src/session.c

   1 /* This file is part of Pazpar2.
   2    Copyright (C) 2006-2013 Index Data
   3
   4 Pazpar2 is free software; you can redistribute it and/or modify it under
   5 the terms of the GNU General Public License as published by the Free
   6 Software Foundation; either version 2, or (at your option) any later
   7 version.
   8
   9 Pazpar2 is distributed in the hope that it will be useful, but WITHOUT ANY
  10 WARRANTY; without even the implied warranty of MERCHANTABILITY or
  11 FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License
  12 for more details.
  13
  14 You should have received a copy of the GNU General Public License
  15 along with this program; if not, write to the Free Software
  16 Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
  17
  18 */
  19
  20 /** \file session.c
  21     \brief high-level logic; mostly user sessions and settings
  22 */
  23
  24 #if HAVE_CONFIG_H
  25 #include <config.h>
  26 #endif
  27
  28 #include <time.h>
  29 #include <stdlib.h>
  30 #include <stdio.h>
  31 #include <string.h>
  32 #if HAVE_SYS_TIME_H
  33 #include <sys/time.h>
  34 #endif
  35 #if HAVE_UNISTD_H
  36 #include <unistd.h>
  37 #endif
  38 #ifdef WIN32
  39 #include <windows.h>
  40 #endif
  41 #include <signal.h>
  42 #include <ctype.h>
  43 #include <assert.h>
  44 #include <math.h>
  45
  46 #include <yaz/marcdisp.h>
  47 #include <yaz/comstack.h>
  48 #include <yaz/tcpip.h>
  49 #include <yaz/proto.h>
  50 #include <yaz/readconf.h>
  51 #include <yaz/pquery.h>
  52 #include <yaz/otherinfo.h>
  53 #include <yaz/yaz-util.h>
  54 #include <yaz/nmem.h>
  55 #include <yaz/query-charset.h>
  56 #include <yaz/querytowrbuf.h>
  57 #include <yaz/oid_db.h>
  58 #include <yaz/snprintf.h>
  59
  60 #define USE_TIMING 0
  61 #if USE_TIMING
  62 #include <yaz/timing.h>
  63 #endif
  64
  65 #include "ppmutex.h"
  66 #include "parameters.h"
  67 #include "session.h"
  68 #include "eventl.h"
  69 #include "http.h"
  70 #include "termlists.h"
  71 #include "reclists.h"
  72 #include "relevance.h"
  73 #include "database.h"
  74 #include "client.h"
  75 #include "settings.h"
  76 #include "normalize7bit.h"
  77
  78 #include <libxml/tree.h>
  79
  80 #define MAX_CHUNK 15
  81
  82 #define MAX(a,b) ((a)>(b)?(a):(b))
  83
  84 // Note: Some things in this structure will eventually move to configuration
  85 struct parameters global_parameters =
  86 {
  87     0,   // dump_records
  88     0,   // debug_mode
  89     0,   // predictable sessions
  90 };
  91
  92 struct client_list {
  93     struct client *client;
  94     struct client_list *next;
  95 };
  96
  97 /* session counting (1) , disable client counting (0) */
  98 static YAZ_MUTEX g_session_mutex = 0;
  99 static int no_sessions = 0;
 100 static int no_session_total = 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     if (delta > 0)
 110         no_session_total += delta;
 111     sessions = no_sessions;
 112     yaz_mutex_leave(g_session_mutex);
 113     yaz_log(YLOG_DEBUG, "%s sessions=%d", delta == 0 ? "" : (delta > 0 ? "INC" : "DEC"), no_sessions);
 114     return sessions;
 115 }
 116
 117 int sessions_count(void)
 118 {
 119     return session_use(0);
 120 }
 121
 122 int session_count_total(void)
 123 {
 124     int total = 0;
 125     if (!g_session_mutex)
 126         return 0;
 127     yaz_mutex_enter(g_session_mutex);
 128     total = no_session_total;
 129     yaz_mutex_leave(g_session_mutex);
 130     return total;
 131 }
 132
 133 static void log_xml_doc(xmlDoc *doc)
 134 {
 135     FILE *lf = yaz_log_file();
 136     xmlChar *result = 0;
 137     int len = 0;
 138 #if LIBXML_VERSION >= 20600
 139     xmlDocDumpFormatMemory(doc, &result, &len, 1);
 140 #else
 141     xmlDocDumpMemory(doc, &result, &len);
 142 #endif
 143     if (lf && len)
 144     {
 145         (void) fwrite(result, 1, len, lf);
 146         fprintf(lf, "\n");
 147     }
 148     xmlFree(result);
 149 }
 150
 151 static void session_enter(struct session *s, const char *caller)
 152 {
 153     if (caller)
 154         session_log(s, YLOG_DEBUG, "Session lock by %s", caller);
 155     yaz_mutex_enter(s->session_mutex);
 156 }
 157
 158 static void session_leave(struct session *s, const char *caller)
 159 {
 160     yaz_mutex_leave(s->session_mutex);
 161     if (caller)
 162         session_log(s, YLOG_DEBUG, "Session unlock by %s", caller);
 163 }
 164
 165 static void session_normalize_facet(struct session *s, const char *type,
 166                                     const char *value,
 167                                     WRBUF display_wrbuf,
 168                                     WRBUF facet_wrbuf)
 169 {
 170     struct conf_service *service = s->service;
 171     pp2_charset_token_t prt;
 172     const char *facet_component;
 173     int i;
 174     const char *icu_chain_id = 0;
 175
 176     for (i = 0; i < service->num_metadata; i++)
 177         if (!strcmp((service->metadata + i)->name, type))
 178             icu_chain_id = (service->metadata + i)->facetrule;
 179     if (!icu_chain_id)
 180         icu_chain_id = "facet";
 181     prt = pp2_charset_token_create(service->charsets, icu_chain_id);
 182     if (!prt)
 183     {
 184         yaz_log(YLOG_FATAL, "Unknown ICU chain '%s' for facet of type '%s'",
 185                 icu_chain_id, type);
 186         wrbuf_destroy(facet_wrbuf);
 187         wrbuf_destroy(display_wrbuf);
 188         return;
 189     }
 190     pp2_charset_token_first(prt, value, 0);
 191     while ((facet_component = pp2_charset_token_next(prt)))
 192     {
 193         const char *display_component;
 194         if (*facet_component)
 195         {
 196             if (wrbuf_len(facet_wrbuf))
 197                 wrbuf_puts(facet_wrbuf, " ");
 198             wrbuf_puts(facet_wrbuf, facet_component);
 199         }
 200         display_component = pp2_get_display(prt);
 201         if (display_component)
 202         {
 203             if (wrbuf_len(display_wrbuf))
 204                 wrbuf_puts(display_wrbuf, " ");
 205             wrbuf_puts(display_wrbuf, display_component);
 206         }
 207     }
 208     pp2_charset_token_destroy(prt);
 209 }
 210
 211 void add_facet(struct session *s, const char *type, const char *value, int count)
 212 {
 213     WRBUF facet_wrbuf = wrbuf_alloc();
 214     WRBUF display_wrbuf = wrbuf_alloc();
 215
 216     session_normalize_facet(s, type, value, display_wrbuf, facet_wrbuf);
 217
 218     if (wrbuf_len(facet_wrbuf))
 219     {
 220         int i;
 221         for (i = 0; i < s->num_termlists; i++)
 222             if (!strcmp(s->termlists[i].name, type))
 223                 break;
 224         if (i == s->num_termlists)
 225         {
 226             if (i == SESSION_MAX_TERMLISTS)
 227             {
 228                 session_log(s, YLOG_FATAL, "Too many termlists");
 229                 wrbuf_destroy(facet_wrbuf);
 230                 wrbuf_destroy(display_wrbuf);
 231                 return;
 232             }
 233
 234             s->termlists[i].name = nmem_strdup(s->nmem, type);
 235             s->termlists[i].termlist = termlist_create(s->nmem);
 236             s->num_termlists = i + 1;
 237         }
 238
 239 #if 0
 240         session_log(s, YLOG_LOG, "Facets for %s: %s norm:%s (%d)", type, value, wrbuf_cstr(facet_wrbuf), count);
 241 #endif
 242         termlist_insert(s->termlists[i].termlist, wrbuf_cstr(display_wrbuf),
 243                         wrbuf_cstr(facet_wrbuf), count);
 244     }
 245     wrbuf_destroy(facet_wrbuf);
 246     wrbuf_destroy(display_wrbuf);
 247 }
 248
 249 static xmlDoc *record_to_xml(struct session *se,
 250                              struct session_database *sdb, const char *rec)
 251 {
 252     struct database *db = sdb->database;
 253     xmlDoc *rdoc = 0;
 254
 255     rdoc = xmlParseMemory(rec, strlen(rec));
 256
 257     if (!rdoc)
 258     {
 259         session_log(se, YLOG_WARN, "Non-wellformed XML");
 260         return 0;
 261     }
 262
 263     if (global_parameters.dump_records)
 264     {
 265         session_log(se, YLOG_LOG, "Un-normalized record from %s", db->id);
 266         log_xml_doc(rdoc);
 267     }
 268
 269     return rdoc;
 270 }
 271
 272 #define MAX_XSLT_ARGS 16
 273
 274 // Add static values from session database settings if applicable
 275 static void insert_settings_parameters(struct session_database *sdb,
 276                                        struct conf_service *service,
 277                                        char **parms,
 278                                        NMEM nmem)
 279 {
 280     int i;
 281     int nparms = 0;
 282     int offset = 0;
 283
 284     for (i = 0; i < service->num_metadata; i++)
 285     {
 286         struct conf_metadata *md = &service->metadata[i];
 287         int setting;
 288
 289         if (md->setting == Metadata_setting_parameter &&
 290             (setting = settings_lookup_offset(service, md->name)) >= 0)
 291         {
 292             const char *val = session_setting_oneval(sdb, setting);
 293             if (val && nparms < MAX_XSLT_ARGS)
 294             {
 295                 char *buf;
 296                 int len = strlen(val);
 297                 buf = nmem_malloc(nmem, len + 3);
 298                 buf[0] = '\'';
 299                 strcpy(buf + 1, val);
 300                 buf[len+1] = '\'';
 301                 buf[len+2] = '\0';
 302                 parms[offset++] = md->name;
 303                 parms[offset++] = buf;
 304                 nparms++;
 305             }
 306         }
 307     }
 308     parms[offset] = 0;
 309 }
 310
 311 // Add static values from session database settings if applicable
 312 static void insert_settings_values(struct session_database *sdb, xmlDoc *doc,
 313     struct conf_service *service)
 314 {
 315     int i;
 316
 317     for (i = 0; i < service->num_metadata; i++)
 318     {
 319         struct conf_metadata *md = &service->metadata[i];
 320         int offset;
 321
 322         if (md->setting == Metadata_setting_postproc &&
 323             (offset = settings_lookup_offset(service, md->name)) >= 0)
 324         {
 325             const char *val = session_setting_oneval(sdb, offset);
 326             if (val)
 327             {
 328                 xmlNode *r = xmlDocGetRootElement(doc);
 329                 xmlNode *n = xmlNewTextChild(r, 0, (xmlChar *) "metadata",
 330                                              (xmlChar *) val);
 331                 xmlSetProp(n, (xmlChar *) "type", (xmlChar *) md->name);
 332             }
 333         }
 334     }
 335 }
 336
 337 static xmlDoc *normalize_record(struct session *se,
 338                                 struct session_database *sdb,
 339                                 struct conf_service *service,
 340                                 const char *rec, NMEM nmem)
 341 {
 342     xmlDoc *rdoc = record_to_xml(se, sdb, rec);
 343
 344     if (rdoc)
 345     {
 346         char *parms[MAX_XSLT_ARGS*2+1];
 347
 348         insert_settings_parameters(sdb, service, parms, nmem);
 349
 350         if (normalize_record_transform(sdb->map, &rdoc, (const char **)parms))
 351         {
 352             session_log(se, YLOG_WARN, "Normalize failed");
 353         }
 354         else
 355         {
 356             insert_settings_values(sdb, rdoc, service);
 357
 358             if (global_parameters.dump_records)
 359             {
 360                 session_log(se, YLOG_LOG, "Normalized record from %s",
 361                             sdb->database->id);
 362                 log_xml_doc(rdoc);
 363             }
 364         }
 365     }
 366     return rdoc;
 367 }
 368
 369 void session_settings_dump(struct session *se,
 370                            struct session_database *db,
 371                            WRBUF w)
 372 {
 373     if (db->settings)
 374     {
 375         int i, num = db->num_settings;
 376         for (i = 0; i < num; i++)
 377         {
 378             struct setting *s = db->settings[i];
 379             for (;s ; s = s->next)
 380             {
 381                 wrbuf_puts(w, "<set name=\"");
 382                 wrbuf_xmlputs(w, s->name);
 383                 wrbuf_puts(w, "\" value=\"");
 384                 wrbuf_xmlputs(w, s->value);
 385                 wrbuf_puts(w, "\"/>");
 386             }
 387             if (db->settings[i])
 388                 wrbuf_puts(w, "\n");
 389         }
 390     }
 391 }
 392
 393 // Retrieve first defined value for 'name' for given database.
 394 // Will be extended to take into account user associated with session
 395 const char *session_setting_oneval(struct session_database *db, int offset)
 396 {
 397     if (offset >= db->num_settings || !db->settings[offset])
 398         return "";
 399     return db->settings[offset]->value;
 400 }
 401
 402 // Prepare XSLT stylesheets for record normalization
 403 // Structures are allocated on the session_wide nmem to avoid having
 404 // to recompute this for every search. This would lead
 405 // to leaking if a single session was to repeatedly change the PZ_XSLT
 406 // setting. However, this is not a realistic use scenario.
 407 static int prepare_map(struct session *se, struct session_database *sdb)
 408 {
 409     if (sdb->settings && !sdb->map)
 410     {
 411         const char *s;
 412
 413         if (sdb->settings[PZ_XSLT] &&
 414             (s = session_setting_oneval(sdb, PZ_XSLT)))
 415         {
 416             char auto_stylesheet[256];
 417
 418             if (!strcmp(s, "auto"))
 419             {
 420                 const char *request_syntax = session_setting_oneval(
 421                     sdb, PZ_REQUESTSYNTAX);
 422                 if (request_syntax)
 423                 {
 424                     char *cp;
 425                     yaz_snprintf(auto_stylesheet, sizeof(auto_stylesheet),
 426                                  "%s.xsl", request_syntax);
 427                     for (cp = auto_stylesheet; *cp; cp++)
 428                     {
 429                         /* deliberately only consider ASCII */
 430                         if (*cp > 32 && *cp < 127)
 431                             *cp = tolower(*cp);
 432                     }
 433                     s = auto_stylesheet;
 434                 }
 435                 else
 436                 {
 437                     session_log(se, YLOG_WARN,
 438                                 "No pz:requestsyntax for auto stylesheet");
 439                 }
 440             }
 441             sdb->map = normalize_cache_get(se->normalize_cache,
 442                                            se->service, s);
 443             if (!sdb->map)
 444                 return -1;
 445         }
 446     }
 447     return 0;
 448 }
 449
 450 // called if watch should be removed because http_channel is to be destroyed
 451 static void session_watch_cancel(void *data, struct http_channel *c,
 452                                  void *data2)
 453 {
 454     struct session_watchentry *ent = data;
 455
 456     ent->fun = 0;
 457     ent->data = 0;
 458     ent->obs = 0;
 459 }
 460
 461 // set watch. Returns 0=OK, -1 if watch is already set
 462 int session_set_watch(struct session *s, int what,
 463                       session_watchfun fun, void *data,
 464                       struct http_channel *chan)
 465 {
 466     int ret;
 467     session_enter(s, "session_set_watch");
 468     if (s->watchlist[what].fun)
 469         ret = -1;
 470     else
 471     {
 472
 473         s->watchlist[what].fun = fun;
 474         s->watchlist[what].data = data;
 475         s->watchlist[what].obs = http_add_observer(chan, &s->watchlist[what],
 476                                                    session_watch_cancel);
 477         ret = 0;
 478     }
 479     session_leave(s, "session_set_watch");
 480     return ret;
 481 }
 482
 483 void session_alert_watch(struct session *s, int what)
 484 {
 485     assert(s);
 486     session_enter(s, "session_alert_watch");
 487     if (s->watchlist[what].fun)
 488     {
 489         /* our watch is no longer associated with http_channel */
 490         void *data;
 491         session_watchfun fun;
 492
 493         http_remove_observer(s->watchlist[what].obs);
 494         fun  = s->watchlist[what].fun;
 495         data = s->watchlist[what].data;
 496
 497         /* reset watch before fun is invoked - in case fun wants to set
 498            it again */
 499         s->watchlist[what].fun = 0;
 500         s->watchlist[what].data = 0;
 501         s->watchlist[what].obs = 0;
 502
 503         session_leave(s, "session_alert_watch");
 504         session_log(s, YLOG_DEBUG,
 505                     "Alert Watch: %d calling function: %p", what, fun);
 506         fun(data);
 507     }
 508     else
 509         session_leave(s,"session_alert_watch");
 510 }
 511
 512 //callback for grep_databases
 513 static void select_targets_callback(struct session *se,
 514                                     struct session_database *db)
 515 {
 516     struct client *cl;
 517     struct client_list *l;
 518
 519     for (l = se->clients_cached; l; l = l->next)
 520         if (client_get_database(l->client) == db)
 521             break;
 522
 523     if (l)
 524         cl = l->client;
 525     else
 526     {
 527         cl = client_create(db->database->id);
 528         client_set_database(cl, db);
 529
 530         l = xmalloc(sizeof(*l));
 531         l->client = cl;
 532         l->next = se->clients_cached;
 533         se->clients_cached = l;
 534     }
 535     /* set session always. If may be 0 if client is not active */
 536     client_set_session(cl, se);
 537
 538     l = xmalloc(sizeof(*l));
 539     l->client = cl;
 540     l->next = se->clients_active;
 541     se->clients_active = l;
 542 }
 543
 544 static void session_reset_active_clients(struct session *se,
 545                                          struct client_list *new_list)
 546 {
 547     struct client_list *l;
 548
 549     session_enter(se, "session_reset_active_clients");
 550     l = se->clients_active;
 551     se->clients_active = new_list;
 552     session_leave(se, "session_reset_active_clients");
 553
 554     while (l)
 555     {
 556         struct client_list *l_next = l->next;
 557
 558         client_lock(l->client);
 559         client_set_session(l->client, 0); /* mark client inactive */
 560         client_unlock(l->client);
 561
 562         xfree(l);
 563         l = l_next;
 564     }
 565 }
 566
 567 static void session_remove_cached_clients(struct session *se)
 568 {
 569     struct client_list *l;
 570
 571     session_reset_active_clients(se, 0);
 572
 573     session_enter(se, "session_remove_cached_clients");
 574     l = se->clients_cached;
 575     se->clients_cached = 0;
 576     session_leave(se, "session_remove_cached_clients");
 577
 578     while (l)
 579     {
 580         struct client_list *l_next = l->next;
 581         client_lock(l->client);
 582         client_set_session(l->client, 0);
 583         client_set_database(l->client, 0);
 584         client_unlock(l->client);
 585         client_destroy(l->client);
 586         xfree(l);
 587         l = l_next;
 588     }
 589 }
 590
 591 // Associates a set of clients with a session;
 592 // Note: Session-databases represent databases with per-session
 593 // setting overrides
 594 static int select_targets(struct session *se, const char *filter)
 595 {
 596     return session_grep_databases(se, filter, select_targets_callback);
 597 }
 598
 599 int session_active_clients(struct session *s)
 600 {
 601     struct client_list *l;
 602     int res = 0;
 603
 604     for (l = s->clients_active; l; l = l->next)
 605         if (client_is_active(l->client))
 606             res++;
 607
 608     return res;
 609 }
 610
 611 int session_is_preferred_clients_ready(struct session *s)
 612 {
 613     struct client_list *l;
 614     int res = 0;
 615
 616     for (l = s->clients_active; l; l = l->next)
 617         if (client_is_active_preferred(l->client))
 618             res++;
 619     session_log(s, YLOG_DEBUG, "Has %d active preferred clients.", res);
 620     return res == 0;
 621 }
 622
 623 static void session_clear_set(struct session *se, struct reclist_sortparms *sp)
 624 {
 625     reclist_destroy(se->reclist);
 626     se->reclist = 0;
 627     if (nmem_total(se->nmem))
 628         session_log(se, YLOG_DEBUG, "NMEN operation usage %zd",
 629                     nmem_total(se->nmem));
 630     nmem_reset(se->nmem);
 631     se->total_records = se->total_merged = 0;
 632     se->num_termlists = 0;
 633
 634     /* reset list of sorted results and clear to relevance search */
 635     se->sorted_results = nmem_malloc(se->nmem, sizeof(*se->sorted_results));
 636     se->sorted_results->name = nmem_strdup(se->nmem, sp->name);
 637     se->sorted_results->increasing = sp->increasing;
 638     se->sorted_results->type = sp->type;
 639     se->sorted_results->next = 0;
 640
 641     session_log(se, YLOG_DEBUG, "clear_set session_sort: field=%s increasing=%d type=%d configured",
 642             sp->name, sp->increasing, sp->type);
 643
 644     se->reclist = reclist_create(se->nmem);
 645 }
 646
 647 static void session_sort_unlocked(struct session *se, struct reclist_sortparms *sp)
 648 {
 649     struct reclist_sortparms *sr;
 650     struct client_list *l;
 651     const char *field = sp->name;
 652     int increasing = sp->increasing;
 653     int type  = sp->type;
 654     int clients_research = 0;
 655
 656     yaz_log(YLOG_LOG, "session_sort field=%s increasing=%d type=%d", field, increasing, type);
 657     /* see if we already have sorted for this criteria */
 658     for (sr = se->sorted_results; sr; sr = sr->next)
 659     {
 660         if (!reclist_sortparms_cmp(sr,sp))
 661             break;
 662     }
 663     if (sr)
 664     {
 665         session_log(se, YLOG_DEBUG, "search_sort: field=%s increasing=%d type=%d already fetched",
 666                     field, increasing, type);
 667         return;
 668     }
 669     session_log(se, YLOG_DEBUG, "search_sort: field=%s increasing=%d type=%d must fetch",
 670                     field, increasing, type);
 671
 672     // We need to reset reclist on every sort that changes the records, not just for position
 673     // So if just one client requires new searching, we need to clear set.
 674     // Ask each of the client if sorting requires re-search due to native sort
 675     // If it does it will require us to
 676     for (l = se->clients_active; l; l = l->next)
 677     {
 678         struct client *cl = l->client;
 679         // Assume no re-search is required.
 680         client_parse_init(cl, 1);
 681         clients_research += client_parse_sort(cl, sp);
 682     }
 683     if (clients_research) {
 684         yaz_log(YLOG_DEBUG, "Reset results due to %d clients researching", clients_research);
 685         session_clear_set(se, sp);
 686     }
 687     else {
 688         // A new sorting based on same record set
 689         sr = nmem_malloc(se->nmem, sizeof(*sr));
 690         sr->name = nmem_strdup(se->nmem, field);
 691         sr->increasing = increasing;
 692         sr->type = type;
 693         sr->next = se->sorted_results;
 694         se->sorted_results = sr;
 695         session_log(se, YLOG_DEBUG, "No research/ingesting done");
 696         return ;
 697     }
 698     session_log(se, YLOG_DEBUG, "Re- search/ingesting for clients due to change in sort order");
 699
 700     for (l = se->clients_active; l; l = l->next)
 701     {
 702         struct client *cl = l->client;
 703         if (client_get_state(cl) == Client_Connecting ||
 704             client_get_state(cl) == Client_Idle ||
 705             client_get_state(cl) == Client_Working) {
 706             client_start_search(cl);
 707         }
 708         else {
 709             yaz_log(YLOG_DEBUG, "Client %s: No re-start/ingest in show. Wrong client state: %d",
 710                         client_get_id(cl), client_get_state(cl));
 711         }
 712
 713     }
 714 }
 715
 716 void session_sort(struct session *se, struct reclist_sortparms *sp) {
 717     //session_enter(se, "session_sort");
 718     session_sort_unlocked(se, sp);
 719     //session_leave(se, "session_sort");
 720 }
 721
 722
 723 enum pazpar2_error_code session_search(struct session *se,
 724                                        const char *query,
 725                                        const char *startrecs,
 726                                        const char *maxrecs,
 727                                        const char *filter,
 728                                        const char *limit,
 729                                        const char **addinfo,
 730                                        struct reclist_sortparms *sp)
 731 {
 732     int live_channels = 0;
 733     int no_working = 0;
 734     int no_failed_query = 0;
 735     int no_failed_limit = 0;
 736     struct client_list *l, *l0;
 737     int same_sort_order = 0;
 738
 739     session_log(se, YLOG_DEBUG, "Search");
 740
 741     *addinfo = 0;
 742
 743     if (se->settings_modified) {
 744         session_remove_cached_clients(se);
 745     }
 746     else
 747         session_reset_active_clients(se, 0);
 748
 749     session_enter(se, "session_search");
 750     se->settings_modified = 0;
 751
 752     if (se->sorted_results) {
 753         if (!reclist_sortparms_cmp(se->sorted_results, sp))
 754             same_sort_order = 1;
 755     }
 756     session_clear_set(se, sp);
 757     relevance_destroy(&se->relevance);
 758
 759     live_channels = select_targets(se, filter);
 760     if (!live_channels)
 761     {
 762         session_leave(se, "session_search");
 763         return PAZPAR2_NO_TARGETS;
 764     }
 765
 766     facet_limits_destroy(se->facet_limits);
 767     se->facet_limits = facet_limits_create(limit);
 768     if (!se->facet_limits)
 769     {
 770         *addinfo = "limit";
 771         session_leave(se, "session_search");
 772         return PAZPAR2_MALFORMED_PARAMETER_VALUE;
 773     }
 774
 775     l0 = se->clients_active;
 776     se->clients_active = 0;
 777     session_leave(se, "session_search");
 778
 779     for (l = l0; l; l = l->next)
 780     {
 781         int parse_ret;
 782         struct client *cl = l->client;
 783         client_parse_init(cl, 1);
 784         if (prepare_map(se, client_get_database(cl)) < 0)
 785             continue;
 786
 787         parse_ret = client_parse_query(cl, query, se->facet_limits);
 788         if (parse_ret == -1)
 789             no_failed_query++;
 790         else if (parse_ret == -2)
 791             no_failed_limit++;
 792         else if (parse_ret < 0)
 793             no_working++; /* other error, such as bad CCL map */
 794         else
 795         {
 796             client_parse_range(cl, startrecs, maxrecs);
 797             client_parse_sort(cl, sp);
 798             client_start_search(cl);
 799             no_working++;
 800         }
 801     }
 802     session_reset_active_clients(se, l0);
 803
 804     if (no_working == 0)
 805     {
 806         if (no_failed_query > 0)
 807         {
 808             *addinfo = "query";
 809             return PAZPAR2_MALFORMED_PARAMETER_VALUE;
 810         }
 811         else if (no_failed_limit > 0)
 812         {
 813             *addinfo = "limit";
 814             return PAZPAR2_MALFORMED_PARAMETER_VALUE;
 815         }
 816         else
 817             return PAZPAR2_NO_TARGETS;
 818     }
 819     session_log(se, YLOG_LOG, "session_start_search done");
 820     return PAZPAR2_NO_ERROR;
 821 }
 822
 823 // Creates a new session_database object for a database
 824 static void session_init_databases_fun(void *context, struct database *db)
 825 {
 826     struct session *se = (struct session *) context;
 827     struct session_database *new = nmem_malloc(se->session_nmem, sizeof(*new));
 828     int i;
 829
 830     new->database = db;
 831
 832     new->map = 0;
 833     assert(db->settings);
 834     new->settings = nmem_malloc(se->session_nmem,
 835                                 sizeof(struct settings *) * db->num_settings);
 836     new->num_settings = db->num_settings;
 837     for (i = 0; i < db->num_settings; i++)
 838     {
 839         struct setting *setting = db->settings[i];
 840         new->settings[i] = setting;
 841     }
 842     new->next = se->databases;
 843     se->databases = new;
 844 }
 845
 846 // Doesn't free memory associated with sdb -- nmem takes care of that
 847 static void session_database_destroy(struct session_database *sdb)
 848 {
 849     sdb->map = 0;
 850 }
 851
 852 // Initialize session_database list -- this represents this session's view
 853 // of the database list -- subject to modification by the settings ws command
 854 void session_init_databases(struct session *se)
 855 {
 856     se->databases = 0;
 857     predef_grep_databases(se, se->service, session_init_databases_fun);
 858 }
 859
 860 // Probably session_init_databases_fun should be refactored instead of
 861 // called here.
 862 static struct session_database *load_session_database(struct session *se,
 863                                                       char *id)
 864 {
 865     struct database *db = new_database_inherit_settings(id, se->session_nmem, se->service->settings);
 866     session_init_databases_fun((void*) se, db);
 867
 868     // New sdb is head of se->databases list
 869     return se->databases;
 870 }
 871
 872 // Find an existing session database. If not found, load it
 873 static struct session_database *find_session_database(struct session *se,
 874                                                       char *id)
 875 {
 876     struct session_database *sdb;
 877
 878     for (sdb = se->databases; sdb; sdb = sdb->next)
 879         if (!strcmp(sdb->database->id, id))
 880             return sdb;
 881     return load_session_database(se, id);
 882 }
 883
 884 // Apply a session override to a database
 885 void session_apply_setting(struct session *se, char *dbname, char *setting,
 886                            char *value)
 887 {
 888     struct session_database *sdb = find_session_database(se, dbname);
 889     struct conf_service *service = se->service;
 890     struct setting *new = nmem_malloc(se->session_nmem, sizeof(*new));
 891     int offset = settings_create_offset(service, setting);
 892
 893     expand_settings_array(&sdb->settings, &sdb->num_settings, offset,
 894                           se->session_nmem);
 895     new->precedence = 0;
 896     new->target = dbname;
 897     new->name = setting;
 898     new->value = value;
 899     new->next = sdb->settings[offset];
 900     sdb->settings[offset] = new;
 901
 902     se->settings_modified = 1;
 903
 904     // Force later recompute of settings-driven data structures
 905     // (happens when a search starts and client connections are prepared)
 906     switch (offset)
 907     {
 908     case PZ_XSLT:
 909         if (sdb->map)
 910         {
 911             sdb->map = 0;
 912         }
 913         break;
 914     }
 915 }
 916
 917 void session_destroy(struct session *se)
 918 {
 919     struct session_database *sdb;
 920     session_log(se, YLOG_DEBUG, "Destroying");
 921     session_use(-1);
 922     session_remove_cached_clients(se);
 923
 924     for (sdb = se->databases; sdb; sdb = sdb->next)
 925         session_database_destroy(sdb);
 926     normalize_cache_destroy(se->normalize_cache);
 927     relevance_destroy(&se->relevance);
 928     reclist_destroy(se->reclist);
 929     if (nmem_total(se->nmem))
 930         session_log(se, YLOG_DEBUG, "NMEN operation usage %zd", nmem_total(se->nmem));
 931     if (nmem_total(se->session_nmem))
 932         session_log(se, YLOG_DEBUG, "NMEN session usage %zd", nmem_total(se->session_nmem));
 933     facet_limits_destroy(se->facet_limits);
 934     nmem_destroy(se->nmem);
 935     service_destroy(se->service);
 936     yaz_mutex_destroy(&se->session_mutex);
 937 }
 938
 939 size_t session_get_memory_status(struct session *session) {
 940     size_t session_nmem;
 941     if (session == 0)
 942         return 0;
 943     session_enter(session, "session_get_memory_status");
 944     session_nmem = nmem_total(session->nmem);
 945     session_leave(session, "session_get_memory_status");
 946     return session_nmem;
 947 }
 948
 949
 950 struct session *new_session(NMEM nmem, struct conf_service *service,
 951                             unsigned session_id)
 952 {
 953     int i;
 954     struct session *session = nmem_malloc(nmem, sizeof(*session));
 955
 956     char tmp_str[50];
 957
 958     sprintf(tmp_str, "session#%u", session_id);
 959
 960     session->session_id = session_id;
 961     session_log(session, YLOG_DEBUG, "New");
 962     session->service = service;
 963     session->relevance = 0;
 964     session->total_records = 0;
 965     session->number_of_warnings_unknown_elements = 0;
 966     session->number_of_warnings_unknown_metadata = 0;
 967     session->num_termlists = 0;
 968     session->reclist = 0;
 969     session->clients_active = 0;
 970     session->clients_cached = 0;
 971     session->settings_modified = 0;
 972     session->session_nmem = nmem;
 973     session->nmem = nmem_create();
 974     session->databases = 0;
 975     session->sorted_results = 0;
 976     session->facet_limits = 0;
 977
 978     for (i = 0; i <= SESSION_WATCH_MAX; i++)
 979     {
 980         session->watchlist[i].data = 0;
 981         session->watchlist[i].fun = 0;
 982     }
 983     session->normalize_cache = normalize_cache_create();
 984     session->session_mutex = 0;
 985     pazpar2_mutex_create(&session->session_mutex, tmp_str);
 986     session_use(1);
 987     return session;
 988 }
 989
 990 const char * client_get_suggestions_xml(struct client *cl, WRBUF wrbuf);
 991
 992 static struct hitsbytarget *hitsbytarget_nb(struct session *se,
 993                                             int *count, NMEM nmem)
 994 {
 995     struct hitsbytarget *res = 0;
 996     struct client_list *l;
 997     size_t sz = 0;
 998
 999     for (l = se->clients_active; l; l = l->next)
1000         sz++;
1001
1002     res = nmem_malloc(nmem, sizeof(*res) * sz);
1003     *count = 0;
1004     for (l = se->clients_active; l; l = l->next)
1005     {
1006         struct client *cl = l->client;
1007         WRBUF w = wrbuf_alloc();
1008         const char *name = session_setting_oneval(client_get_database(cl),
1009                                                   PZ_NAME);
1010
1011         res[*count].id = client_get_id(cl);
1012         res[*count].name = *name ? name : "Unknown";
1013         res[*count].hits = client_get_hits(cl);
1014         res[*count].approximation = client_get_approximation(cl);
1015         res[*count].records = client_get_num_records(cl);
1016         res[*count].filtered = client_get_num_records_filtered(cl);
1017         res[*count].diagnostic =
1018             client_get_diagnostic(cl, &res[*count].message,
1019                                   &res[*count].addinfo);
1020         res[*count].state = client_get_state_str(cl);
1021         res[*count].connected  = client_get_connection(cl) ? 1 : 0;
1022         session_settings_dump(se, client_get_database(cl), w);
1023         res[*count].settings_xml = nmem_strdup(nmem, wrbuf_cstr(w));
1024         wrbuf_rewind(w);
1025         wrbuf_puts(w, "");
1026         res[*count].suggestions_xml = nmem_strdup(nmem, client_get_suggestions_xml(cl, w));
1027         wrbuf_destroy(w);
1028         (*count)++;
1029     }
1030     return res;
1031 }
1032
1033 struct hitsbytarget *get_hitsbytarget(struct session *se, int *count, NMEM nmem)
1034 {
1035     struct hitsbytarget *p;
1036     session_enter(se, "get_hitsbytarget");
1037     p = hitsbytarget_nb(se, count, nmem);
1038     session_leave(se, "get_hitsbytarget");
1039     return p;
1040 }
1041
1042 // Compares two hitsbytarget nodes by hitcount
1043 static int cmp_ht(const void *p1, const void *p2)
1044 {
1045     const struct hitsbytarget *h1 = p1;
1046     const struct hitsbytarget *h2 = p2;
1047     return h2->hits - h1->hits;
1048 }
1049
1050 // Compares two hitsbytarget nodes by hitcount
1051 static int cmp_ht_approx(const void *p1, const void *p2)
1052 {
1053     const struct hitsbytarget *h1 = p1;
1054     const struct hitsbytarget *h2 = p2;
1055     return h2->approximation - h1->approximation;
1056 }
1057
1058 static int targets_termlist_nb(WRBUF wrbuf, struct session *se, int num,
1059                                NMEM nmem, int version)
1060 {
1061     struct hitsbytarget *ht;
1062     int count, i;
1063
1064     ht = hitsbytarget_nb(se, &count, nmem);
1065     if (version >= 2)
1066         qsort(ht, count, sizeof(struct hitsbytarget), cmp_ht_approx);
1067     else
1068         qsort(ht, count, sizeof(struct hitsbytarget), cmp_ht);
1069     for (i = 0; i < count && i < num && ht[i].hits > 0; i++)
1070     {
1071
1072         // do only print terms which have display names
1073
1074         wrbuf_puts(wrbuf, "<term>\n");
1075
1076         wrbuf_puts(wrbuf, "<id>");
1077         wrbuf_xmlputs(wrbuf, ht[i].id);
1078         wrbuf_puts(wrbuf, "</id>\n");
1079
1080         wrbuf_puts(wrbuf, "<name>");
1081         if (!ht[i].name || !ht[i].name[0])
1082             wrbuf_xmlputs(wrbuf, "NO TARGET NAME");
1083         else
1084             wrbuf_xmlputs(wrbuf, ht[i].name);
1085         wrbuf_puts(wrbuf, "</name>\n");
1086
1087         wrbuf_printf(wrbuf, "<frequency>" ODR_INT_PRINTF "</frequency>\n",
1088                      ht[i].hits);
1089
1090         if (version >= 2) {
1091             // Should not print if we know it isn't a approximation.
1092             wrbuf_printf(wrbuf, "<approximation>" ODR_INT_PRINTF "</approximation>\n", ht[i].approximation);
1093             wrbuf_printf(wrbuf, "<records>%d</records>\n", ht[i].records - ht[i].filtered);
1094             wrbuf_printf(wrbuf, "<filtered>%d</filtered>\n", ht[i].filtered);
1095         }
1096
1097         wrbuf_puts(wrbuf, "<state>");
1098         wrbuf_xmlputs(wrbuf, ht[i].state);
1099         wrbuf_puts(wrbuf, "</state>\n");
1100
1101         wrbuf_printf(wrbuf, "<diagnostic>%d</diagnostic>\n",
1102                      ht[i].diagnostic);
1103         wrbuf_puts(wrbuf, "</term>\n");
1104     }
1105     return count;
1106 }
1107
1108 void perform_termlist(struct http_channel *c, struct session *se,
1109                       const char *name, int num, int version)
1110 {
1111     int i, j;
1112     NMEM nmem_tmp = nmem_create();
1113     char **names;
1114     int num_names = 0;
1115
1116     if (!name)
1117         name = "*";
1118
1119     nmem_strsplit(nmem_tmp, ",", name, &names, &num_names);
1120
1121     session_enter(se, "perform_termlist");
1122
1123     for (j = 0; j < num_names; j++)
1124     {
1125         const char *tname;
1126         int must_generate_empty = 1; /* bug 5350 */
1127
1128         for (i = 0; i < se->num_termlists; i++)
1129         {
1130             tname = se->termlists[i].name;
1131             if (!strcmp(names[j], tname) || !strcmp(names[j], "*"))
1132             {
1133                 struct termlist_score **p = 0;
1134                 int len;
1135
1136                 wrbuf_puts(c->wrbuf, "<list name=\"");
1137                 wrbuf_xmlputs(c->wrbuf, tname);
1138                 wrbuf_puts(c->wrbuf, "\">\n");
1139                 must_generate_empty = 0;
1140
1141                 p = termlist_highscore(se->termlists[i].termlist, &len,
1142                                        nmem_tmp);
1143                 if (p)
1144                 {
1145                     int i;
1146                     for (i = 0; i < len && i < num; i++)
1147                     {
1148                         // prevent sending empty term elements
1149                         if (!p[i]->display_term || !p[i]->display_term[0])
1150                             continue;
1151
1152                         wrbuf_puts(c->wrbuf, "<term>");
1153                         wrbuf_puts(c->wrbuf, "<name>");
1154                         wrbuf_xmlputs(c->wrbuf, p[i]->display_term);
1155                         wrbuf_puts(c->wrbuf, "</name>");
1156
1157                         wrbuf_printf(c->wrbuf,
1158                                      "<frequency>%d</frequency>",
1159                                      p[i]->frequency);
1160                         wrbuf_puts(c->wrbuf, "</term>\n");
1161                     }
1162                 }
1163                 wrbuf_puts(c->wrbuf, "</list>\n");
1164             }
1165         }
1166         tname = "xtargets";
1167         if (!strcmp(names[j], tname) || !strcmp(names[j], "*"))
1168         {
1169             wrbuf_puts(c->wrbuf, "<list name=\"");
1170             wrbuf_xmlputs(c->wrbuf, tname);
1171             wrbuf_puts(c->wrbuf, "\">\n");
1172
1173             targets_termlist_nb(c->wrbuf, se, num, c->nmem, version);
1174             wrbuf_puts(c->wrbuf, "</list>\n");
1175             must_generate_empty = 0;
1176         }
1177         if (must_generate_empty)
1178         {
1179             wrbuf_puts(c->wrbuf, "<list name=\"");
1180             wrbuf_xmlputs(c->wrbuf, names[j]);
1181             wrbuf_puts(c->wrbuf, "\"/>\n");
1182         }
1183     }
1184     session_leave(se, "perform_termlist");
1185     nmem_destroy(nmem_tmp);
1186 }
1187
1188 #ifdef MISSING_HEADERS
1189 void report_nmem_stats(void)
1190 {
1191     size_t in_use, is_free;
1192
1193     nmem_get_memory_in_use(&in_use);
1194     nmem_get_memory_free(&is_free);
1195
1196     yaz_log(YLOG_LOG, "nmem stat: use=%ld free=%ld",
1197             (long) in_use, (long) is_free);
1198 }
1199 #endif
1200
1201 struct record_cluster *show_single_start(struct session *se, const char *id,
1202                                          struct record_cluster **prev_r,
1203                                          struct record_cluster **next_r)
1204 {
1205     struct record_cluster *r = 0;
1206
1207     session_enter(se, "show_single_start");
1208     *prev_r = 0;
1209     *next_r = 0;
1210     if (se->reclist)
1211     {
1212         reclist_limit(se->reclist, se);
1213
1214         reclist_enter(se->reclist);
1215         while ((r = reclist_read_record(se->reclist)))
1216         {
1217             if (!strcmp(r->recid, id))
1218             {
1219                 *next_r = reclist_read_record(se->reclist);
1220                 break;
1221             }
1222             *prev_r = r;
1223         }
1224         reclist_leave(se->reclist);
1225     }
1226     if (!r)
1227         session_leave(se, "show_single_start");
1228     return r;
1229 }
1230
1231 void show_single_stop(struct session *se, struct record_cluster *rec)
1232 {
1233     session_leave(se, "show_single_stop");
1234 }
1235
1236
1237 struct record_cluster **show_range_start(struct session *se,
1238                                          struct reclist_sortparms *sp,
1239                                          int start, int *num, int *total,
1240                                          Odr_int *sumhits, Odr_int *approx_hits)
1241 {
1242     struct record_cluster **recs = 0;
1243     struct reclist_sortparms *spp;
1244     struct client_list *l;
1245     int i;
1246 #if USE_TIMING
1247     yaz_timing_t t = yaz_timing_create();
1248 #endif
1249     session_enter(se, "show_range_start");
1250     *sumhits = 0;
1251     *approx_hits = 0;
1252     *total = 0;
1253     reclist_limit(se->reclist, se);
1254     if (se->relevance)
1255     {
1256         for (spp = sp; spp; spp = spp->next)
1257             if (spp->type == Metadata_sortkey_relevance)
1258             {
1259                 relevance_prepare_read(se->relevance, se->reclist);
1260                 break;
1261             }
1262         for (l = se->clients_active; l; l = l->next) {
1263             *sumhits += client_get_hits(l->client);
1264             *approx_hits += client_get_approximation(l->client);
1265         }
1266     }
1267     reclist_sort(se->reclist, sp);
1268
1269     reclist_enter(se->reclist);
1270     *total = reclist_get_num_records(se->reclist);
1271
1272     for (i = 0; i < start; i++)
1273         if (!reclist_read_record(se->reclist))
1274         {
1275             *num = 0;
1276             break;
1277         }
1278
1279     if (*num > 0)
1280         recs =
1281             nmem_malloc(se->nmem, *num * sizeof(struct record_cluster *));
1282     for (i = 0; i < *num; i++)
1283     {
1284         struct record_cluster *r = reclist_read_record(se->reclist);
1285         if (!r)
1286         {
1287             *num = i;
1288             break;
1289         }
1290         recs[i] = r;
1291     }
1292     reclist_leave(se->reclist);
1293 #if USE_TIMING
1294     yaz_timing_stop(t);
1295     yaz_log(YLOG_LOG, "show %6.5f %3.2f %3.2f",
1296             yaz_timing_get_real(t), yaz_timing_get_user(t),
1297             yaz_timing_get_sys(t));
1298     yaz_timing_destroy(&t);
1299 #endif
1300     return recs;
1301 }
1302
1303 void show_range_stop(struct session *se, struct record_cluster **recs)
1304 {
1305     session_leave(se, "show_range_stop");
1306 }
1307
1308 void statistics(struct session *se, struct statistics *stat)
1309 {
1310     struct client_list *l;
1311     int count = 0;
1312
1313     memset(stat, 0, sizeof(*stat));
1314     stat->num_hits = 0;
1315     for (l = se->clients_active; l; l = l->next)
1316     {
1317         struct client *cl = l->client;
1318         if (!client_get_connection(cl))
1319             stat->num_no_connection++;
1320         stat->num_hits += client_get_hits(cl);
1321         switch (client_get_state(cl))
1322         {
1323         case Client_Connecting: stat->num_connecting++; break;
1324         case Client_Working: stat->num_working++; break;
1325         case Client_Idle: stat->num_idle++; break;
1326         case Client_Failed: stat->num_failed++; break;
1327         case Client_Error: stat->num_error++; break;
1328         default: break;
1329         }
1330         count++;
1331     }
1332     stat->num_records = se->total_records;
1333
1334     stat->num_clients = count;
1335 }
1336
1337 static struct record_metadata *record_metadata_init(
1338     NMEM nmem, const char *value, enum conf_metadata_type type,
1339     struct _xmlAttr *attr)
1340 {
1341     struct record_metadata *rec_md = record_metadata_create(nmem);
1342     struct record_metadata_attr **attrp = &rec_md->attributes;
1343
1344     for (; attr; attr = attr->next)
1345     {
1346         if (attr->children && attr->children->content)
1347         {
1348             if (strcmp((const char *) attr->name, "type"))
1349             {  /* skip the "type" attribute.. Its value is already part of
1350                   the element in output (md-%s) and so repeating it here
1351                   is redundant */
1352                 *attrp = nmem_malloc(nmem, sizeof(**attrp));
1353                 (*attrp)->name =
1354                     nmem_strdup(nmem, (const char *) attr->name);
1355                 (*attrp)->value =
1356                     nmem_strdup(nmem, (const char *) attr->children->content);
1357                 attrp = &(*attrp)->next;
1358             }
1359         }
1360     }
1361     *attrp = 0;
1362
1363     if (type == Metadata_type_generic)
1364     {
1365         char *p = nmem_strdup(nmem, value);
1366
1367         p = normalize7bit_generic(p, " ,/.:([");
1368
1369         rec_md->data.text.disp = p;
1370         rec_md->data.text.sort = 0;
1371     }
1372     else if (type == Metadata_type_year || type == Metadata_type_date)
1373     {
1374         int first, last;
1375         int longdate = 0;
1376
1377         if (type == Metadata_type_date)
1378             longdate = 1;
1379         if (extract7bit_dates((char *) value, &first, &last, longdate) < 0)
1380             return 0;
1381
1382         rec_md->data.number.min = first;
1383         rec_md->data.number.max = last;
1384     }
1385     else
1386         return 0;
1387     return rec_md;
1388 }
1389
1390 static int get_mergekey_from_doc(xmlDoc *doc, xmlNode *root, const char *name,
1391                                  struct conf_service *service, WRBUF norm_wr)
1392 {
1393     xmlNode *n;
1394     int no_found = 0;
1395     for (n = root->children; n; n = n->next)
1396     {
1397         if (n->type != XML_ELEMENT_NODE)
1398             continue;
1399         if (!strcmp((const char *) n->name, "metadata"))
1400         {
1401             xmlChar *type = xmlGetProp(n, (xmlChar *) "type");
1402             if (type == NULL) {
1403                 yaz_log(YLOG_FATAL, "Missing type attribute on metadata element. Skipping!");
1404             }
1405             else if (!strcmp(name, (const char *) type))
1406             {
1407                 xmlChar *value = xmlNodeListGetString(doc, n->children, 1);
1408                 if (value)
1409                 {
1410                     const char *norm_str;
1411                     pp2_charset_token_t prt =
1412                         pp2_charset_token_create(service->charsets, "mergekey");
1413
1414                     pp2_charset_token_first(prt, (const char *) value, 0);
1415                     if (wrbuf_len(norm_wr) > 0)
1416                         wrbuf_puts(norm_wr, " ");
1417                     wrbuf_puts(norm_wr, name);
1418                     while ((norm_str =
1419                             pp2_charset_token_next(prt)))
1420                     {
1421                         if (*norm_str)
1422                         {
1423                             wrbuf_puts(norm_wr, " ");
1424                             wrbuf_puts(norm_wr, norm_str);
1425                         }
1426                     }
1427                     xmlFree(value);
1428                     pp2_charset_token_destroy(prt);
1429                     no_found++;
1430                 }
1431             }
1432             xmlFree(type);
1433         }
1434     }
1435     return no_found;
1436 }
1437
1438 static const char *get_mergekey(xmlDoc *doc, struct client *cl, int record_no,
1439                                 struct conf_service *service, NMEM nmem)
1440 {
1441     char *mergekey_norm = 0;
1442     xmlNode *root = xmlDocGetRootElement(doc);
1443     WRBUF norm_wr = wrbuf_alloc();
1444
1445     /* consider mergekey from XSL first */
1446     xmlChar *mergekey = xmlGetProp(root, (xmlChar *) "mergekey");
1447     if (mergekey)
1448     {
1449         const char *norm_str;
1450         pp2_charset_token_t prt =
1451             pp2_charset_token_create(service->charsets, "mergekey");
1452
1453         pp2_charset_token_first(prt, (const char *) mergekey, 0);
1454         while ((norm_str = pp2_charset_token_next(prt)))
1455         {
1456             if (*norm_str)
1457             {
1458                 if (wrbuf_len(norm_wr))
1459                     wrbuf_puts(norm_wr, " ");
1460                 wrbuf_puts(norm_wr, norm_str);
1461             }
1462         }
1463         pp2_charset_token_destroy(prt);
1464         xmlFree(mergekey);
1465     }
1466     else
1467     {
1468         /* no mergekey defined in XSL. Look for mergekey metadata instead */
1469         int field_id;
1470         for (field_id = 0; field_id < service->num_metadata; field_id++)
1471         {
1472             struct conf_metadata *ser_md = &service->metadata[field_id];
1473             if (ser_md->mergekey != Metadata_mergekey_no)
1474             {
1475                 int r = get_mergekey_from_doc(doc, root, ser_md->name,
1476                                               service, norm_wr);
1477                 if (r == 0 && ser_md->mergekey == Metadata_mergekey_required)
1478                 {
1479                     /* no mergekey on this one and it is required..
1480                        Generate unique key instead */
1481                     wrbuf_rewind(norm_wr);
1482                     break;
1483                 }
1484             }
1485         }
1486     }
1487
1488     /* generate unique key if none is not generated already or is empty */
1489     if (wrbuf_len(norm_wr) == 0)
1490     {
1491         wrbuf_printf(norm_wr, "position: %s-%d",
1492                      client_get_id(cl), record_no);
1493     }
1494     else
1495     {
1496         const char *lead = "content: ";
1497         wrbuf_insert(norm_wr, 0, lead, strlen(lead));
1498     }
1499     if (wrbuf_len(norm_wr) > 0)
1500         mergekey_norm = nmem_strdup(nmem, wrbuf_cstr(norm_wr));
1501     wrbuf_destroy(norm_wr);
1502     return mergekey_norm;
1503 }
1504
1505 /** \brief see if metadata for pz:recordfilter exists
1506     \param root xml root element of normalized record
1507     \param sdb session database for client
1508     \retval 0 if there is no metadata for pz:recordfilter
1509     \retval 1 if there is metadata for pz:recordfilter
1510
1511     If there is no pz:recordfilter defined, this function returns 1
1512     as well.
1513 */
1514
1515 static int check_record_filter(xmlNode *root, struct session_database *sdb)
1516 {
1517     int match = 0;
1518     xmlNode *n;
1519     const char *s;
1520     s = session_setting_oneval(sdb, PZ_RECORDFILTER);
1521
1522     if (!s || !*s)
1523         return 1;
1524
1525     for (n = root->children; n; n = n->next)
1526     {
1527         if (n->type != XML_ELEMENT_NODE)
1528             continue;
1529         if (!strcmp((const char *) n->name, "metadata"))
1530         {
1531             xmlChar *type = xmlGetProp(n, (xmlChar *) "type");
1532             if (type)
1533             {
1534                 size_t len;
1535                 int substring;
1536                 const char *eq;
1537
1538                 if ((eq = strchr(s, '=')))
1539                     substring = 0;
1540                 else if ((eq = strchr(s, '~')))
1541                     substring = 1;
1542                 if (eq)
1543                     len = eq - s;
1544                 else
1545                     len = strlen(s);
1546                 if (len == strlen((const char *)type) &&
1547                     !memcmp((const char *) type, s, len))
1548                 {
1549                     xmlChar *value = xmlNodeGetContent(n);
1550                     if (value && *value)
1551                     {
1552                         if (!eq ||
1553                             (substring && strstr((const char *) value, eq+1)) ||
1554                             (!substring && !strcmp((const char *) value, eq + 1)))
1555                             match = 1;
1556                     }
1557                     xmlFree(value);
1558                 }
1559                 xmlFree(type);
1560             }
1561         }
1562     }
1563     return match;
1564 }
1565
1566
1567 static int ingest_to_cluster(struct client *cl,
1568                              xmlDoc *xdoc,
1569                              xmlNode *root,
1570                              int record_no,
1571                              const char *mergekey_norm);
1572
1573 /** \brief ingest XML record
1574     \param cl client holds the result set for record
1575     \param rec record buffer (0 terminated)
1576     \param record_no record position (1, 2, ..)
1577     \param nmem working NMEM
1578     \retval 0 OK
1579     \retval -1 failure
1580     \retval -2 Filtered
1581 */
1582 int ingest_record(struct client *cl, const char *rec,
1583                   int record_no, NMEM nmem)
1584 {
1585     struct session *se = client_get_session(cl);
1586     int ret = 0;
1587     struct session_database *sdb = client_get_database(cl);
1588     struct conf_service *service = se->service;
1589     xmlDoc *xdoc = normalize_record(se, sdb, service, rec, nmem);
1590     xmlNode *root;
1591     const char *mergekey_norm;
1592
1593     if (!xdoc)
1594         return -1;
1595
1596     root = xmlDocGetRootElement(xdoc);
1597
1598     if (!check_record_filter(root, sdb))
1599     {
1600         session_log(se, YLOG_LOG, "Filtered out record no %d from %s", record_no, sdb->database->id);
1601         xmlFreeDoc(xdoc);
1602         return -2;
1603     }
1604
1605     mergekey_norm = get_mergekey(xdoc, cl, record_no, service, nmem);
1606     if (!mergekey_norm)
1607     {
1608         session_log(se, YLOG_WARN, "Got no mergekey");
1609         xmlFreeDoc(xdoc);
1610         return -1;
1611     }
1612     session_enter(se, "ingest_record");
1613     if (client_get_session(cl) == se)
1614         ret = ingest_to_cluster(cl, xdoc, root, record_no, mergekey_norm);
1615     session_leave(se, "ingest_record");
1616
1617     xmlFreeDoc(xdoc);
1618     return ret;
1619 }
1620
1621 //    struct conf_metadata *ser_md = &service->metadata[md_field_id];
1622 //    struct record_metadata *rec_md = record->metadata[md_field_id];
1623 static int match_metadata_local(struct conf_metadata *ser_md,
1624                                 struct record_metadata *rec_md0,
1625                                 char **values, int num_v)
1626 {
1627     int i;
1628     struct record_metadata *rec_md = rec_md0;
1629     for (i = 0; i < num_v; )
1630     {
1631         if (rec_md)
1632         {
1633             if (ser_md->type == Metadata_type_year
1634                 || ser_md->type == Metadata_type_date)
1635             {
1636                 int y = atoi(values[i]);
1637                 if (y >= rec_md->data.number.min
1638                     && y <= rec_md->data.number.max)
1639                     break;
1640             }
1641             else
1642             {
1643                 yaz_log(YLOG_DEBUG, "cmp: '%s' '%s'", rec_md->data.text.disp, values[i]);
1644                 if (!strcmp(rec_md->data.text.disp, values[i]))
1645                 {
1646                     // Value equals, should not be filtered.
1647                     break;
1648                 }
1649             }
1650             rec_md = rec_md->next;
1651         }
1652         else
1653         {
1654             rec_md = rec_md0;
1655             i++;
1656         }
1657     }
1658     return i < num_v ? 1 : 0;
1659 }
1660
1661 int session_check_cluster_limit(struct session *se, struct record_cluster *rec)
1662 {
1663     int i;
1664     struct conf_service *service = se->service;
1665     int ret = 1;
1666     const char *name;
1667     const char *value;
1668     NMEM nmem_tmp = nmem_create();
1669
1670     for (i = 0; (name = facet_limits_get(se->facet_limits, i, &value)); i++)
1671     {
1672         int j;
1673         for (j = 0; j < service->num_metadata; j++)
1674         {
1675             struct conf_metadata *md = service->metadata + j;
1676             if (!strcmp(md->name, name) && md->limitcluster)
1677             {
1678                 char **values = 0;
1679                 int num = 0;
1680                 int md_field_id =
1681                     conf_service_metadata_field_id(service,
1682                                                    md->limitcluster);
1683
1684                 if (md_field_id < 0)
1685                 {
1686                     ret = 0;
1687                     break;
1688                 }
1689
1690                 nmem_strsplit_escape2(nmem_tmp, "|", value, &values,
1691                                       &num, 1, '\\', 1);
1692
1693                 if (!match_metadata_local(&service->metadata[md_field_id],
1694                                           rec->metadata[md_field_id],
1695                                           values, num))
1696                 {
1697                     ret = 0;
1698                     break;
1699                 }
1700             }
1701         }
1702     }
1703     nmem_destroy(nmem_tmp);
1704     return ret;
1705 }
1706
1707 // Skip record on non-zero
1708 static int check_limit_local(struct client *cl,
1709                              struct record *record,
1710                              int record_no)
1711 {
1712     int skip_record = 0;
1713     struct session *se = client_get_session(cl);
1714     struct conf_service *service = se->service;
1715     NMEM nmem_tmp = nmem_create();
1716     struct session_database *sdb = client_get_database(cl);
1717     int l = 0;
1718     while (!skip_record)
1719     {
1720         int md_field_id;
1721         char **values = 0;
1722         int num_v = 0;
1723         const char *name =
1724             client_get_facet_limit_local(cl, sdb, &l, nmem_tmp,
1725                                          &num_v, &values);
1726         if (!name)
1727             break;
1728
1729         if (!strcmp(name, "*"))
1730         {
1731             for (md_field_id = 0; md_field_id < service->num_metadata;
1732                  md_field_id++)
1733             {
1734                 if (match_metadata_local(
1735                         &service->metadata[md_field_id],
1736                         record->metadata[md_field_id],
1737                         values, num_v))
1738                     break;
1739             }
1740             if (md_field_id == service->num_metadata)
1741                 skip_record = 1;
1742         }
1743         else
1744         {
1745             md_field_id = conf_service_metadata_field_id(service, name);
1746             if (md_field_id < 0)
1747             {
1748                 skip_record = 1;
1749                 break;
1750             }
1751             if (!match_metadata_local(
1752                     &service->metadata[md_field_id],
1753                     record->metadata[md_field_id],
1754                     values, num_v))
1755             {
1756                 skip_record = 1;
1757             }
1758         }
1759     }
1760     nmem_destroy(nmem_tmp);
1761     return skip_record;
1762 }
1763
1764 static int ingest_to_cluster(struct client *cl,
1765                              xmlDoc *xdoc,
1766                              xmlNode *root,
1767                              int record_no,
1768                              const char *mergekey_norm)
1769 {
1770     xmlNode *n;
1771     xmlChar *type = 0;
1772     xmlChar *value = 0;
1773     struct session *se = client_get_session(cl);
1774     struct conf_service *service = se->service;
1775     int term_factor = 1;
1776     struct record_cluster *cluster;
1777     struct record_metadata **metadata0;
1778     struct session_database *sdb = client_get_database(cl);
1779     struct record *record = record_create(se->nmem,
1780                                           service->num_metadata,
1781                                           service->num_sortkeys, cl,
1782                                           record_no);
1783
1784     for (n = root->children; n; n = n->next)
1785     {
1786         if (type)
1787             xmlFree(type);
1788         if (value)
1789             xmlFree(value);
1790         type = value = 0;
1791
1792         if (n->type != XML_ELEMENT_NODE)
1793             continue;
1794         if (!strcmp((const char *) n->name, "metadata"))
1795         {
1796             struct conf_metadata *ser_md = 0;
1797             struct record_metadata **wheretoput = 0;
1798             struct record_metadata *rec_md = 0;
1799             int md_field_id = -1;
1800
1801             type = xmlGetProp(n, (xmlChar *) "type");
1802             value = xmlNodeListGetString(xdoc, n->children, 1);
1803
1804             if (!type || !value || !*value)
1805                 continue;
1806
1807             md_field_id
1808                 = conf_service_metadata_field_id(service, (const char *) type);
1809             if (md_field_id < 0)
1810             {
1811                 if (se->number_of_warnings_unknown_metadata == 0)
1812                 {
1813                     session_log(se, YLOG_WARN,
1814                             "Ignoring unknown metadata element: %s", type);
1815                 }
1816                 se->number_of_warnings_unknown_metadata++;
1817                 continue;
1818             }
1819
1820             ser_md = &service->metadata[md_field_id];
1821
1822             // non-merged metadata
1823             rec_md = record_metadata_init(se->nmem, (const char *) value,
1824                                           ser_md->type, n->properties);
1825             if (!rec_md)
1826             {
1827                 session_log(se, YLOG_WARN, "bad metadata data '%s' "
1828                             "for element '%s'", value, type);
1829                 continue;
1830             }
1831             wheretoput = &record->metadata[md_field_id];
1832             while (*wheretoput)
1833                 wheretoput = &(*wheretoput)->next;
1834             *wheretoput = rec_md;
1835         }
1836     }
1837
1838     if (check_limit_local(cl, record, record_no))
1839     {
1840         session_log(se, YLOG_LOG, "Facet filtered out record no %d from %s",
1841                     record_no, sdb->database->id);
1842         if (type)
1843             xmlFree(type);
1844         if (value)
1845             xmlFree(value);
1846         return -2;
1847     }
1848     cluster = reclist_insert(se->reclist, service, record,
1849                              mergekey_norm, &se->total_merged);
1850     if (!cluster)
1851         return -1;
1852
1853     {
1854         const char *use_term_factor_str =
1855             session_setting_oneval(sdb, PZ_TERMLIST_TERM_FACTOR);
1856         if (use_term_factor_str && use_term_factor_str[0] == '1')
1857         {
1858             int maxrecs = client_get_maxrecs(cl);
1859             int hits = (int) client_get_hits(cl);
1860             term_factor = MAX(hits, maxrecs) /  MAX(1, maxrecs);
1861             assert(term_factor >= 1);
1862             yaz_log(YLOG_DEBUG, "Using term factor: %d (%d / %d)", term_factor, MAX(hits, maxrecs), MAX(1, maxrecs));
1863         }
1864     }
1865
1866     if (global_parameters.dump_records)
1867         session_log(se, YLOG_LOG, "Cluster id %s from %s (#%d)", cluster->recid,
1868                     sdb->database->id, record_no);
1869
1870
1871     relevance_newrec(se->relevance, cluster);
1872
1873     // original metadata, to check if first existence of a field
1874     metadata0 = xmalloc(sizeof(*metadata0) * service->num_metadata);
1875     memcpy(metadata0, cluster->metadata,
1876            sizeof(*metadata0) * service->num_metadata);
1877
1878     // now parsing XML record and adding data to cluster or record metadata
1879     for (n = root->children; n; n = n->next)
1880     {
1881         pp2_charset_token_t prt;
1882         if (type)
1883             xmlFree(type);
1884         if (value)
1885             xmlFree(value);
1886         type = value = 0;
1887
1888         if (n->type != XML_ELEMENT_NODE)
1889             continue;
1890         if (!strcmp((const char *) n->name, "metadata"))
1891         {
1892             struct conf_metadata *ser_md = 0;
1893             struct conf_sortkey *ser_sk = 0;
1894             struct record_metadata **wheretoput = 0;
1895             struct record_metadata *rec_md = 0;
1896             int md_field_id = -1;
1897             int sk_field_id = -1;
1898             const char *rank;
1899             xmlChar *xml_rank;
1900
1901             type = xmlGetProp(n, (xmlChar *) "type");
1902             value = xmlNodeListGetString(xdoc, n->children, 1);
1903
1904             if (!type || !value || !*value)
1905                 continue;
1906
1907             md_field_id
1908                 = conf_service_metadata_field_id(service, (const char *) type);
1909             if (md_field_id < 0)
1910                 continue;
1911
1912             ser_md = &service->metadata[md_field_id];
1913
1914             if (ser_md->sortkey_offset >= 0)
1915             {
1916                 sk_field_id = ser_md->sortkey_offset;
1917                 ser_sk = &service->sortkeys[sk_field_id];
1918             }
1919
1920             // merged metadata
1921             rec_md = record_metadata_init(se->nmem, (const char *) value,
1922                                           ser_md->type, 0);
1923
1924             // see if the field was not in cluster already (from beginning)
1925
1926             if (!rec_md)
1927                 continue;
1928
1929             xml_rank = xmlGetProp(n, (xmlChar *) "rank");
1930             rank = xml_rank ? (const char *) xml_rank : ser_md->rank;
1931
1932             wheretoput = &cluster->metadata[md_field_id];
1933
1934             if (ser_md->merge == Metadata_merge_first)
1935             {
1936                 if (!metadata0[md_field_id])
1937                 {
1938                     while (*wheretoput)
1939                         wheretoput = &(*wheretoput)->next;
1940                     *wheretoput = rec_md;
1941                 }
1942             }
1943             else if (ser_md->merge == Metadata_merge_unique)
1944             {
1945                 while (*wheretoput)
1946                 {
1947                     if (!strcmp((const char *) (*wheretoput)->data.text.disp,
1948                                 rec_md->data.text.disp))
1949                         break;
1950                     wheretoput = &(*wheretoput)->next;
1951                 }
1952                 if (!*wheretoput)
1953                     *wheretoput = rec_md;
1954             }
1955             else if (ser_md->merge == Metadata_merge_longest)
1956             {
1957                 if (!*wheretoput
1958                     || strlen(rec_md->data.text.disp)
1959                     > strlen((*wheretoput)->data.text.disp))
1960                 {
1961                     *wheretoput = rec_md;
1962                     if (ser_sk)
1963                     {
1964                         const char *sort_str = 0;
1965                         int skip_article =
1966                             ser_sk->type == Metadata_sortkey_skiparticle;
1967
1968                         if (!cluster->sortkeys[sk_field_id])
1969                             cluster->sortkeys[sk_field_id] =
1970                                 nmem_malloc(se->nmem,
1971                                             sizeof(union data_types));
1972
1973                         prt =
1974                             pp2_charset_token_create(service->charsets, "sort");
1975
1976                         pp2_charset_token_first(prt, rec_md->data.text.disp,
1977                                                 skip_article);
1978
1979                         pp2_charset_token_next(prt);
1980
1981                         sort_str = pp2_get_sort(prt);
1982
1983                         cluster->sortkeys[sk_field_id]->text.disp =
1984                             rec_md->data.text.disp;
1985                         if (!sort_str)
1986                         {
1987                             sort_str = rec_md->data.text.disp;
1988                             session_log(se, YLOG_WARN,
1989                                     "Could not make sortkey. Bug #1858");
1990                         }
1991                         cluster->sortkeys[sk_field_id]->text.sort =
1992                             nmem_strdup(se->nmem, sort_str);
1993                         pp2_charset_token_destroy(prt);
1994                     }
1995                 }
1996             }
1997             else if (ser_md->merge == Metadata_merge_all)
1998             {
1999                 while (*wheretoput)
2000                     wheretoput = &(*wheretoput)->next;
2001                 *wheretoput = rec_md;
2002             }
2003             else if (ser_md->merge == Metadata_merge_range)
2004             {
2005                 if (!*wheretoput)
2006                 {
2007                     *wheretoput = rec_md;
2008                     if (ser_sk)
2009                         cluster->sortkeys[sk_field_id]
2010                             = &rec_md->data;
2011                 }
2012                 else
2013                 {
2014                     int this_min = rec_md->data.number.min;
2015                     int this_max = rec_md->data.number.max;
2016                     if (this_min < (*wheretoput)->data.number.min)
2017                         (*wheretoput)->data.number.min = this_min;
2018                     if (this_max > (*wheretoput)->data.number.max)
2019                         (*wheretoput)->data.number.max = this_max;
2020                 }
2021             }
2022
2023             // ranking of _all_ fields enabled ...
2024             if (rank)
2025             {
2026                 relevance_countwords(se->relevance, cluster,
2027                                      (char *) value, rank, ser_md->name);
2028             }
2029
2030             // construct facets ... unless the client already has reported them
2031             if (ser_md->termlist && !client_has_facet(cl, (char *) type))
2032             {
2033                 if (ser_md->type == Metadata_type_year)
2034                 {
2035                     char year[64];
2036                     sprintf(year, "%d", rec_md->data.number.max);
2037
2038                     add_facet(se, (char *) type, year, term_factor);
2039                     if (rec_md->data.number.max != rec_md->data.number.min)
2040                     {
2041                         sprintf(year, "%d", rec_md->data.number.min);
2042                         add_facet(se, (char *) type, year, term_factor);
2043                     }
2044                 }
2045                 else
2046                     add_facet(se, (char *) type, (char *) value, term_factor);
2047             }
2048
2049             // cleaning up
2050             if (xml_rank)
2051                 xmlFree(xml_rank);
2052             xmlFree(type);
2053             xmlFree(value);
2054             type = value = 0;
2055         }
2056         else
2057         {
2058             if (se->number_of_warnings_unknown_elements == 0)
2059                 session_log(se, YLOG_WARN,
2060                         "Unexpected element in internal record: %s", n->name);
2061             se->number_of_warnings_unknown_elements++;
2062         }
2063     }
2064     if (type)
2065         xmlFree(type);
2066     if (value)
2067         xmlFree(value);
2068
2069     xfree(metadata0);
2070     relevance_donerecord(se->relevance, cluster);
2071     se->total_records++;
2072
2073     return 0;
2074 }
2075
2076 void session_log(struct session *s, int level, const char *fmt, ...)
2077 {
2078     char buf[1024];
2079     va_list ap;
2080     va_start(ap, fmt);
2081
2082     yaz_vsnprintf(buf, sizeof(buf)-30, fmt, ap);
2083     yaz_log(level, "Session %u: %s", s->session_id, buf);
2084
2085     va_end(ap);
2086 }
2087
2088 /*
2089  * Local variables:
2090  * c-basic-offset: 4
2091  * c-file-style: "Stroustrup"
2092  * indent-tabs-mode: nil
2093  * End:
2094  * vim: shiftwidth=4 tabstop=8 expandtab
2095  */
2096