src/session.c

   1 /* This file is part of Pazpar2.
   2    Copyright (C) 2006-2012 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/gettimeofday.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 TERMLIST_HIGH_SCORE 25
  82
  83 #define MAX_CHUNK 15
  84
  85 #define MAX(a,b) ((a)>(b)?(a):(b))
  86
  87 // Note: Some things in this structure will eventually move to configuration
  88 struct parameters global_parameters =
  89 {
  90     0,   // dump_records
  91     0,   // debug_mode
  92     0,   // predictable sessions
  93 };
  94
  95 struct client_list {
  96     struct client *client;
  97     struct client_list *next;
  98 };
  99
 100 /* session counting (1) , disable client counting (0) */
 101 static YAZ_MUTEX g_session_mutex = 0;
 102 static int no_sessions = 0;
 103 static int no_session_total = 0;
 104
 105 static int session_use(int delta)
 106 {
 107     int sessions;
 108     if (!g_session_mutex)
 109         yaz_mutex_create(&g_session_mutex);
 110     yaz_mutex_enter(g_session_mutex);
 111     no_sessions += delta;
 112     if (delta > 0)
 113         no_session_total += delta;
 114     sessions = no_sessions;
 115     yaz_mutex_leave(g_session_mutex);
 116     yaz_log(YLOG_DEBUG, "%s sessions=%d", delta == 0 ? "" : (delta > 0 ? "INC" : "DEC"), no_sessions);
 117     return sessions;
 118 }
 119
 120 int sessions_count(void)
 121 {
 122     return session_use(0);
 123 }
 124
 125 int session_count_total(void)
 126 {
 127     int total = 0;
 128     if (!g_session_mutex)
 129         return 0;
 130     yaz_mutex_enter(g_session_mutex);
 131     total = no_session_total;
 132     yaz_mutex_leave(g_session_mutex);
 133     return total;
 134 }
 135
 136 static void log_xml_doc(xmlDoc *doc)
 137 {
 138     FILE *lf = yaz_log_file();
 139     xmlChar *result = 0;
 140     int len = 0;
 141 #if LIBXML_VERSION >= 20600
 142     xmlDocDumpFormatMemory(doc, &result, &len, 1);
 143 #else
 144     xmlDocDumpMemory(doc, &result, &len);
 145 #endif
 146     if (lf && len)
 147     {
 148         (void) fwrite(result, 1, len, lf);
 149         fprintf(lf, "\n");
 150     }
 151     xmlFree(result);
 152 }
 153
 154 static void session_enter(struct session *s)
 155 {
 156     yaz_mutex_enter(s->session_mutex);
 157 }
 158
 159 static void session_leave(struct session *s)
 160 {
 161     yaz_mutex_leave(s->session_mutex);
 162 }
 163
 164 static void session_normalize_facet(struct session *s, const char *type,
 165                                     const char *value,
 166                                     WRBUF display_wrbuf,
 167                                     WRBUF facet_wrbuf)
 168 {
 169     struct conf_service *service = s->service;
 170     pp2_charset_token_t prt;
 171     const char *facet_component;
 172     int i;
 173     const char *icu_chain_id = 0;
 174
 175     for (i = 0; i < service->num_metadata; i++)
 176         if (!strcmp((service->metadata + i)->name, type))
 177             icu_chain_id = (service->metadata + i)->facetrule;
 178     if (!icu_chain_id)
 179         icu_chain_id = "facet";
 180     prt = pp2_charset_token_create(service->charsets, icu_chain_id);
 181     if (!prt)
 182     {
 183         yaz_log(YLOG_FATAL, "Unknown ICU chain '%s' for facet of type '%s'",
 184                 icu_chain_id, type);
 185         wrbuf_destroy(facet_wrbuf);
 186         wrbuf_destroy(display_wrbuf);
 187         return;
 188     }
 189     pp2_charset_token_first(prt, value, 0);
 190     while ((facet_component = pp2_charset_token_next(prt)))
 191     {
 192         const char *display_component;
 193         if (*facet_component)
 194         {
 195             if (wrbuf_len(facet_wrbuf))
 196                 wrbuf_puts(facet_wrbuf, " ");
 197             wrbuf_puts(facet_wrbuf, facet_component);
 198         }
 199         display_component = pp2_get_display(prt);
 200         if (display_component)
 201         {
 202             if (wrbuf_len(display_wrbuf))
 203                 wrbuf_puts(display_wrbuf, " ");
 204             wrbuf_puts(display_wrbuf, display_component);
 205         }
 206     }
 207     pp2_charset_token_destroy(prt);
 208 }
 209
 210 void add_facet(struct session *s, const char *type, const char *value, int count)
 211 {
 212     WRBUF facet_wrbuf = wrbuf_alloc();
 213     WRBUF display_wrbuf = wrbuf_alloc();
 214
 215     session_normalize_facet(s, type, value, display_wrbuf, facet_wrbuf);
 216
 217     if (wrbuf_len(facet_wrbuf))
 218     {
 219         int i;
 220         for (i = 0; i < s->num_termlists; i++)
 221             if (!strcmp(s->termlists[i].name, type))
 222                 break;
 223         if (i == s->num_termlists)
 224         {
 225             if (i == SESSION_MAX_TERMLISTS)
 226             {
 227                 session_log(s, YLOG_FATAL, "Too many termlists");
 228                 wrbuf_destroy(facet_wrbuf);
 229                 wrbuf_destroy(display_wrbuf);
 230                 return;
 231             }
 232
 233             s->termlists[i].name = nmem_strdup(s->nmem, type);
 234             s->termlists[i].termlist
 235                 = termlist_create(s->nmem, TERMLIST_HIGH_SCORE);
 236             s->num_termlists = i + 1;
 237         }
 238
 239 #if 0
 240         session_log(s, YLOG_DEBUG, "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_FATAL, "Non-wellformed XML received from %s",
 260                     db->id);
 261         return 0;
 262     }
 263
 264     if (global_parameters.dump_records)
 265     {
 266         session_log(se, YLOG_LOG, "Un-normalized record from %s", db->id);
 267         log_xml_doc(rdoc);
 268     }
 269
 270     return rdoc;
 271 }
 272
 273 #define MAX_XSLT_ARGS 16
 274
 275 // Add static values from session database settings if applicable
 276 static void insert_settings_parameters(struct session_database *sdb,
 277                                        struct conf_service *service,
 278                                        char **parms,
 279                                        NMEM nmem)
 280 {
 281     int i;
 282     int nparms = 0;
 283     int offset = 0;
 284
 285     for (i = 0; i < service->num_metadata; i++)
 286     {
 287         struct conf_metadata *md = &service->metadata[i];
 288         int setting;
 289
 290         if (md->setting == Metadata_setting_parameter &&
 291             (setting = settings_lookup_offset(service, md->name)) >= 0)
 292         {
 293             const char *val = session_setting_oneval(sdb, setting);
 294             if (val && nparms < MAX_XSLT_ARGS)
 295             {
 296                 char *buf;
 297                 int len = strlen(val);
 298                 buf = nmem_malloc(nmem, len + 3);
 299                 buf[0] = '\'';
 300                 strcpy(buf + 1, val);
 301                 buf[len+1] = '\'';
 302                 buf[len+2] = '\0';
 303                 parms[offset++] = md->name;
 304                 parms[offset++] = buf;
 305                 nparms++;
 306             }
 307         }
 308     }
 309     parms[offset] = 0;
 310 }
 311
 312 // Add static values from session database settings if applicable
 313 static void insert_settings_values(struct session_database *sdb, xmlDoc *doc,
 314     struct conf_service *service)
 315 {
 316     int i;
 317
 318     for (i = 0; i < service->num_metadata; i++)
 319     {
 320         struct conf_metadata *md = &service->metadata[i];
 321         int offset;
 322
 323         if (md->setting == Metadata_setting_postproc &&
 324             (offset = settings_lookup_offset(service, md->name)) >= 0)
 325         {
 326             const char *val = session_setting_oneval(sdb, offset);
 327             if (val)
 328             {
 329                 xmlNode *r = xmlDocGetRootElement(doc);
 330                 xmlNode *n = xmlNewTextChild(r, 0, (xmlChar *) "metadata",
 331                                              (xmlChar *) val);
 332                 xmlSetProp(n, (xmlChar *) "type", (xmlChar *) md->name);
 333             }
 334         }
 335     }
 336 }
 337
 338 static xmlDoc *normalize_record(struct session *se,
 339                                 struct session_database *sdb,
 340                                 struct conf_service *service,
 341                                 const char *rec, NMEM nmem)
 342 {
 343     xmlDoc *rdoc = record_to_xml(se, sdb, rec);
 344
 345     if (rdoc)
 346     {
 347         char *parms[MAX_XSLT_ARGS*2+1];
 348
 349         insert_settings_parameters(sdb, service, parms, nmem);
 350
 351         if (normalize_record_transform(sdb->map, &rdoc, (const char **)parms))
 352         {
 353             session_log(se, YLOG_WARN, "Normalize failed from %s",
 354                         sdb->database->id);
 355         }
 356         else
 357         {
 358             insert_settings_values(sdb, rdoc, service);
 359
 360             if (global_parameters.dump_records)
 361             {
 362                 session_log(se, YLOG_LOG, "Normalized record from %s",
 363                             sdb->database->id);
 364                 log_xml_doc(rdoc);
 365             }
 366         }
 367     }
 368     return rdoc;
 369 }
 370
 371 void session_settings_dump(struct session *se,
 372                            struct session_database *db,
 373                            WRBUF w)
 374 {
 375     if (db->settings)
 376     {
 377         int i, num = db->num_settings;
 378         for (i = 0; i < num; i++)
 379         {
 380             struct setting *s = db->settings[i];
 381             for (;s ; s = s->next)
 382             {
 383                 wrbuf_puts(w, "<set name=\"");
 384                 wrbuf_xmlputs(w, s->name);
 385                 wrbuf_puts(w, "\" value=\"");
 386                 wrbuf_xmlputs(w, s->value);
 387                 wrbuf_puts(w, "\"/>");
 388             }
 389             if (db->settings[i])
 390                 wrbuf_puts(w, "\n");
 391         }
 392     }
 393 }
 394
 395 // Retrieve first defined value for 'name' for given database.
 396 // Will be extended to take into account user associated with session
 397 const char *session_setting_oneval(struct session_database *db, int offset)
 398 {
 399     if (offset >= db->num_settings || !db->settings[offset])
 400         return "";
 401     return db->settings[offset]->value;
 402 }
 403
 404 // Prepare XSLT stylesheets for record normalization
 405 // Structures are allocated on the session_wide nmem to avoid having
 406 // to recompute this for every search. This would lead
 407 // to leaking if a single session was to repeatedly change the PZ_XSLT
 408 // setting. However, this is not a realistic use scenario.
 409 static int prepare_map(struct session *se, struct session_database *sdb)
 410 {
 411     const char *s;
 412
 413     if (sdb->settings && sdb->settings[PZ_XSLT] && !sdb->map &&
 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     return 0;
 447 }
 448
 449 // called if watch should be removed because http_channel is to be destroyed
 450 static void session_watch_cancel(void *data, struct http_channel *c,
 451                                  void *data2)
 452 {
 453     struct session_watchentry *ent = data;
 454
 455     ent->fun = 0;
 456     ent->data = 0;
 457     ent->obs = 0;
 458 }
 459
 460 // set watch. Returns 0=OK, -1 if watch is already set
 461 int session_set_watch(struct session *s, int what,
 462                       session_watchfun fun, void *data,
 463                       struct http_channel *chan)
 464 {
 465     int ret;
 466     session_enter(s);
 467     if (s->watchlist[what].fun)
 468         ret = -1;
 469     else
 470     {
 471
 472         s->watchlist[what].fun = fun;
 473         s->watchlist[what].data = data;
 474         s->watchlist[what].obs = http_add_observer(chan, &s->watchlist[what],
 475                                                    session_watch_cancel);
 476         ret = 0;
 477     }
 478     session_leave(s);
 479     return ret;
 480 }
 481
 482 void session_alert_watch(struct session *s, int what)
 483 {
 484     assert(s);
 485     session_enter(s);
 486     if (s->watchlist[what].fun)
 487     {
 488         /* our watch is no longer associated with http_channel */
 489         void *data;
 490         session_watchfun fun;
 491
 492         http_remove_observer(s->watchlist[what].obs);
 493         fun  = s->watchlist[what].fun;
 494         data = s->watchlist[what].data;
 495
 496         /* reset watch before fun is invoked - in case fun wants to set
 497            it again */
 498         s->watchlist[what].fun = 0;
 499         s->watchlist[what].data = 0;
 500         s->watchlist[what].obs = 0;
 501
 502         session_leave(s);
 503         session_log(s, YLOG_DEBUG,
 504                     "Alert Watch: %d calling function: %p", what, fun);
 505         fun(data);
 506     }
 507     else
 508         session_leave(s);
 509 }
 510
 511 //callback for grep_databases
 512 static void select_targets_callback(struct session *se,
 513                                     struct session_database *db)
 514 {
 515     struct client *cl;
 516     struct client_list *l;
 517
 518     for (l = se->clients_cached; l; l = l->next)
 519         if (client_get_database(l->client) == db)
 520             break;
 521
 522     if (l)
 523         cl = l->client;
 524     else
 525     {
 526         cl = client_create(db->database->id);
 527         client_set_database(cl, db);
 528
 529         l = xmalloc(sizeof(*l));
 530         l->client = cl;
 531         l->next = se->clients_cached;
 532         se->clients_cached = l;
 533     }
 534     /* set session always. If may be 0 if client is not active */
 535     client_set_session(cl, se);
 536
 537     l = xmalloc(sizeof(*l));
 538     l->client = cl;
 539     l->next = se->clients_active;
 540     se->clients_active = l;
 541 }
 542
 543 static void session_reset_active_clients(struct session *se,
 544                                          struct client_list *new_list)
 545 {
 546     struct client_list *l;
 547
 548     session_enter(se);
 549     l = se->clients_active;
 550     se->clients_active = new_list;
 551     session_leave(se);
 552
 553     while (l)
 554     {
 555         struct client_list *l_next = l->next;
 556
 557         client_lock(l->client);
 558         client_set_session(l->client, 0); /* mark client inactive */
 559         client_unlock(l->client);
 560
 561         xfree(l);
 562         l = l_next;
 563     }
 564 }
 565
 566 static void session_remove_cached_clients(struct session *se)
 567 {
 568     struct client_list *l;
 569
 570     session_reset_active_clients(se, 0);
 571
 572     session_enter(se);
 573     l = se->clients_cached;
 574     se->clients_cached = 0;
 575     session_leave(se);
 576
 577     while (l)
 578     {
 579         struct client_list *l_next = l->next;
 580         client_lock(l->client);
 581         client_set_session(l->client, 0);
 582         client_set_database(l->client, 0);
 583         client_unlock(l->client);
 584         client_destroy(l->client);
 585         xfree(l);
 586         l = l_next;
 587     }
 588 }
 589
 590 // Associates a set of clients with a session;
 591 // Note: Session-databases represent databases with per-session
 592 // setting overrides
 593 static int select_targets(struct session *se, const char *filter)
 594 {
 595     return session_grep_databases(se, filter, select_targets_callback);
 596 }
 597
 598 int session_active_clients(struct session *s)
 599 {
 600     struct client_list *l;
 601     int res = 0;
 602
 603     for (l = s->clients_active; l; l = l->next)
 604         if (client_is_active(l->client))
 605             res++;
 606
 607     return res;
 608 }
 609
 610 int session_is_preferred_clients_ready(struct session *s)
 611 {
 612     struct client_list *l;
 613     int res = 0;
 614
 615     for (l = s->clients_active; l; l = l->next)
 616         if (client_is_active_preferred(l->client))
 617             res++;
 618     session_log(s, YLOG_DEBUG, "Has %d active preferred clients.", res);
 619     return res == 0;
 620 }
 621
 622 void session_sort(struct session *se, const char *field, int increasing)
 623 {
 624     struct session_sorted_results *sr;
 625     struct client_list *l;
 626
 627     session_enter(se);
 628
 629     /* see if we already have sorted for this critieria */
 630     for (sr = se->sorted_results; sr; sr = sr->next)
 631     {
 632         if (!strcmp(field, sr->field) && increasing == sr->increasing)
 633             break;
 634     }
 635     if (sr)
 636     {
 637         yaz_log(YLOG_LOG, "search_sort: field=%s increasing=%d already fetched",
 638                 field, increasing);
 639         session_leave(se);
 640         return;
 641     }
 642     yaz_log(YLOG_LOG, "search_sort: field=%s increasing=%d must fetch",
 643             field, increasing);
 644     sr = nmem_malloc(se->nmem, sizeof(*sr));
 645     sr->field = nmem_strdup(se->nmem, field);
 646     sr->increasing = increasing;
 647     sr->next = se->sorted_results;
 648     se->sorted_results = sr;
 649
 650     for (l = se->clients_active; l; l = l->next)
 651     {
 652         struct client *cl = l->client;
 653         if (client_get_state(cl) == Client_Connecting ||
 654             client_get_state(cl) == Client_Idle ||
 655             client_get_state(cl) == Client_Working)
 656             client_start_search(cl);
 657     }
 658     session_leave(se);
 659 }
 660
 661 enum pazpar2_error_code session_search(struct session *se,
 662                                        const char *query,
 663                                        const char *startrecs,
 664                                        const char *maxrecs,
 665                                        const char *filter,
 666                                        const char *limit,
 667                                        const char **addinfo,
 668                                        const char *sort_field, int increasing)
 669 {
 670     int live_channels = 0;
 671     int no_working = 0;
 672     int no_failed_query = 0;
 673     int no_failed_limit = 0;
 674     struct client_list *l, *l0;
 675     struct timeval tval;
 676     facet_limits_t facet_limits;
 677
 678     session_log(se, YLOG_DEBUG, "Search");
 679
 680     *addinfo = 0;
 681
 682     if (se->settings_modified)
 683         session_remove_cached_clients(se);
 684     else
 685         session_reset_active_clients(se, 0);
 686
 687     session_enter(se);
 688     reclist_destroy(se->reclist);
 689     se->reclist = 0;
 690     se->settings_modified = 0;
 691     relevance_destroy(&se->relevance);
 692     nmem_reset(se->nmem);
 693     se->total_records = se->total_merged = 0;
 694     se->num_termlists = 0;
 695
 696     /* reset list of sorted results and clear to relevance search */
 697     se->sorted_results = nmem_malloc(se->nmem, sizeof(*se->sorted_results));
 698     se->sorted_results->field = nmem_strdup(se->nmem, sort_field);
 699     se->sorted_results->increasing = increasing;
 700     se->sorted_results->next = 0;
 701
 702     live_channels = select_targets(se, filter);
 703     if (!live_channels)
 704     {
 705         session_leave(se);
 706         return PAZPAR2_NO_TARGETS;
 707     }
 708     se->reclist = reclist_create(se->nmem);
 709
 710     yaz_gettimeofday(&tval);
 711
 712     tval.tv_sec += 5;
 713
 714     facet_limits = facet_limits_create(limit);
 715     if (!facet_limits)
 716     {
 717         *addinfo = "limit";
 718         session_leave(se);
 719         return PAZPAR2_MALFORMED_PARAMETER_VALUE;
 720     }
 721
 722     l0 = se->clients_active;
 723     se->clients_active = 0;
 724     session_leave(se);
 725
 726     for (l = l0; l; l = l->next)
 727     {
 728         int parse_ret;
 729         struct client *cl = l->client;
 730
 731         if (prepare_map(se, client_get_database(cl)) < 0)
 732             continue;
 733
 734         parse_ret = client_parse_query(cl, query, facet_limits, startrecs,
 735             maxrecs);
 736         if (parse_ret == -1)
 737             no_failed_query++;
 738         else if (parse_ret == -2)
 739             no_failed_limit++;
 740         else
 741         {
 742             int r =
 743                 client_prep_connection(cl, se->service->z3950_operation_timeout,
 744                                        se->service->z3950_session_timeout,
 745                                        se->service->server->iochan_man,
 746                                        &tval);
 747             if (parse_ret == 1 && r == 2)
 748             {
 749                 session_log(se, YLOG_LOG, "client REUSE %s", client_get_id(cl));
 750                 client_reingest(cl);
 751             }
 752             else if (r)
 753             {
 754                 session_log(se, YLOG_LOG, "client NEW %s", client_get_id(cl));
 755                 client_start_search(cl);
 756             }
 757             no_working++;
 758         }
 759     }
 760     facet_limits_destroy(facet_limits);
 761     session_reset_active_clients(se, l0);
 762
 763     if (no_working == 0)
 764     {
 765         if (no_failed_query > 0)
 766         {
 767             *addinfo = "query";
 768             return PAZPAR2_MALFORMED_PARAMETER_VALUE;
 769         }
 770         else if (no_failed_limit > 0)
 771         {
 772             *addinfo = "limit";
 773             return PAZPAR2_MALFORMED_PARAMETER_VALUE;
 774         }
 775         else
 776             return PAZPAR2_NO_TARGETS;
 777     }
 778     yaz_log(YLOG_LOG, "session_start_search done");
 779     return PAZPAR2_NO_ERROR;
 780 }
 781
 782 // Creates a new session_database object for a database
 783 static void session_init_databases_fun(void *context, struct database *db)
 784 {
 785     struct session *se = (struct session *) context;
 786     struct session_database *new = nmem_malloc(se->session_nmem, sizeof(*new));
 787     int i;
 788
 789     new->database = db;
 790
 791     new->map = 0;
 792     assert(db->settings);
 793     new->settings = nmem_malloc(se->session_nmem,
 794                                 sizeof(struct settings *) * db->num_settings);
 795     new->num_settings = db->num_settings;
 796     for (i = 0; i < db->num_settings; i++)
 797     {
 798         struct setting *setting = db->settings[i];
 799         new->settings[i] = setting;
 800     }
 801     new->next = se->databases;
 802     se->databases = new;
 803 }
 804
 805 // Doesn't free memory associated with sdb -- nmem takes care of that
 806 static void session_database_destroy(struct session_database *sdb)
 807 {
 808     sdb->map = 0;
 809 }
 810
 811 // Initialize session_database list -- this represents this session's view
 812 // of the database list -- subject to modification by the settings ws command
 813 void session_init_databases(struct session *se)
 814 {
 815     se->databases = 0;
 816     predef_grep_databases(se, se->service, session_init_databases_fun);
 817 }
 818
 819 // Probably session_init_databases_fun should be refactored instead of
 820 // called here.
 821 static struct session_database *load_session_database(struct session *se,
 822                                                       char *id)
 823 {
 824     struct database *db = new_database(id, se->session_nmem);
 825
 826     session_init_databases_fun((void*) se, db);
 827
 828     // New sdb is head of se->databases list
 829     return se->databases;
 830 }
 831
 832 // Find an existing session database. If not found, load it
 833 static struct session_database *find_session_database(struct session *se,
 834                                                       char *id)
 835 {
 836     struct session_database *sdb;
 837
 838     for (sdb = se->databases; sdb; sdb = sdb->next)
 839         if (!strcmp(sdb->database->id, id))
 840             return sdb;
 841     return load_session_database(se, id);
 842 }
 843
 844 // Apply a session override to a database
 845 void session_apply_setting(struct session *se, char *dbname, char *setting,
 846                            char *value)
 847 {
 848     struct session_database *sdb = find_session_database(se, dbname);
 849     struct conf_service *service = se->service;
 850     struct setting *new = nmem_malloc(se->session_nmem, sizeof(*new));
 851     int offset = settings_create_offset(service, setting);
 852
 853     expand_settings_array(&sdb->settings, &sdb->num_settings, offset,
 854                           se->session_nmem);
 855     new->precedence = 0;
 856     new->target = dbname;
 857     new->name = setting;
 858     new->value = value;
 859     new->next = sdb->settings[offset];
 860     sdb->settings[offset] = new;
 861
 862     se->settings_modified = 1;
 863
 864     // Force later recompute of settings-driven data structures
 865     // (happens when a search starts and client connections are prepared)
 866     switch (offset)
 867     {
 868     case PZ_XSLT:
 869         if (sdb->map)
 870         {
 871             sdb->map = 0;
 872         }
 873         break;
 874     }
 875 }
 876
 877 void session_destroy(struct session *se)
 878 {
 879     struct session_database *sdb;
 880     session_log(se, YLOG_DEBUG, "Destroying");
 881     session_use(-1);
 882     session_remove_cached_clients(se);
 883
 884     for (sdb = se->databases; sdb; sdb = sdb->next)
 885         session_database_destroy(sdb);
 886     normalize_cache_destroy(se->normalize_cache);
 887     relevance_destroy(&se->relevance);
 888     reclist_destroy(se->reclist);
 889     nmem_destroy(se->nmem);
 890     service_destroy(se->service);
 891     yaz_mutex_destroy(&se->session_mutex);
 892 }
 893
 894 size_t session_get_memory_status(struct session *session) {
 895     size_t session_nmem;
 896     if (session == 0)
 897         return 0;
 898     session_enter(session);
 899     session_nmem = nmem_total(session->nmem);
 900     session_leave(session);
 901     return session_nmem;
 902 }
 903
 904
 905 struct session *new_session(NMEM nmem, struct conf_service *service,
 906                             unsigned session_id)
 907 {
 908     int i;
 909     struct session *session = nmem_malloc(nmem, sizeof(*session));
 910
 911     char tmp_str[50];
 912
 913     sprintf(tmp_str, "session#%u", session_id);
 914
 915     session->session_id = session_id;
 916     session_log(session, YLOG_DEBUG, "New");
 917     session->service = service;
 918     session->relevance = 0;
 919     session->total_records = 0;
 920     session->number_of_warnings_unknown_elements = 0;
 921     session->number_of_warnings_unknown_metadata = 0;
 922     session->num_termlists = 0;
 923     session->reclist = 0;
 924     session->clients_active = 0;
 925     session->clients_cached = 0;
 926     session->settings_modified = 0;
 927     session->session_nmem = nmem;
 928     session->nmem = nmem_create();
 929     session->databases = 0;
 930     for (i = 0; i <= SESSION_WATCH_MAX; i++)
 931     {
 932         session->watchlist[i].data = 0;
 933         session->watchlist[i].fun = 0;
 934     }
 935     session->normalize_cache = normalize_cache_create();
 936     session->session_mutex = 0;
 937     pazpar2_mutex_create(&session->session_mutex, tmp_str);
 938     session_use(1);
 939     return session;
 940 }
 941
 942 const char * client_get_suggestions_xml(struct client *cl, WRBUF wrbuf);
 943
 944 static struct hitsbytarget *hitsbytarget_nb(struct session *se,
 945                                             int *count, NMEM nmem)
 946 {
 947     struct hitsbytarget *res = 0;
 948     struct client_list *l;
 949     size_t sz = 0;
 950
 951     for (l = se->clients_active; l; l = l->next)
 952         sz++;
 953
 954     res = nmem_malloc(nmem, sizeof(*res) * sz);
 955     *count = 0;
 956     for (l = se->clients_active; l; l = l->next)
 957     {
 958         struct client *cl = l->client;
 959         WRBUF w = wrbuf_alloc();
 960         const char *name = session_setting_oneval(client_get_database(cl),
 961                                                   PZ_NAME);
 962
 963         res[*count].id = client_get_id(cl);
 964         res[*count].name = *name ? name : "Unknown";
 965         res[*count].hits = client_get_hits(cl);
 966         res[*count].records = client_get_num_records(cl);
 967         res[*count].filtered = client_get_num_records_filtered(cl);
 968         res[*count].diagnostic =
 969             client_get_diagnostic(cl, &res[*count].addinfo);
 970         res[*count].state = client_get_state_str(cl);
 971         res[*count].connected  = client_get_connection(cl) ? 1 : 0;
 972         session_settings_dump(se, client_get_database(cl), w);
 973         res[*count].settings_xml = nmem_strdup(nmem, wrbuf_cstr(w));
 974         wrbuf_rewind(w);
 975         wrbuf_puts(w, "");
 976         res[*count].suggestions_xml = nmem_strdup(nmem, client_get_suggestions_xml(cl, w));
 977         wrbuf_destroy(w);
 978         (*count)++;
 979     }
 980     return res;
 981 }
 982
 983 struct hitsbytarget *get_hitsbytarget(struct session *se, int *count, NMEM nmem)
 984 {
 985     struct hitsbytarget *p;
 986     session_enter(se);
 987     p = hitsbytarget_nb(se, count, nmem);
 988     session_leave(se);
 989     return p;
 990 }
 991
 992 struct termlist_score **get_termlist_score(struct session *se,
 993                                            const char *name, int *num)
 994 {
 995     int i;
 996     struct termlist_score **tl = 0;
 997
 998     session_enter(se);
 999     for (i = 0; i < se->num_termlists; i++)
1000         if (!strcmp((const char *) se->termlists[i].name, name))
1001         {
1002             tl = termlist_highscore(se->termlists[i].termlist, num);
1003             break;
1004         }
1005     session_leave(se);
1006     return tl;
1007 }
1008
1009 // Compares two hitsbytarget nodes by hitcount
1010 static int cmp_ht(const void *p1, const void *p2)
1011 {
1012     const struct hitsbytarget *h1 = p1;
1013     const struct hitsbytarget *h2 = p2;
1014     return h2->hits - h1->hits;
1015 }
1016
1017 static int targets_termlist_nb(WRBUF wrbuf, struct session *se, int num,
1018                                NMEM nmem, int version)
1019 {
1020     struct hitsbytarget *ht;
1021     int count, i;
1022
1023     ht = hitsbytarget_nb(se, &count, nmem);
1024     qsort(ht, count, sizeof(struct hitsbytarget), cmp_ht);
1025     for (i = 0; i < count && i < num && ht[i].hits > 0; i++)
1026     {
1027
1028         // do only print terms which have display names
1029
1030         wrbuf_puts(wrbuf, "<term>\n");
1031
1032         wrbuf_puts(wrbuf, "<id>");
1033         wrbuf_xmlputs(wrbuf, ht[i].id);
1034         wrbuf_puts(wrbuf, "</id>\n");
1035
1036         wrbuf_puts(wrbuf, "<name>");
1037         if (!ht[i].name || !ht[i].name[0])
1038             wrbuf_xmlputs(wrbuf, "NO TARGET NAME");
1039         else
1040             wrbuf_xmlputs(wrbuf, ht[i].name);
1041         wrbuf_puts(wrbuf, "</name>\n");
1042
1043         wrbuf_printf(wrbuf, "<frequency>" ODR_INT_PRINTF "</frequency>\n",
1044                      ht[i].hits);
1045
1046         if (version >= 2) {
1047             wrbuf_printf(wrbuf, "<records>%d</records>\n", ht[i].records - ht[i].filtered);
1048             wrbuf_printf(wrbuf, "<filtered>%d</filtered>\n", ht[i].filtered);
1049         }
1050
1051         wrbuf_puts(wrbuf, "<state>");
1052         wrbuf_xmlputs(wrbuf, ht[i].state);
1053         wrbuf_puts(wrbuf, "</state>\n");
1054
1055         wrbuf_printf(wrbuf, "<diagnostic>%d</diagnostic>\n",
1056                      ht[i].diagnostic);
1057         wrbuf_puts(wrbuf, "</term>\n");
1058     }
1059     return count;
1060 }
1061
1062 void perform_termlist(struct http_channel *c, struct session *se,
1063                       const char *name, int num, int version)
1064 {
1065     int i, j;
1066     NMEM nmem_tmp = nmem_create();
1067     char **names;
1068     int num_names = 0;
1069
1070     if (!name)
1071         name = "*";
1072
1073     nmem_strsplit(nmem_tmp, ",", name, &names, &num_names);
1074
1075     session_enter(se);
1076
1077     for (j = 0; j < num_names; j++)
1078     {
1079         const char *tname;
1080         int must_generate_empty = 1; /* bug 5350 */
1081
1082         for (i = 0; i < se->num_termlists; i++)
1083         {
1084             tname = se->termlists[i].name;
1085             if (!strcmp(names[j], tname) || !strcmp(names[j], "*"))
1086             {
1087                 struct termlist_score **p = 0;
1088                 int len;
1089
1090                 wrbuf_puts(c->wrbuf, "<list name=\"");
1091                 wrbuf_xmlputs(c->wrbuf, tname);
1092                 wrbuf_puts(c->wrbuf, "\">\n");
1093                 must_generate_empty = 0;
1094
1095                 p = termlist_highscore(se->termlists[i].termlist, &len);
1096                 if (p)
1097                 {
1098                     int i;
1099                     for (i = 0; i < len && i < num; i++)
1100                     {
1101                         // prevent sending empty term elements
1102                         if (!p[i]->display_term || !p[i]->display_term[0])
1103                             continue;
1104
1105                         wrbuf_puts(c->wrbuf, "<term>");
1106                         wrbuf_puts(c->wrbuf, "<name>");
1107                         wrbuf_xmlputs(c->wrbuf, p[i]->display_term);
1108                         wrbuf_puts(c->wrbuf, "</name>");
1109
1110                         wrbuf_printf(c->wrbuf,
1111                                      "<frequency>%d</frequency>",
1112                                      p[i]->frequency);
1113                         wrbuf_puts(c->wrbuf, "</term>\n");
1114                     }
1115                 }
1116                 wrbuf_puts(c->wrbuf, "</list>\n");
1117             }
1118         }
1119         tname = "xtargets";
1120         if (!strcmp(names[j], tname) || !strcmp(names[j], "*"))
1121         {
1122             wrbuf_puts(c->wrbuf, "<list name=\"");
1123             wrbuf_xmlputs(c->wrbuf, tname);
1124             wrbuf_puts(c->wrbuf, "\">\n");
1125
1126             targets_termlist_nb(c->wrbuf, se, num, c->nmem, version);
1127             wrbuf_puts(c->wrbuf, "</list>\n");
1128             must_generate_empty = 0;
1129         }
1130         if (must_generate_empty)
1131         {
1132             wrbuf_puts(c->wrbuf, "<list name=\"");
1133             wrbuf_xmlputs(c->wrbuf, names[j]);
1134             wrbuf_puts(c->wrbuf, "\"/>\n");
1135         }
1136     }
1137     session_leave(se);
1138     nmem_destroy(nmem_tmp);
1139 }
1140
1141 #ifdef MISSING_HEADERS
1142 void report_nmem_stats(void)
1143 {
1144     size_t in_use, is_free;
1145
1146     nmem_get_memory_in_use(&in_use);
1147     nmem_get_memory_free(&is_free);
1148
1149     yaz_log(YLOG_LOG, "nmem stat: use=%ld free=%ld",
1150             (long) in_use, (long) is_free);
1151 }
1152 #endif
1153
1154 struct record_cluster *show_single_start(struct session *se, const char *id,
1155                                          struct record_cluster **prev_r,
1156                                          struct record_cluster **next_r)
1157 {
1158     struct record_cluster *r = 0;
1159
1160     session_enter(se);
1161     *prev_r = 0;
1162     *next_r = 0;
1163     if (se->reclist)
1164     {
1165         reclist_enter(se->reclist);
1166         while ((r = reclist_read_record(se->reclist)))
1167         {
1168             if (!strcmp(r->recid, id))
1169             {
1170                 *next_r = reclist_read_record(se->reclist);
1171                 break;
1172             }
1173             *prev_r = r;
1174         }
1175         reclist_leave(se->reclist);
1176     }
1177     if (!r)
1178         session_leave(se);
1179     return r;
1180 }
1181
1182 void show_single_stop(struct session *se, struct record_cluster *rec)
1183 {
1184     session_leave(se);
1185 }
1186
1187 struct record_cluster **show_range_start(struct session *se,
1188                                          struct reclist_sortparms *sp,
1189                                          int start, int *num, int *total, Odr_int *sumhits)
1190 {
1191     struct record_cluster **recs;
1192     struct reclist_sortparms *spp;
1193     int i;
1194 #if USE_TIMING
1195     yaz_timing_t t = yaz_timing_create();
1196 #endif
1197     session_enter(se);
1198     recs = nmem_malloc(se->nmem, *num * sizeof(struct record_cluster *));
1199     if (!se->relevance)
1200     {
1201         *num = 0;
1202         *total = 0;
1203         *sumhits = 0;
1204         recs = 0;
1205     }
1206     else
1207     {
1208         struct client_list *l;
1209
1210         for (spp = sp; spp; spp = spp->next)
1211             if (spp->type == Metadata_sortkey_relevance)
1212             {
1213                 relevance_prepare_read(se->relevance, se->reclist);
1214                 break;
1215             }
1216         reclist_sort(se->reclist, sp);
1217
1218         reclist_enter(se->reclist);
1219         *total = reclist_get_num_records(se->reclist);
1220
1221         *sumhits = 0;
1222         for (l = se->clients_active; l; l = l->next)
1223             *sumhits += client_get_hits(l->client);
1224
1225         for (i = 0; i < start; i++)
1226             if (!reclist_read_record(se->reclist))
1227             {
1228                 *num = 0;
1229                 recs = 0;
1230                 break;
1231             }
1232
1233         for (i = 0; i < *num; i++)
1234         {
1235             struct record_cluster *r = reclist_read_record(se->reclist);
1236             if (!r)
1237             {
1238                 *num = i;
1239                 break;
1240             }
1241             recs[i] = r;
1242         }
1243         reclist_leave(se->reclist);
1244     }
1245 #if USE_TIMING
1246     yaz_timing_stop(t);
1247     yaz_log(YLOG_LOG, "show %6.5f %3.2f %3.2f",
1248             yaz_timing_get_real(t), yaz_timing_get_user(t),
1249             yaz_timing_get_sys(t));
1250     yaz_timing_destroy(&t);
1251 #endif
1252     return recs;
1253 }
1254
1255 void show_range_stop(struct session *se, struct record_cluster **recs)
1256 {
1257     session_leave(se);
1258 }
1259
1260 void statistics(struct session *se, struct statistics *stat)
1261 {
1262     struct client_list *l;
1263     int count = 0;
1264
1265     memset(stat, 0, sizeof(*stat));
1266     stat->num_hits = 0;
1267     for (l = se->clients_active; l; l = l->next)
1268     {
1269         struct client *cl = l->client;
1270         if (!client_get_connection(cl))
1271             stat->num_no_connection++;
1272         stat->num_hits += client_get_hits(cl);
1273         switch (client_get_state(cl))
1274         {
1275         case Client_Connecting: stat->num_connecting++; break;
1276         case Client_Working: stat->num_working++; break;
1277         case Client_Idle: stat->num_idle++; break;
1278         case Client_Failed: stat->num_failed++; break;
1279         case Client_Error: stat->num_error++; break;
1280         default: break;
1281         }
1282         count++;
1283     }
1284     stat->num_records = se->total_records;
1285
1286     stat->num_clients = count;
1287 }
1288
1289 static struct record_metadata *record_metadata_init(
1290     NMEM nmem, const char *value, enum conf_metadata_type type,
1291     struct _xmlAttr *attr)
1292 {
1293     struct record_metadata *rec_md = record_metadata_create(nmem);
1294     struct record_metadata_attr **attrp = &rec_md->attributes;
1295
1296     for (; attr; attr = attr->next)
1297     {
1298         if (attr->children && attr->children->content)
1299         {
1300             if (strcmp((const char *) attr->name, "type"))
1301             {  /* skip the "type" attribute.. Its value is already part of
1302                   the element in output (md-%s) and so repeating it here
1303                   is redundant */
1304                 *attrp = nmem_malloc(nmem, sizeof(**attrp));
1305                 (*attrp)->name =
1306                     nmem_strdup(nmem, (const char *) attr->name);
1307                 (*attrp)->value =
1308                     nmem_strdup(nmem, (const char *) attr->children->content);
1309                 attrp = &(*attrp)->next;
1310             }
1311         }
1312     }
1313     *attrp = 0;
1314
1315     if (type == Metadata_type_generic)
1316     {
1317         char *p = nmem_strdup(nmem, value);
1318
1319         p = normalize7bit_generic(p, " ,/.:([");
1320
1321         rec_md->data.text.disp = p;
1322         rec_md->data.text.sort = 0;
1323     }
1324     else if (type == Metadata_type_year || type == Metadata_type_date)
1325     {
1326         int first, last;
1327         int longdate = 0;
1328
1329         if (type == Metadata_type_date)
1330             longdate = 1;
1331         if (extract7bit_dates((char *) value, &first, &last, longdate) < 0)
1332             return 0;
1333
1334         rec_md->data.number.min = first;
1335         rec_md->data.number.max = last;
1336     }
1337     else
1338         return 0;
1339     return rec_md;
1340 }
1341
1342 static int get_mergekey_from_doc(xmlDoc *doc, xmlNode *root, const char *name,
1343                                  struct conf_service *service, WRBUF norm_wr)
1344 {
1345     xmlNode *n;
1346     int no_found = 0;
1347     for (n = root->children; n; n = n->next)
1348     {
1349         if (n->type != XML_ELEMENT_NODE)
1350             continue;
1351         if (!strcmp((const char *) n->name, "metadata"))
1352         {
1353             xmlChar *type = xmlGetProp(n, (xmlChar *) "type");
1354             if (type == NULL) {
1355                 yaz_log(YLOG_FATAL, "Missing type attribute on metadata element. Skipping!");
1356             }
1357             else if (!strcmp(name, (const char *) type))
1358             {
1359                 xmlChar *value = xmlNodeListGetString(doc, n->children, 1);
1360                 if (value)
1361                 {
1362                     const char *norm_str;
1363                     pp2_charset_token_t prt =
1364                         pp2_charset_token_create(service->charsets, "mergekey");
1365
1366                     pp2_charset_token_first(prt, (const char *) value, 0);
1367                     if (wrbuf_len(norm_wr) > 0)
1368                         wrbuf_puts(norm_wr, " ");
1369                     wrbuf_puts(norm_wr, name);
1370                     while ((norm_str =
1371                             pp2_charset_token_next(prt)))
1372                     {
1373                         if (*norm_str)
1374                         {
1375                             wrbuf_puts(norm_wr, " ");
1376                             wrbuf_puts(norm_wr, norm_str);
1377                         }
1378                     }
1379                     xmlFree(value);
1380                     pp2_charset_token_destroy(prt);
1381                     no_found++;
1382                 }
1383             }
1384             xmlFree(type);
1385         }
1386     }
1387     return no_found;
1388 }
1389
1390 static const char *get_mergekey(xmlDoc *doc, struct client *cl, int record_no,
1391                                 struct conf_service *service, NMEM nmem)
1392 {
1393     char *mergekey_norm = 0;
1394     xmlNode *root = xmlDocGetRootElement(doc);
1395     WRBUF norm_wr = wrbuf_alloc();
1396
1397     /* consider mergekey from XSL first */
1398     xmlChar *mergekey = xmlGetProp(root, (xmlChar *) "mergekey");
1399     if (mergekey)
1400     {
1401         const char *norm_str;
1402         pp2_charset_token_t prt =
1403             pp2_charset_token_create(service->charsets, "mergekey");
1404
1405         pp2_charset_token_first(prt, (const char *) mergekey, 0);
1406         while ((norm_str = pp2_charset_token_next(prt)))
1407         {
1408             if (*norm_str)
1409             {
1410                 if (wrbuf_len(norm_wr))
1411                     wrbuf_puts(norm_wr, " ");
1412                 wrbuf_puts(norm_wr, norm_str);
1413             }
1414         }
1415         pp2_charset_token_destroy(prt);
1416         xmlFree(mergekey);
1417     }
1418     else
1419     {
1420         /* no mergekey defined in XSL. Look for mergekey metadata instead */
1421         int field_id;
1422         for (field_id = 0; field_id < service->num_metadata; field_id++)
1423         {
1424             struct conf_metadata *ser_md = &service->metadata[field_id];
1425             if (ser_md->mergekey != Metadata_mergekey_no)
1426             {
1427                 int r = get_mergekey_from_doc(doc, root, ser_md->name,
1428                                               service, norm_wr);
1429                 if (r == 0 && ser_md->mergekey == Metadata_mergekey_required)
1430                 {
1431                     /* no mergekey on this one and it is required..
1432                        Generate unique key instead */
1433                     wrbuf_rewind(norm_wr);
1434                     break;
1435                 }
1436             }
1437         }
1438     }
1439
1440     /* generate unique key if none is not generated already or is empty */
1441     if (wrbuf_len(norm_wr) == 0)
1442     {
1443         wrbuf_printf(norm_wr, "position: %s-%d",
1444                      client_get_id(cl), record_no);
1445     }
1446     else
1447     {
1448         const char *lead = "content: ";
1449         wrbuf_insert(norm_wr, 0, lead, strlen(lead));
1450     }
1451     if (wrbuf_len(norm_wr) > 0)
1452         mergekey_norm = nmem_strdup(nmem, wrbuf_cstr(norm_wr));
1453     wrbuf_destroy(norm_wr);
1454     return mergekey_norm;
1455 }
1456
1457 /** \brief see if metadata for pz:recordfilter exists
1458     \param root xml root element of normalized record
1459     \param sdb session database for client
1460     \retval 0 if there is no metadata for pz:recordfilter
1461     \retval 1 if there is metadata for pz:recordfilter
1462
1463     If there is no pz:recordfilter defined, this function returns 1
1464     as well.
1465 */
1466
1467 static int check_record_filter(xmlNode *root, struct session_database *sdb)
1468 {
1469     int match = 0;
1470     xmlNode *n;
1471     const char *s;
1472     s = session_setting_oneval(sdb, PZ_RECORDFILTER);
1473
1474     if (!s || !*s)
1475         return 1;
1476
1477     for (n = root->children; n; n = n->next)
1478     {
1479         if (n->type != XML_ELEMENT_NODE)
1480             continue;
1481         if (!strcmp((const char *) n->name, "metadata"))
1482         {
1483             xmlChar *type = xmlGetProp(n, (xmlChar *) "type");
1484             if (type)
1485             {
1486                 size_t len;
1487                 int substring;
1488                 const char *eq;
1489
1490                 if ((eq = strchr(s, '=')))
1491                     substring = 0;
1492                 else if ((eq = strchr(s, '~')))
1493                     substring = 1;
1494                 if (eq)
1495                     len = eq - s;
1496                 else
1497                     len = strlen(s);
1498                 if (len == strlen((const char *)type) &&
1499                     !memcmp((const char *) type, s, len))
1500                 {
1501                     xmlChar *value = xmlNodeGetContent(n);
1502                     if (value && *value)
1503                     {
1504                         if (!eq ||
1505                             (substring && strstr((const char *) value, eq+1)) ||
1506                             (!substring && !strcmp((const char *) value, eq + 1)))
1507                             match = 1;
1508                     }
1509                     xmlFree(value);
1510                 }
1511                 xmlFree(type);
1512             }
1513         }
1514     }
1515     return match;
1516 }
1517
1518
1519 static int ingest_to_cluster(struct client *cl,
1520                              xmlDoc *xdoc,
1521                              xmlNode *root,
1522                              int record_no,
1523                              const char *mergekey_norm);
1524
1525 /** \brief ingest XML record
1526     \param cl client holds the result set for record
1527     \param rec record buffer (0 terminated)
1528     \param record_no record position (1, 2, ..)
1529     \param nmem working NMEM
1530     \retval 0 OK
1531     \retval -1 failure
1532     \retval -2 Filtered
1533 */
1534 int ingest_record(struct client *cl, const char *rec,
1535                   int record_no, NMEM nmem)
1536 {
1537     struct session *se = client_get_session(cl);
1538     int ret = 0;
1539     struct session_database *sdb = client_get_database(cl);
1540     struct conf_service *service = se->service;
1541     xmlDoc *xdoc = normalize_record(se, sdb, service, rec, nmem);
1542     xmlNode *root;
1543     const char *mergekey_norm;
1544
1545     if (!xdoc)
1546         return -1;
1547
1548     root = xmlDocGetRootElement(xdoc);
1549
1550     if (!check_record_filter(root, sdb))
1551     {
1552         session_log(se, YLOG_LOG, "Filtered out record no %d from %s", record_no, sdb->database->id);
1553         xmlFreeDoc(xdoc);
1554         return -2;
1555     }
1556
1557     mergekey_norm = get_mergekey(xdoc, cl, record_no, service, nmem);
1558     if (!mergekey_norm)
1559     {
1560         session_log(se, YLOG_WARN, "Got no mergekey");
1561         xmlFreeDoc(xdoc);
1562         return -1;
1563     }
1564     session_enter(se);
1565     if (client_get_session(cl) == se)
1566         ret = ingest_to_cluster(cl, xdoc, root, record_no, mergekey_norm);
1567     session_leave(se);
1568
1569     xmlFreeDoc(xdoc);
1570     return ret;
1571 }
1572
1573 static int check_limit_local(struct client *cl,
1574                              struct record *record,
1575                              int record_no)
1576 {
1577     int skip_record = 0;
1578     struct session *se = client_get_session(cl);
1579     struct conf_service *service = se->service;
1580     NMEM nmem_tmp = nmem_create();
1581     struct session_database *sdb = client_get_database(cl);
1582     int l = 0;
1583     while (!skip_record)
1584     {
1585         struct conf_metadata *ser_md = 0;
1586         struct record_metadata *rec_md = 0;
1587         int md_field_id;
1588         char **values = 0;
1589         int i, num_v = 0;
1590
1591         const char *name =
1592             client_get_facet_limit_local(cl, sdb, &l, nmem_tmp, &num_v,
1593                                          &values);
1594         if (!name)
1595             break;
1596
1597         md_field_id = conf_service_metadata_field_id(service, name);
1598         if (md_field_id < 0)
1599         {
1600             skip_record = 1;
1601             break;
1602         }
1603         ser_md = &service->metadata[md_field_id];
1604         rec_md = record->metadata[md_field_id];
1605         yaz_log(YLOG_LOG, "check limit local %s", name);
1606         for (i = 0; i < num_v; )
1607         {
1608             if (rec_md)
1609             {
1610                 if (ser_md->type == Metadata_type_year
1611                     || ser_md->type == Metadata_type_date)
1612                 {
1613                     int y = atoi(values[i]);
1614                     if (y >= rec_md->data.number.min
1615                         && y <= rec_md->data.number.max)
1616                         break;
1617                 }
1618                 else
1619                 {
1620                     yaz_log(YLOG_LOG, "cmp: '%s' '%s'",
1621                             rec_md->data.text.disp, values[i]);
1622                     if (!strcmp(rec_md->data.text.disp, values[i]))
1623                     {
1624                         break;
1625                     }
1626                 }
1627                 rec_md = rec_md->next;
1628             }
1629             else
1630             {
1631                 rec_md = record->metadata[md_field_id];
1632                 i++;
1633             }
1634         }
1635         if (i == num_v)
1636         {
1637             skip_record = 1;
1638             break;
1639         }
1640     }
1641     nmem_destroy(nmem_tmp);
1642     return skip_record;
1643 }
1644
1645 static int ingest_to_cluster(struct client *cl,
1646                              xmlDoc *xdoc,
1647                              xmlNode *root,
1648                              int record_no,
1649                              const char *mergekey_norm)
1650 {
1651     xmlNode *n;
1652     xmlChar *type = 0;
1653     xmlChar *value = 0;
1654     struct session *se = client_get_session(cl);
1655     struct conf_service *service = se->service;
1656     int term_factor = 1;
1657     struct record_cluster *cluster;
1658     struct session_database *sdb = client_get_database(cl);
1659     struct record *record = record_create(se->nmem,
1660                                           service->num_metadata,
1661                                           service->num_sortkeys, cl,
1662                                           record_no);
1663
1664     for (n = root->children; n; n = n->next)
1665     {
1666         if (type)
1667             xmlFree(type);
1668         if (value)
1669             xmlFree(value);
1670         type = value = 0;
1671
1672         if (n->type != XML_ELEMENT_NODE)
1673             continue;
1674         if (!strcmp((const char *) n->name, "metadata"))
1675         {
1676             struct conf_metadata *ser_md = 0;
1677             struct record_metadata **wheretoput = 0;
1678             struct record_metadata *rec_md = 0;
1679             int md_field_id = -1;
1680
1681             type = xmlGetProp(n, (xmlChar *) "type");
1682             value = xmlNodeListGetString(xdoc, n->children, 1);
1683
1684             if (!type || !value || !*value)
1685                 continue;
1686
1687             md_field_id
1688                 = conf_service_metadata_field_id(service, (const char *) type);
1689             if (md_field_id < 0)
1690             {
1691                 if (se->number_of_warnings_unknown_metadata == 0)
1692                 {
1693                     session_log(se, YLOG_WARN,
1694                             "Ignoring unknown metadata element: %s", type);
1695                 }
1696                 se->number_of_warnings_unknown_metadata++;
1697                 continue;
1698             }
1699
1700             ser_md = &service->metadata[md_field_id];
1701
1702             // non-merged metadata
1703             rec_md = record_metadata_init(se->nmem, (const char *) value,
1704                                           ser_md->type, n->properties);
1705             if (!rec_md)
1706             {
1707                 session_log(se, YLOG_WARN, "bad metadata data '%s' "
1708                             "for element '%s'", value, type);
1709                 continue;
1710             }
1711             wheretoput = &record->metadata[md_field_id];
1712             while (*wheretoput)
1713                 wheretoput = &(*wheretoput)->next;
1714             *wheretoput = rec_md;
1715         }
1716     }
1717
1718     if (check_limit_local(cl, record, record_no))
1719     {
1720         session_log(se, YLOG_LOG, "Facet filtered out record no %d from %s",
1721                     record_no, sdb->database->id);
1722         if (type)
1723             xmlFree(type);
1724         if (value)
1725             xmlFree(value);
1726         return -2;
1727     }
1728     cluster = reclist_insert(se->reclist, service, record,
1729                              mergekey_norm, &se->total_merged);
1730     if (!cluster)
1731         return -1;
1732
1733     {
1734         const char *use_term_factor_str =
1735             session_setting_oneval(sdb, PZ_TERMLIST_TERM_FACTOR);
1736         if (use_term_factor_str && use_term_factor_str[0] == '1')
1737         {
1738             int maxrecs = client_get_maxrecs(cl);
1739             int hits = (int) client_get_hits(cl);
1740             term_factor = MAX(hits, maxrecs) /  MAX(1, maxrecs);
1741             assert(term_factor >= 1);
1742             yaz_log(YLOG_DEBUG, "Using term factor: %d (%d / %d)", term_factor, MAX(hits, maxrecs), MAX(1, maxrecs));
1743         }
1744     }
1745
1746     if (global_parameters.dump_records)
1747         session_log(se, YLOG_LOG, "Cluster id %s from %s (#%d)", cluster->recid,
1748                     sdb->database->id, record_no);
1749
1750
1751     relevance_newrec(se->relevance, cluster);
1752
1753     // now parsing XML record and adding data to cluster or record metadata
1754     for (n = root->children; n; n = n->next)
1755     {
1756         pp2_charset_token_t prt;
1757         if (type)
1758             xmlFree(type);
1759         if (value)
1760             xmlFree(value);
1761         type = value = 0;
1762
1763         if (n->type != XML_ELEMENT_NODE)
1764             continue;
1765         if (!strcmp((const char *) n->name, "metadata"))
1766         {
1767             struct conf_metadata *ser_md = 0;
1768             struct conf_sortkey *ser_sk = 0;
1769             struct record_metadata **wheretoput = 0;
1770             struct record_metadata *rec_md = 0;
1771             int md_field_id = -1;
1772             int sk_field_id = -1;
1773
1774             type = xmlGetProp(n, (xmlChar *) "type");
1775             value = xmlNodeListGetString(xdoc, n->children, 1);
1776
1777             if (!type || !value || !*value)
1778                 continue;
1779
1780             md_field_id
1781                 = conf_service_metadata_field_id(service, (const char *) type);
1782             if (md_field_id < 0)
1783                 continue;
1784
1785             ser_md = &service->metadata[md_field_id];
1786
1787             if (ser_md->sortkey_offset >= 0)
1788             {
1789                 sk_field_id = ser_md->sortkey_offset;
1790                 ser_sk = &service->sortkeys[sk_field_id];
1791             }
1792
1793             // merged metadata
1794             rec_md = record_metadata_init(se->nmem, (const char *) value,
1795                                           ser_md->type, 0);
1796             if (!rec_md)
1797                 continue;
1798
1799             wheretoput = &cluster->metadata[md_field_id];
1800
1801             // and polulate with data:
1802             // assign cluster or record based on merge action
1803             if (ser_md->merge == Metadata_merge_unique)
1804             {
1805                 while (*wheretoput)
1806                 {
1807                     if (!strcmp((const char *) (*wheretoput)->data.text.disp,
1808                                 rec_md->data.text.disp))
1809                         break;
1810                     wheretoput = &(*wheretoput)->next;
1811                 }
1812                 if (!*wheretoput)
1813                     *wheretoput = rec_md;
1814             }
1815             else if (ser_md->merge == Metadata_merge_longest)
1816             {
1817                 if (!*wheretoput
1818                     || strlen(rec_md->data.text.disp)
1819                     > strlen((*wheretoput)->data.text.disp))
1820                 {
1821                     *wheretoput = rec_md;
1822                     if (ser_sk)
1823                     {
1824                         const char *sort_str = 0;
1825                         int skip_article =
1826                             ser_sk->type == Metadata_sortkey_skiparticle;
1827
1828                         if (!cluster->sortkeys[sk_field_id])
1829                             cluster->sortkeys[sk_field_id] =
1830                                 nmem_malloc(se->nmem,
1831                                             sizeof(union data_types));
1832
1833                         prt =
1834                             pp2_charset_token_create(service->charsets, "sort");
1835
1836                         pp2_charset_token_first(prt, rec_md->data.text.disp,
1837                                                 skip_article);
1838
1839                         pp2_charset_token_next(prt);
1840
1841                         sort_str = pp2_get_sort(prt);
1842
1843                         cluster->sortkeys[sk_field_id]->text.disp =
1844                             rec_md->data.text.disp;
1845                         if (!sort_str)
1846                         {
1847                             sort_str = rec_md->data.text.disp;
1848                             session_log(se, YLOG_WARN,
1849                                     "Could not make sortkey. Bug #1858");
1850                         }
1851                         cluster->sortkeys[sk_field_id]->text.sort =
1852                             nmem_strdup(se->nmem, sort_str);
1853                         pp2_charset_token_destroy(prt);
1854                     }
1855                 }
1856             }
1857             else if (ser_md->merge == Metadata_merge_all)
1858             {
1859                 while (*wheretoput)
1860                     wheretoput = &(*wheretoput)->next;
1861                 *wheretoput = rec_md;
1862             }
1863             else if (ser_md->merge == Metadata_merge_range)
1864             {
1865                 if (!*wheretoput)
1866                 {
1867                     *wheretoput = rec_md;
1868                     if (ser_sk)
1869                         cluster->sortkeys[sk_field_id]
1870                             = &rec_md->data;
1871                 }
1872                 else
1873                 {
1874                     int this_min = rec_md->data.number.min;
1875                     int this_max = rec_md->data.number.max;
1876                     if (this_min < (*wheretoput)->data.number.min)
1877                         (*wheretoput)->data.number.min = this_min;
1878                     if (this_max > (*wheretoput)->data.number.max)
1879                         (*wheretoput)->data.number.max = this_max;
1880                 }
1881             }
1882
1883
1884             // ranking of _all_ fields enabled ...
1885             if (ser_md->rank)
1886                 relevance_countwords(se->relevance, cluster,
1887                                      (char *) value, ser_md->rank,
1888                                      ser_md->name);
1889
1890             // construct facets ... unless the client already has reported them
1891             if (ser_md->termlist && !client_has_facet(cl, (char *) type))
1892             {
1893                 if (ser_md->type == Metadata_type_year)
1894                 {
1895                     char year[64];
1896                     sprintf(year, "%d", rec_md->data.number.max);
1897
1898                     add_facet(se, (char *) type, year, term_factor);
1899                     if (rec_md->data.number.max != rec_md->data.number.min)
1900                     {
1901                         sprintf(year, "%d", rec_md->data.number.min);
1902                         add_facet(se, (char *) type, year, term_factor);
1903                     }
1904                 }
1905                 else
1906                     add_facet(se, (char *) type, (char *) value, term_factor);
1907             }
1908
1909             // cleaning up
1910             xmlFree(type);
1911             xmlFree(value);
1912             type = value = 0;
1913         }
1914         else
1915         {
1916             if (se->number_of_warnings_unknown_elements == 0)
1917                 session_log(se, YLOG_WARN,
1918                         "Unexpected element in internal record: %s", n->name);
1919             se->number_of_warnings_unknown_elements++;
1920         }
1921     }
1922     if (type)
1923         xmlFree(type);
1924     if (value)
1925         xmlFree(value);
1926
1927     relevance_donerecord(se->relevance, cluster);
1928     se->total_records++;
1929
1930     return 0;
1931 }
1932
1933 void session_log(struct session *s, int level, const char *fmt, ...)
1934 {
1935     char buf[1024];
1936     va_list ap;
1937     va_start(ap, fmt);
1938
1939     yaz_vsnprintf(buf, sizeof(buf)-30, fmt, ap);
1940     yaz_log(level, "Session %u: %s", s->session_id, buf);
1941
1942     va_end(ap);
1943 }
1944
1945 /*
1946  * Local variables:
1947  * c-basic-offset: 4
1948  * c-file-style: "Stroustrup"
1949  * indent-tabs-mode: nil
1950  * End:
1951  * vim: shiftwidth=4 tabstop=8 expandtab
1952  */
1953