = reclist_get_num_records(se->reclist);
1472
1473     for (l = se->clients_active; l; l = l->next)
1474         client_update_show_stat(l->client, 0);
1475
1476     for (i = 0; i < start; i++)
1477     {
1478         struct record_cluster *r = reclist_read_record(se->reclist);
1479         if (!r)
1480         {
1481             *num = 0;
1482             break;
1483         }
1484         else
1485         {
1486             struct record *rec = r->records;
1487             for (;rec; rec = rec->next)
1488                 client_update_show_stat(rec->client, 1);
1489         }
1490     }
1491     recs = nmem_malloc(se->nmem, (*num > 0 ? *num : 1) * sizeof(*recs));
1492     for (i = 0; i < *num; i++)
1493     {
1494         struct record_cluster *r = reclist_read_record(se->reclist);
1495         if (!r)
1496         {
1497             *num = i;
1498             break;
1499         }
1500         else
1501         {
1502             struct record *rec = r->records;
1503             for (;rec; rec = rec->next)
1504                 client_update_show_stat(rec->client, 1);
1505             recs[i] = r;
1506         }
1507     }
1508     reclist_leave(se->reclist);
1509 #if USE_TIMING
1510     yaz_timing_stop(t);
1511     session_log(se, YLOG_LOG, "show %6.5f %3.2f %3.2f",
1512             yaz_timing_get_real(t), yaz_timing_get_user(t),
1513             yaz_timing_get_sys(t));
1514     yaz_timing_destroy(&t);
1515 #endif
1516
1517     if (!session_fetch_more(se))
1518         session_log(se, YLOG_LOG, "can not fetch more");
1519     else
1520     {
1521         show_range_stop(se, recs);
1522         session_log(se, YLOG_LOG, "fetching more in progress");
1523         if (session_set_watch(se, SESSION_WATCH_SHOW,
1524                               show_records_ready, chan, chan))
1525         {
1526             session_log(se, YLOG_WARN, "Ignoring show block");
1527             session_enter(se, "show_range_start");
1528         }
1529         else
1530         {
1531             session_log(se, YLOG_LOG, "session watch OK");
1532             return 0;
1533         }
1534     }
1535     return recs;
1536 }
1537
1538 void show_range_stop(struct session *se, struct record_cluster **recs)
1539 {
1540     session_leave(se, "show_range_stop");
1541 }
1542
1543 void statistics(struct session *se, struct statistics *stat)
1544 {
1545     struct client_list *l;
1546     int count = 0;
1547
1548     memset(stat, 0, sizeof(*stat));
1549     stat->num_hits = 0;
1550     for (l = se->clients_active; l; l = l->next)
1551     {
1552         struct client *cl = l->client;
1553         if (!client_get_connection(cl))
1554             stat->num_no_connection++;
1555         stat->num_hits += client_get_hits(cl);
1556         switch (client_get_state(cl))
1557         {
1558         case Client_Connecting: stat->num_connecting++; break;
1559         case Client_Working: stat->num_working++; break;
1560         case Client_Idle: stat->num_idle++; break;
1561         case Client_Failed: stat->num_failed++; break;
1562         case Client_Error: stat->num_error++; break;
1563         default: break;
1564         }
1565         count++;
1566     }
1567     stat->num_records = se->total_records;
1568
1569     stat->num_clients = count;
1570 }
1571
1572 static struct record_metadata *record_metadata_init(
1573     NMEM nmem, const char *value, const char *norm,
1574     enum conf_metadata_type type,
1575     struct _xmlAttr *attr)
1576 {
1577     struct record_metadata *rec_md = record_metadata_create(nmem);
1578     struct record_metadata_attr **attrp = &rec_md->attributes;
1579
1580     for (; attr; attr = attr->next)
1581     {
1582         if (attr->children && attr->children->content)
1583         {
1584             if (strcmp((const char *) attr->name, "type")
1585                 && strcmp((const char *) attr->name, "empty"))
1586             {  /* skip the "type" + "empty" attribute..
1587                   The "Type" is already part of the element in output
1588                   (md-%s) and so repeating it here is redundant */
1589                 *attrp = nmem_malloc(nmem, sizeof(**attrp));
1590                 (*attrp)->name =
1591                     nmem_strdup(nmem, (const char *) attr->name);
1592                 (*attrp)->value =
1593                     nmem_strdup(nmem, (const char *) attr->children->content);
1594                 attrp = &(*attrp)->next;
1595             }
1596         }
1597     }
1598     *attrp = 0;
1599
1600     switch (type)
1601     {
1602     case Metadata_type_generic:
1603     case Metadata_type_skiparticle:
1604         if (norm)
1605         {
1606             rec_md->data.text.disp = nmem_strdup(nmem, value);
1607             rec_md->data.text.norm = nmem_strdup(nmem, norm);
1608         }
1609         else
1610         {
1611             if (strstr(value, "://")) /* looks like a URL */
1612                 rec_md->data.text.disp = nmem_strdup(nmem, value);
1613             else
1614                 rec_md->data.text.disp =
1615                     normalize7bit_generic(nmem_strdup(nmem, value), " ,/.:([");
1616             rec_md->data.text.norm = rec_md->data.text.disp;
1617         }
1618         rec_md->data.text.sort = 0;
1619         rec_md->data.text.snippet = 0;
1620         break;
1621     case Metadata_type_year:
1622     case Metadata_type_date:
1623     {
1624         int first, last;
1625         int longdate = 0;
1626
1627         if (type == Metadata_type_date)
1628             longdate = 1;
1629         if (extract7bit_dates((char *) value, &first, &last, longdate) < 0)
1630             return 0;
1631
1632         rec_md->data.number.min = first;
1633         rec_md->data.number.max = last;
1634     }
1635     break;
1636     case Metadata_type_float:
1637         rec_md->data.fnumber = atof(value);
1638         break;
1639     case Metadata_type_relevance:
1640     case Metadata_type_position:
1641     case Metadata_type_retrieval:
1642         return 0;
1643     }
1644     return rec_md;
1645 }
1646
1647 static void mergekey_norm_wr(pp2_charset_fact_t charsets,
1648                              WRBUF norm_wr, const char *value)
1649 {
1650     const char *norm_str;
1651     pp2_charset_token_t prt =
1652         pp2_charset_token_create(charsets, "mergekey");
1653
1654     pp2_charset_token_first(prt, value, 0);
1655     while ((norm_str = pp2_charset_token_next(prt)))
1656     {
1657         if (*norm_str)
1658         {
1659             if (wrbuf_len(norm_wr))
1660                 wrbuf_puts(norm_wr, " ");
1661             wrbuf_puts(norm_wr, norm_str);
1662         }
1663     }
1664     pp2_charset_token_destroy(prt);
1665 }
1666
1667 static int get_mergekey_from_doc(xmlDoc *doc, xmlNode *root, const char *name,
1668                                  struct conf_service *service, WRBUF norm_wr)
1669 {
1670     xmlNode *n;
1671     int no_found = 0;
1672     for (n = root->children; n; n = n->next)
1673     {
1674         if (n->type != XML_ELEMENT_NODE)
1675             continue;
1676         if (!strcmp((const char *) n->name, "metadata"))
1677         {
1678             const char *type = yaz_xml_get_prop(n, "type");
1679             if (type == NULL) {
1680                 yaz_log(YLOG_FATAL, "Missing type attribute on metadata element. Skipping!");
1681             }
1682             else if (!strcmp(name, (const char *) type))
1683             {
1684                 xmlChar *value = xmlNodeListGetString(doc, n->children, 1);
1685                 if (value && *value)
1686                 {
1687                     if (wrbuf_len(norm_wr) > 0)
1688                         wrbuf_puts(norm_wr, " ");
1689                     wrbuf_puts(norm_wr, name);
1690                     mergekey_norm_wr(service->charsets, norm_wr,
1691                                      (const char *) value);
1692                     no_found++;
1693                 }
1694                 if (value)
1695                     xmlFree(value);
1696             }
1697         }
1698     }
1699     return no_found;
1700 }
1701
1702 static const char *get_mergekey(xmlDoc *doc, xmlNode *root,
1703                                 struct client *cl, int record_no,
1704                                 struct conf_service *service, NMEM nmem,
1705                                 const char *session_mergekey)
1706 {
1707     char *mergekey_norm = 0;
1708     WRBUF norm_wr = wrbuf_alloc();
1709     const char *mergekey;
1710
1711     if (session_mergekey)
1712     {
1713         int i, num = 0;
1714         char **values = 0;
1715         nmem_strsplit_escape2(nmem, ",", session_mergekey, &values,
1716                               &num, 1, '\\', 1);
1717
1718         for (i = 0; i < num; i++)
1719             get_mergekey_from_doc(doc, root, values[i], service, norm_wr);
1720     }
1721     else if ((mergekey = yaz_xml_get_prop(root, "mergekey")))
1722     {
1723         mergekey_norm_wr(service->charsets, norm_wr, mergekey);
1724     }
1725     else
1726     {
1727         /* no mergekey defined in XSL. Look for mergekey metadata instead */
1728         int field_id;
1729         for (field_id = 0; field_id < service->num_metadata; field_id++)
1730         {
1731             struct conf_metadata *ser_md = &service->metadata[field_id];
1732             if (ser_md->mergekey != Metadata_mergekey_no)
1733             {
1734                 int r = get_mergekey_from_doc(doc, root, ser_md->name,
1735                                               service, norm_wr);
1736                 if (r == 0 && ser_md->mergekey == Metadata_mergekey_required)
1737                 {
1738                     /* no mergekey on this one and it is required..
1739                        Generate unique key instead */
1740                     wrbuf_rewind(norm_wr);
1741                     break;
1742                 }
1743             }
1744         }
1745     }
1746
1747     /* generate unique key if none is not generated already or is empty */
1748     if (wrbuf_len(norm_wr) == 0)
1749     {
1750         wrbuf_printf(norm_wr, "position: %s-%06d",
1751                      client_get_id(cl), record_no);
1752     }
1753     else
1754     {
1755         const char *lead = "content: ";
1756         wrbuf_insert(norm_wr, 0, lead, strlen(lead));
1757     }
1758     if (wrbuf_len(norm_wr) > 0)
1759         mergekey_norm = nmem_strdup(nmem, wrbuf_cstr(norm_wr));
1760     wrbuf_destroy(norm_wr);
1761     return mergekey_norm;
1762 }
1763
1764 /** \brief see if metadata for pz:recordfilter exists
1765     \param root xml root element of normalized record
1766     \param sdb session database for client
1767     \retval 0 if there is no metadata for pz:recordfilter
1768     \retval 1 if there is metadata for pz:recordfilter
1769
1770     If there is no pz:recordfilter defined, this function returns 1
1771     as well.
1772 */
1773
1774 static int check_record_filter(xmlNode *root, struct session_database *sdb)
1775 {
1776     int match = 0;
1777     xmlNode *n;
1778     const char *s;
1779     s = session_setting_oneval(sdb, PZ_RECORDFILTER);
1780
1781     if (!s || !*s)
1782         return 1;
1783
1784     for (n = root->children; n; n = n->next)
1785     {
1786         if (n->type != XML_ELEMENT_NODE)
1787             continue;
1788         if (!strcmp((const char *) n->name, "metadata"))
1789         {
1790             const char *type = yaz_xml_get_prop(n, "type");
1791             if (type)
1792             {
1793                 size_t len;
1794                 int substring;
1795                 const char *eq;
1796
1797                 if ((eq = strchr(s, '=')))
1798                     substring = 0;
1799                 else if ((eq = strchr(s, '~')))
1800                     substring = 1;
1801                 if (eq)
1802                     len = eq - s;
1803                 else
1804                     len = strlen(s);
1805                 if (len == strlen((const char *)type) &&
1806                     !memcmp((const char *) type, s, len))
1807                 {
1808                     xmlChar *value = xmlNodeGetContent(n);
1809                     if (value && *value)
1810                     {
1811                         if (!eq ||
1812                             (substring && strstr((const char *) value, eq+1)) ||
1813                             (!substring && !strcmp((const char *) value, eq + 1)))
1814                             match = 1;
1815                     }
1816                     xmlFree(value);
1817                 }
1818             }
1819         }
1820     }
1821     return match;
1822 }
1823
1824 static int ingest_to_cluster(struct client *cl,
1825                              WRBUF wrbuf_disp,
1826                              WRBUF wrbuf_norm,
1827                              xmlDoc *xdoc,
1828                              xmlNode *root,
1829                              int record_no,
1830                              struct record_metadata_attr *mergekey);
1831
1832 static int ingest_sub_record(struct client *cl, xmlDoc *xdoc, xmlNode *root,
1833                              int record_no, NMEM nmem,
1834                              struct session_database *sdb,
1835                              struct record_metadata_attr *mergekeys)
1836 {
1837     int ret = 0;
1838     struct session *se = client_get_session(cl);
1839     WRBUF wrbuf_disp, wrbuf_norm;
1840
1841     if (!check_record_filter(root, sdb))
1842     {
1843         session_log(se, YLOG_LOG,
1844                     "Filtered out record no %d from %s",
1845                     record_no, sdb->database->id);
1846         return 0;
1847     }
1848     wrbuf_disp = wrbuf_alloc();
1849     wrbuf_norm = wrbuf_alloc();
1850     session_enter(se, "ingest_sub_record");
1851     if (client_get_session(cl) == se && se->relevance)
1852         ret = ingest_to_cluster(cl, wrbuf_disp, wrbuf_norm,
1853                                 xdoc, root, record_no, mergekeys);
1854     session_leave(se, "ingest_sub_record");
1855     wrbuf_destroy(wrbuf_norm);
1856     wrbuf_destroy(wrbuf_disp);
1857     return ret;
1858 }
1859
1860 /** \brief ingest XML record
1861     \param cl client holds the result set for record
1862     \param rec record buffer (0 terminated)
1863     \param record_no record position (1, 2, ..)
1864     \param nmem working NMEM
1865     \retval 0 OK
1866     \retval -1 failure
1867     \retval -2 Filtered
1868 */
1869 int ingest_record(struct client *cl, const char *rec,
1870                   int record_no, NMEM nmem)
1871 {
1872     struct session *se = client_get_session(cl);
1873     struct session_database *sdb = client_get_database(cl);
1874     struct conf_service *service = se->service;
1875     xmlDoc *xdoc = normalize_record(se, sdb, service, rec, nmem);
1876     int r = ingest_xml_record(cl, xdoc, record_no, nmem, 0);
1877     client_store_xdoc(cl, record_no, xdoc);
1878     return r;
1879 }
1880
1881 int ingest_xml_record(struct client *cl, xmlDoc *xdoc,
1882                       int record_no, NMEM nmem, int cached_copy)
1883 {
1884     struct session *se = client_get_session(cl);
1885     struct session_database *sdb = client_get_database(cl);
1886     struct conf_service *service = se->service;
1887     xmlNode *root;
1888     int r = 0;
1889     if (!xdoc)
1890         return -1;
1891
1892     if (global_parameters.dump_records)
1893     {
1894         session_log(se, YLOG_LOG, "Normalized record from %s",
1895                     sdb->database->id);
1896         log_xml_doc(xdoc);
1897     }
1898
1899     root = xmlDocGetRootElement(xdoc);
1900
1901     if (!strcmp((const char *) root->name, "cluster"))
1902     {
1903         int no_merge_keys = 0;
1904         int no_merge_dups = 0;
1905         xmlNode *sroot;
1906         struct record_metadata_attr *mk = 0;
1907
1908         for (sroot = root->children; sroot; sroot = sroot->next)
1909             if (sroot->type == XML_ELEMENT_NODE &&
1910                 !strcmp((const char *) sroot->name, "record"))
1911             {
1912                 struct record_metadata_attr **mkp;
1913                 const char *mergekey_norm =
1914                     get_mergekey(xdoc, sroot, cl, record_no, service, nmem,
1915                                  se->mergekey);
1916                 if (!mergekey_norm)
1917                 {
1918                     r = -1;
1919                     break;
1920                 }
1921                 for (mkp = &mk; *mkp; mkp = &(*mkp)->next)
1922                     if (!strcmp((*mkp)->value, mergekey_norm))
1923                         break;
1924                 if (!*mkp)
1925                 {
1926                     *mkp = (struct record_metadata_attr*)
1927                         nmem_malloc(nmem, sizeof(**mkp));
1928                     (*mkp)->name = 0;
1929                     (*mkp)->value = nmem_strdup(nmem, mergekey_norm);
1930                     (*mkp)->next = 0;
1931                     no_merge_keys++;
1932                 }
1933                 else
1934                     no_merge_dups++;
1935             }
1936         if (no_merge_keys > 1 || no_merge_dups > 0)
1937         {
1938             yaz_log(YLOG_LOG, "Got %d mergekeys, %d dups for position %d",
1939                     no_merge_keys, no_merge_dups, record_no);
1940         }
1941         for (sroot = root->children; !r && sroot; sroot = sroot->next)
1942             if (sroot->type == XML_ELEMENT_NODE &&
1943                 !strcmp((const char *) sroot->name, "record"))
1944             {
1945                 if (!cached_copy)
1946                     insert_settings_values(sdb, xdoc, root, service);
1947                 r = ingest_sub_record(cl, xdoc, sroot, record_no, nmem, sdb,
1948                                       mk);
1949             }
1950     }
1951     else if (!strcmp((const char *) root->name, "record"))
1952     {
1953         const char *mergekey_norm =
1954             get_mergekey(xdoc, root, cl, record_no, service, nmem,
1955                          se->mergekey);
1956         if (mergekey_norm)
1957         {
1958             struct record_metadata_attr *mk = (struct record_metadata_attr*)
1959                 nmem_malloc(nmem, sizeof(*mk));
1960             mk->name = 0;
1961             mk->value = nmem_strdup(nmem, mergekey_norm);
1962             mk->next = 0;
1963
1964             if (!cached_copy)
1965                 insert_settings_values(sdb, xdoc, root, service);
1966             r = ingest_sub_record(cl, xdoc, root, record_no, nmem, sdb, mk);
1967         }
1968     }
1969     else
1970     {
1971         session_log(se, YLOG_WARN, "Bad pz root element: %s",
1972                     (const char *) root->name);
1973         r = -1;
1974     }
1975     return r;
1976 }
1977
1978
1979 //    struct conf_metadata *ser_md = &service->metadata[md_field_id];
1980 //    struct record_metadata *rec_md = record->metadata[md_field_id];
1981 static int match_metadata_local(struct conf_service *service,
1982                                 struct conf_metadata *ser_md,
1983                                 struct record_metadata *rec_md0,
1984                                 char **values, int num_v)
1985 {
1986     int i;
1987     struct record_metadata *rec_md = rec_md0;
1988     WRBUF val_wr = 0;
1989     WRBUF text_wr = wrbuf_alloc();
1990     for (i = 0; i < num_v; )
1991     {
1992         if (rec_md)
1993         {
1994             if (ser_md->type == Metadata_type_year
1995                 || ser_md->type == Metadata_type_date)
1996             {
1997                 int y = atoi(values[i]);
1998                 if (y >= rec_md->data.number.min
1999                     && y <= rec_md->data.number.max)
2000                     break;
2001             }
2002             else
2003             {
2004                 if (!val_wr)
2005                 {
2006                     val_wr = wrbuf_alloc();
2007                     mergekey_norm_wr(service->charsets, val_wr, values[i]);
2008                 }
2009                 wrbuf_rewind(text_wr);
2010                 mergekey_norm_wr(service->charsets, text_wr,
2011                                  rec_md->data.text.disp);
2012                 if (!strcmp(wrbuf_cstr(val_wr), wrbuf_cstr(text_wr)))
2013                     break;
2014             }
2015             rec_md = rec_md->next;
2016         }
2017         else
2018         {
2019             rec_md = rec_md0;
2020             wrbuf_destroy(val_wr);
2021             val_wr = 0;
2022             i++;
2023         }
2024     }
2025     wrbuf_destroy(val_wr);
2026     wrbuf_destroy(text_wr);
2027     return i < num_v ? 1 : 0;
2028 }
2029
2030 int session_check_cluster_limit(struct session *se, struct record_cluster *rec)
2031 {
2032     int i;
2033     struct conf_service *service = se->service;
2034     int ret = 1;
2035     const char *name;
2036     const char *value;
2037     NMEM nmem_tmp = nmem_create();
2038
2039     for (i = 0; (name = facet_limits_get(se->facet_limits, i, &value)); i++)
2040     {
2041         int j;
2042         for (j = 0; j < service->num_metadata; j++)
2043         {
2044             struct conf_metadata *md = service->metadata + j;
2045             if (!strcmp(md->name, name) && md->limitcluster)
2046             {
2047                 char **values = 0;
2048                 int num = 0;
2049                 int md_field_id =
2050                     conf_service_metadata_field_id(service,
2051                                                    md->limitcluster);
2052
2053                 if (md_field_id < 0)
2054                 {
2055                     ret = 0;
2056                     break;
2057                 }
2058
2059                 nmem_strsplit_escape2(nmem_tmp, "|", value, &values,
2060                                       &num, 1, '\\', 1);
2061
2062                 if (!match_metadata_local(service,
2063                                           &service->metadata[md_field_id],
2064                                           rec->metadata[md_field_id],
2065                                           values, num))
2066                 {
2067                     ret = 0;
2068                     break;
2069                 }
2070             }
2071         }
2072     }
2073     nmem_destroy(nmem_tmp);
2074     return ret;
2075 }
2076
2077 // Skip record on non-zero
2078 static int check_limit_local(struct client *cl,
2079                              struct record *record,
2080                              int record_no)
2081 {
2082     int skip_record = 0;
2083     struct session *se = client_get_session(cl);
2084     struct conf_service *service = se->service;
2085     NMEM nmem_tmp = nmem_create();
2086     struct session_database *sdb = client_get_database(cl);
2087     int l = 0;
2088     while (!skip_record)
2089     {
2090         int md_field_id;
2091         char **values = 0;
2092         int num_v = 0;
2093         const char *name =
2094             client_get_facet_limit_local(cl, sdb, &l, nmem_tmp,
2095                                          &num_v, &values);
2096         if (!name)
2097             break;
2098
2099         if (!strcmp(name, "*"))
2100         {
2101             for (md_field_id = 0; md_field_id < service->num_metadata;
2102                  md_field_id++)
2103             {
2104                 if (match_metadata_local(
2105                         service,
2106                         &service->metadata[md_field_id],
2107                         record->metadata[md_field_id],
2108                         values, num_v))
2109                     break;
2110             }
2111             if (md_field_id == service->num_metadata)
2112                 skip_record = 1;
2113         }
2114         else
2115         {
2116             md_field_id = conf_service_metadata_field_id(service, name);
2117             if (md_field_id < 0)
2118             {
2119                 skip_record = 1;
2120                 break;
2121             }
2122             if (!match_metadata_local(
2123                     service,
2124                     &service->metadata[md_field_id],
2125                     record->metadata[md_field_id],
2126                     values, num_v))
2127             {
2128                 skip_record = 1;
2129             }
2130         }
2131     }
2132     nmem_destroy(nmem_tmp);
2133     return skip_record;
2134 }
2135
2136 static int ingest_to_cluster(struct client *cl,
2137                              WRBUF wrbuf_disp,
2138                              WRBUF wrbuf_norm,
2139                              xmlDoc *xdoc,
2140                              xmlNode *root,
2141                              int record_no,
2142                              struct record_metadata_attr *merge_keys)
2143 {
2144     xmlNode *n;
2145     struct session *se = client_get_session(cl);
2146     struct conf_service *service = se->service;
2147     int term_factor = 1;
2148     struct record_cluster *cluster;
2149     struct record_metadata **metadata0;
2150     struct session_database *sdb = client_get_database(cl);
2151     NMEM ingest_nmem = 0;
2152     char **rank_values = 0;
2153     int rank_num = 0;
2154     struct record *record = record_create(se->nmem,
2155                                           service->num_metadata,
2156                                           service->num_sortkeys, cl,
2157                                           record_no);
2158
2159     for (n = root->children; n; n = n->next)
2160     {
2161         if (n->type != XML_ELEMENT_NODE)
2162             continue;
2163         if (!strcmp((const char *) n->name, "metadata"))
2164         {
2165             struct conf_metadata *ser_md = 0;
2166             struct record_metadata **wheretoput = 0;
2167             struct record_metadata *rec_md = 0;
2168             int md_field_id = -1;
2169             xmlChar *value0;
2170             const char *type = yaz_xml_get_prop(n, "type");
2171
2172             if (!type)
2173                 continue;
2174
2175             md_field_id
2176                 = conf_service_metadata_field_id(service, (const char *) type);
2177             if (md_field_id < 0)
2178             {
2179                 if (se->number_of_warnings_unknown_metadata == 0)
2180                 {
2181                     session_log(se, YLOG_WARN,
2182                             "Ignoring unknown metadata element: %s", type);
2183                 }
2184                 se->number_of_warnings_unknown_metadata++;
2185                 continue;
2186             }
2187
2188             wrbuf_rewind(wrbuf_disp);
2189             value0 = xmlNodeListGetString(xdoc, n->children, 1);
2190             if (!value0 || !*value0)
2191             {
2192                 const char *empty = yaz_xml_get_prop(n, "empty");
2193                 if (!empty)
2194                     continue;
2195                 wrbuf_puts(wrbuf_disp, (const char *) empty);
2196             }
2197             else
2198             {
2199                 wrbuf_puts(wrbuf_disp, (const char *) value0);
2200             }
2201             if (value0)
2202                 xmlFree(value0);
2203             ser_md = &service->metadata[md_field_id];
2204
2205             // non-merged metadata
2206             rec_md = record_metadata_init(se->nmem, wrbuf_cstr(wrbuf_disp), 0,
2207                                           ser_md->type, n->properties);
2208             if (!rec_md)
2209             {
2210                 session_log(se, YLOG_WARN, "bad metadata data '%s' "
2211                             "for element '%s'", wrbuf_cstr(wrbuf_disp), type);
2212                 continue;
2213             }
2214
2215             if (ser_md->type == Metadata_type_generic)
2216             {
2217                 WRBUF w = wrbuf_alloc();
2218                 if (relevance_snippet(se->relevance,
2219                                       wrbuf_cstr(wrbuf_disp), ser_md->name, w))
2220                     rec_md->data.text.snippet = nmem_strdup(se->nmem,
2221                                                             wrbuf_cstr(w));
2222                 wrbuf_destroy(w);
2223             }
2224
2225
2226             wheretoput = &record->metadata[md_field_id];
2227             while (*wheretoput)
2228                 wheretoput = &(*wheretoput)->next;
2229             *wheretoput = rec_md;
2230         }
2231     }
2232
2233     if (check_limit_local(cl, record, record_no))
2234     {
2235         return -2;
2236     }
2237     cluster = reclist_insert(se->reclist, se->relevance, service, record,
2238                              merge_keys, &se->total_merged);
2239     if (!cluster)
2240     {
2241         return 0; // complete match with existing record
2242     }
2243
2244     {
2245         const char *use_term_factor_str =
2246             session_setting_oneval(sdb, PZ_TERMLIST_TERM_FACTOR);
2247         if (use_term_factor_str && use_term_factor_str[0] == '1')
2248         {
2249             int maxrecs = client_get_maxrecs(cl);
2250             int hits = (int) client_get_hits(cl);
2251             term_factor = MAX(hits, maxrecs) /  MAX(1, maxrecs);
2252             assert(term_factor >= 1);
2253             session_log(se, YLOG_DEBUG, "Using term factor: %d (%d / %d)",
2254                         term_factor, MAX(hits, maxrecs), MAX(1, maxrecs));
2255         }
2256     }
2257
2258     if (global_parameters.dump_records)
2259         session_log(se, YLOG_LOG, "Cluster id %s from %s (#%d)", cluster->recid,
2260                     sdb->database->id, record_no);
2261
2262     // original metadata, to check if first existence of a field
2263     metadata0 = xmalloc(sizeof(*metadata0) * service->num_metadata);
2264     memcpy(metadata0, cluster->metadata,
2265            sizeof(*metadata0) * service->num_metadata);
2266
2267     ingest_nmem = nmem_create();
2268     if (se->rank)
2269     {
2270         yaz_log(YLOG_LOG, "local in sort : %s", se->rank);
2271         nmem_strsplit_escape2(ingest_nmem, ",", se->rank, &rank_values,
2272                               &rank_num, 1, '\\', 1);
2273     }
2274
2275     // now parsing XML record and adding data to cluster or record metadata
2276     for (n = root->children; n; n = n->next)
2277     {
2278         if (n->type != XML_ELEMENT_NODE)
2279             continue;
2280         if (!strcmp((const char *) n->name, "metadata"))
2281         {
2282             struct conf_metadata *ser_md = 0;
2283             struct conf_sortkey *ser_sk = 0;
2284             struct record_metadata **wheretoput = 0;
2285             struct record_metadata *rec_md = 0;
2286             int md_field_id = -1;
2287             int sk_field_id = -1;
2288             const char *rank = 0;
2289             const char *xml_rank = 0;
2290             const char *type = 0;
2291             xmlChar *value0;
2292
2293             type = yaz_xml_get_prop(n, "type");
2294             if (!type)
2295                 continue;
2296
2297             md_field_id
2298                 = conf_service_metadata_field_id(service, (const char *) type);
2299             if (md_field_id < 0)
2300                 continue;
2301
2302             ser_md = &service->metadata[md_field_id];
2303
2304             if (ser_md->sortkey_offset >= 0)
2305             {
2306                 sk_field_id = ser_md->sortkey_offset;
2307                 ser_sk = &service->sortkeys[sk_field_id];
2308             }
2309
2310             wrbuf_rewind(wrbuf_disp);
2311             wrbuf_rewind(wrbuf_norm);
2312
2313             value0 = xmlNodeListGetString(xdoc, n->children, 1);
2314             if (!value0 || !*value0)
2315             {
2316                 if (value0)
2317                     xmlFree(value0);
2318                 continue;
2319             }
2320
2321             if (ser_md->icurule)
2322             {
2323                 run_icu(se, ser_md->icurule, (const char *) value0,
2324                         wrbuf_norm, wrbuf_disp);
2325                 yaz_log(YLOG_LOG, "run_icu input=%s norm=%s disp=%s",
2326                         (const char *) value0,
2327                         wrbuf_cstr(wrbuf_norm), wrbuf_cstr(wrbuf_disp));
2328                 rec_md = record_metadata_init(se->nmem, wrbuf_cstr(wrbuf_disp),
2329                                               wrbuf_cstr(wrbuf_norm),
2330                                               ser_md->type, 0);
2331             }
2332             else
2333             {
2334                 wrbuf_puts(wrbuf_disp, (const char *) value0);
2335                 rec_md = record_metadata_init(se->nmem, wrbuf_cstr(wrbuf_disp),
2336                                               0,
2337                                               ser_md->type, 0);
2338             }
2339
2340             xmlFree(value0);
2341
2342             // see if the field was not in cluster already (from beginning)
2343             if (!rec_md)
2344                 continue;
2345
2346             if (rank_num)
2347             {
2348                 int i;
2349                 for (i = 0; i < rank_num; i++)
2350                 {
2351                     const char *val = rank_values[i];
2352                     const char *cp = strchr(val, '=');
2353                     if (!cp)
2354                         continue;
2355                     if ((cp - val) == strlen((const char *) type)
2356                         && !memcmp(val, type, cp - val))
2357                     {
2358                         rank = cp + 1;
2359                         break;
2360                     }
2361                 }
2362             }
2363             else
2364             {
2365                 xml_rank = yaz_xml_get_prop(n, "rank");
2366                 rank = xml_rank ? (const char *) xml_rank : ser_md->rank;
2367             }
2368
2369             wheretoput = &cluster->metadata[md_field_id];
2370
2371             if (ser_md->merge == Metadata_merge_first)
2372             {
2373                 if (!metadata0[md_field_id])
2374                 {
2375                     while (*wheretoput)
2376                         wheretoput = &(*wheretoput)->next;
2377                     *wheretoput = rec_md;
2378                 }
2379             }
2380             else if (ser_md->merge == Metadata_merge_unique)
2381             {
2382                 while (*wheretoput)
2383                 {
2384                     if (!strcmp((const char *) (*wheretoput)->data.text.norm,
2385                                 rec_md->data.text.norm))
2386                         break;
2387                     wheretoput = &(*wheretoput)->next;
2388                 }
2389                 if (!*wheretoput)
2390                     *wheretoput = rec_md;
2391             }
2392             else if (ser_md->merge == Metadata_merge_longest)
2393             {
2394                 if (!*wheretoput
2395                     || strlen(rec_md->data.text.norm)
2396                     > strlen((*wheretoput)->data.text.norm))
2397                 {
2398                     *wheretoput = rec_md;
2399                     if (ser_sk)
2400                     {
2401                         pp2_charset_token_t prt;
2402                         const char *sort_str = 0;
2403                         int skip_article =
2404                             ser_sk->type == Metadata_type_skiparticle;
2405
2406                         if (!cluster->sortkeys[sk_field_id])
2407                             cluster->sortkeys[sk_field_id] =
2408                                 nmem_malloc(se->nmem,
2409                                             sizeof(union data_types));
2410
2411                         prt =
2412                             pp2_charset_token_create(service->charsets, "sort");
2413
2414                         pp2_charset_token_first(prt, rec_md->data.text.disp,
2415                                                 skip_article);
2416
2417                         pp2_charset_token_next(prt);
2418
2419                         sort_str = pp2_get_sort(prt);
2420
2421                         cluster->sortkeys[sk_field_id]->text.disp =
2422                             rec_md->data.text.disp;
2423                         if (!sort_str)
2424                         {
2425                             sort_str = rec_md->data.text.disp;
2426                             session_log(se, YLOG_WARN,
2427                                     "Could not make sortkey. Bug #1858");
2428                         }
2429                         cluster->sortkeys[sk_field_id]->text.sort =
2430                             nmem_strdup(se->nmem, sort_str);
2431                         pp2_charset_token_destroy(prt);
2432                     }
2433                 }
2434             }
2435             else if (ser_md->merge == Metadata_merge_all)
2436             {
2437                 while (*wheretoput)
2438                     wheretoput = &(*wheretoput)->next;
2439                 *wheretoput = rec_md;
2440             }
2441             else if (ser_md->merge == Metadata_merge_range)
2442             {
2443                 if (!*wheretoput)
2444                 {
2445                     *wheretoput = rec_md;
2446                     if (ser_sk)
2447                         cluster->sortkeys[sk_field_id]
2448                             = &rec_md->data;
2449                 }
2450                 else
2451                 {
2452                     int this_min = rec_md->data.number.min;
2453                     int this_max = rec_md->data.number.max;
2454                     if (this_min < (*wheretoput)->data.number.min)
2455                         (*wheretoput)->data.number.min = this_min;
2456                     if (this_max > (*wheretoput)->data.number.max)
2457                         (*wheretoput)->data.number.max = this_max;
2458                 }
2459             }
2460
2461             // ranking of _all_ fields enabled ...
2462             if (rank)
2463             {
2464                 relevance_countwords(se->relevance, cluster,
2465                                      wrbuf_cstr(wrbuf_disp),
2466                                      rank, ser_md->name);
2467             }
2468             // construct facets ... unless the client already has reported them
2469             if (ser_md->termlist && !client_has_facet(cl, (char *) type))
2470             {
2471                 if (ser_md->type == Metadata_type_year)
2472                 {
2473                     char year[64];
2474                     sprintf(year, "%d", rec_md->data.number.max);
2475
2476                     add_facet(se, (char *) type, year, term_factor, cl);
2477                     if (rec_md->data.number.max != rec_md->data.number.min)
2478                     {
2479                         sprintf(year, "%d", rec_md->data.number.min);
2480                         add_facet(se, (char *) type, year, term_factor, cl);
2481                     }
2482                 }
2483                 else
2484                     add_facet(se, type, wrbuf_cstr(wrbuf_disp), term_factor, cl);
2485             }
2486         }
2487         else
2488         {
2489             if (se->number_of_warnings_unknown_elements == 0)
2490                 session_log(se, YLOG_WARN,
2491                         "Unexpected element in internal record: %s", n->name);
2492             se->number_of_warnings_unknown_elements++;
2493         }
2494     }
2495     nmem_destroy(ingest_nmem);
2496     xfree(metadata0);
2497     relevance_donerecord(se->relevance, cluster);
2498     se->total_records++;
2499
2500     return 0;
2501 }
2502
2503 void session_log(struct session *s, int level, const char *fmt, ...)
2504 {
2505     char buf[1024];
2506     va_list ap;
2507     va_start(ap, fmt);
2508
2509     yaz_vsnprintf(buf, sizeof(buf)-30, fmt, ap);
2510     yaz_log(level, "Session %u: %s", s ? s->session_id : 0, buf);
2511
2512     va_end(ap);
2513 }
2514
2515 /*
2516  * Local variables:
2517  * c-basic-offset: 4
2518  * c-file-style: "Stroustrup"
2519  * indent-tabs-mode: nil
2520  * End:
2521  * vim: shiftwidth=4 tabstop=8 expandtab
2522  */
2523