src/session.c

   1 /* This file is part of Pazpar2.
   2    Copyright (C) 2006-2011 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 #define TERMLIST_HIGH_SCORE 25
  80
  81 #define MAX_CHUNK 15
  82
  83 #define MAX(a,b) ((a)>(b)?(a):(b))
  84
  85 // Note: Some things in this structure will eventually move to configuration
  86 struct parameters global_parameters =
  87 {
  88     0,   // dump_records
  89     0,   // debug_mode
  90     0,   // predictable sessions
  91 };
  92
  93 struct client_list {
  94     struct client *client;
  95     struct client_list *next;
  96 };
  97
  98 /* session counting (1) , disable client counting (0) */
  99 static YAZ_MUTEX g_session_mutex = 0;
 100 static int no_sessions = 0;
 101 static int no_session_total = 0;
 102
 103 static int session_use(int delta)
 104 {
 105     int sessions;
 106     if (!g_session_mutex)
 107         yaz_mutex_create(&g_session_mutex);
 108     yaz_mutex_enter(g_session_mutex);
 109     no_sessions += delta;
 110     if (delta > 0)
 111         no_session_total += delta;
 112     sessions = no_sessions;
 113     yaz_mutex_leave(g_session_mutex);
 114     yaz_log(YLOG_DEBUG, "%s sessions=%d", delta == 0 ? "" : (delta > 0 ? "INC" : "DEC"), no_sessions);
 115     return sessions;
 116 }
 117
 118 int sessions_count(void)
 119 {
 120     return session_use(0);
 121 }
 122
 123 int session_count_total(void)
 124 {
 125     int total = 0;
 126     if (!g_session_mutex)
 127         return 0;
 128     yaz_mutex_enter(g_session_mutex);
 129     total = no_session_total;
 130     yaz_mutex_leave(g_session_mutex);
 131     return total;
 132 }
 133
 134 static void log_xml_doc(xmlDoc *doc)
 135 {
 136     FILE *lf = yaz_log_file();
 137     xmlChar *result = 0;
 138     int len = 0;
 139 #if LIBXML_VERSION >= 20600
 140     xmlDocDumpFormatMemory(doc, &result, &len, 1);
 141 #else
 142     xmlDocDumpMemory(doc, &result, &len);
 143 #endif
 144     if (lf && len)
 145     {
 146         (void) fwrite(result, 1, len, lf);
 147         fprintf(lf, "\n");
 148     }
 149     xmlFree(result);
 150 }
 151
 152 static void session_enter(struct session *s)
 153 {
 154     yaz_mutex_enter(s->session_mutex);
 155 }
 156
 157 static void session_leave(struct session *s)
 158 {
 159     yaz_mutex_leave(s->session_mutex);
 160 }
 161
 162 static void session_normalize_facet(struct session *s, const char *type,
 163                                     const char *value,
 164                                     WRBUF display_wrbuf,
 165                                     WRBUF facet_wrbuf)
 166 {
 167     struct conf_service *service = s->service;
 168     pp2_charset_token_t prt;
 169     const char *facet_component;
 170     int i;
 171     const char *icu_chain_id = 0;
 172
 173     for (i = 0; i < service->num_metadata; i++)
 174         if (!strcmp((service->metadata + i)->name, type))
 175             icu_chain_id = (service->metadata + i)->facetrule;
 176     if (!icu_chain_id)
 177         icu_chain_id = "facet";
 178     prt = pp2_charset_token_create(service->charsets, icu_chain_id);
 179     if (!prt)
 180     {
 181         yaz_log(YLOG_FATAL, "Unknown ICU chain '%s' for facet of type '%s'",
 182                 icu_chain_id, type);
 183         wrbuf_destroy(facet_wrbuf);
 184         wrbuf_destroy(display_wrbuf);
 185         return;
 186     }
 187     pp2_charset_token_first(prt, value, 0);
 188     while ((facet_component = pp2_charset_token_next(prt)))
 189     {
 190         const char *display_component;
 191         if (*facet_component)
 192         {
 193             if (wrbuf_len(facet_wrbuf))
 194                 wrbuf_puts(facet_wrbuf, " ");
 195             wrbuf_puts(facet_wrbuf, facet_component);
 196         }
 197         display_component = pp2_get_display(prt);
 198         if (display_component)
 199         {
 200             if (wrbuf_len(display_wrbuf))
 201                 wrbuf_puts(display_wrbuf, " ");
 202             wrbuf_puts(display_wrbuf, display_component);
 203         }
 204     }
 205     pp2_charset_token_destroy(prt);
 206 }
 207
 208 void add_facet(struct session *s, const char *type, const char *value, int count)
 209 {
 210     WRBUF facet_wrbuf = wrbuf_alloc();
 211     WRBUF display_wrbuf = wrbuf_alloc();
 212
 213     session_normalize_facet(s, type, value, display_wrbuf, facet_wrbuf);
 214
 215     if (wrbuf_len(facet_wrbuf))
 216     {
 217         int i;
 218         for (i = 0; i < s->num_termlists; i++)
 219             if (!strcmp(s->termlists[i].name, type))
 220                 break;
 221         if (i == s->num_termlists)
 222         {
 223             if (i == SESSION_MAX_TERMLISTS)
 224             {
 225                 session_log(s, YLOG_FATAL, "Too many termlists");
 226                 wrbuf_destroy(facet_wrbuf);
 227                 wrbuf_destroy(display_wrbuf);
 228                 return;
 229             }
 230
 231             s->termlists[i].name = nmem_strdup(s->nmem, type);
 232             s->termlists[i].termlist
 233                 = termlist_create(s->nmem, TERMLIST_HIGH_SCORE);
 234             s->num_termlists = i + 1;
 235         }
 236
 237 #if 0
 238         session_log(s, YLOG_DEBUG, "Facets for %s: %s norm:%s (%d)", type, value, wrbuf_cstr(facet_wrbuf), count);
 239 #endif
 240         termlist_insert(s->termlists[i].termlist, wrbuf_cstr(display_wrbuf),
 241                         wrbuf_cstr(facet_wrbuf), count);
 242     }
 243     wrbuf_destroy(facet_wrbuf);
 244     wrbuf_destroy(display_wrbuf);
 245 }
 246
 247 static xmlDoc *record_to_xml(struct session *se,
 248                              struct session_database *sdb, const char *rec)
 249 {
 250     struct database *db = sdb->database;
 251     xmlDoc *rdoc = 0;
 252
 253     rdoc = xmlParseMemory(rec, strlen(rec));
 254
 255     if (!rdoc)
 256     {
 257         session_log(se, YLOG_FATAL, "Non-wellformed XML received from %s",
 258                     db->id);
 259         return 0;
 260     }
 261
 262     if (global_parameters.dump_records)
 263     {
 264         session_log(se, YLOG_LOG, "Un-normalized record from %s", db->id);
 265         log_xml_doc(rdoc);
 266     }
 267
 268     return rdoc;
 269 }
 270
 271 #define MAX_XSLT_ARGS 16
 272
 273 // Add static values from session database settings if applicable
 274 static void insert_settings_parameters(struct session_database *sdb,
 275                                        struct conf_service *service,
 276                                        char **parms,
 277                                        NMEM nmem)
 278 {
 279     int i;
 280     int nparms = 0;
 281     int offset = 0;
 282
 283     for (i = 0; i < service->num_metadata; i++)
 284     {
 285         struct conf_metadata *md = &service->metadata[i];
 286         int setting;
 287
 288         if (md->setting == Metadata_setting_parameter &&
 289             (setting = settings_lookup_offset(service, md->name)) >= 0)
 290         {
 291             const char *val = session_setting_oneval(sdb, setting);
 292             if (val && nparms < MAX_XSLT_ARGS)
 293             {
 294                 char *buf;
 295                 int len = strlen(val);
 296                 buf = nmem_malloc(nmem, len + 3);
 297                 buf[0] = '\'';
 298                 strcpy(buf + 1, val);
 299                 buf[len+1] = '\'';
 300                 buf[len+2] = '\0';
 301                 parms[offset++] = md->name;
 302                 parms[offset++] = buf;
 303                 nparms++;
 304             }
 305         }
 306     }
 307     parms[offset] = 0;
 308 }
 309
 310 // Add static values from session database settings if applicable
 311 static void insert_settings_values(struct session_database *sdb, xmlDoc *doc,
 312     struct conf_service *service)
 313 {
 314     int i;
 315
 316     for (i = 0; i < service->num_metadata; i++)
 317     {
 318         struct conf_metadata *md = &service->metadata[i];
 319         int offset;
 320
 321         if (md->setting == Metadata_setting_postproc &&
 322             (offset = settings_lookup_offset(service, md->name)) >= 0)
 323         {
 324             const char *val = session_setting_oneval(sdb, offset);
 325             if (val)
 326             {
 327                 xmlNode *r = xmlDocGetRootElement(doc);
 328                 xmlNode *n = xmlNewTextChild(r, 0, (xmlChar *) "metadata",
 329                                              (xmlChar *) val);
 330                 xmlSetProp(n, (xmlChar *) "type", (xmlChar *) md->name);
 331             }
 332         }
 333     }
 334 }
 335
 336 static xmlDoc *normalize_record(struct session *se,
 337                                 struct session_database *sdb,
 338                                 struct conf_service *service,
 339                                 const char *rec, NMEM nmem)
 340 {
 341     xmlDoc *rdoc = record_to_xml(se, sdb, rec);
 342
 343     if (rdoc)
 344     {
 345         char *parms[MAX_XSLT_ARGS*2+1];
 346
 347         insert_settings_parameters(sdb, service, parms, nmem);
 348
 349         if (normalize_record_transform(sdb->map, &rdoc, (const char **)parms))
 350         {
 351             session_log(se, YLOG_WARN, "Normalize failed from %s",
 352                         sdb->database->id);
 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     const char *s;
 410
 411     if (sdb->settings && sdb->settings[PZ_XSLT] && !sdb->map &&
 412         (s = session_setting_oneval(sdb, PZ_XSLT)))
 413     {
 414         char auto_stylesheet[256];
 415
 416         if (!strcmp(s, "auto"))
 417         {
 418             const char *request_syntax = session_setting_oneval(
 419                 sdb, PZ_REQUESTSYNTAX);
 420             if (request_syntax)
 421             {
 422                 char *cp;
 423                 yaz_snprintf(auto_stylesheet, sizeof(auto_stylesheet),
 424                              "%s.xsl", request_syntax);
 425                 for (cp = auto_stylesheet; *cp; cp++)
 426                 {
 427                     /* deliberately only consider ASCII */
 428                     if (*cp > 32 && *cp < 127)
 429                         *cp = tolower(*cp);
 430                 }
 431                 s = auto_stylesheet;
 432             }
 433             else
 434             {
 435                 session_log(se, YLOG_WARN,
 436                             "No pz:requestsyntax for auto stylesheet");
 437             }
 438         }
 439         sdb->map = normalize_cache_get(se->normalize_cache,
 440                                        se->service, s);
 441         if (!sdb->map)
 442             return -1;
 443     }
 444     return 0;
 445 }
 446
 447 // called if watch should be removed because http_channel is to be destroyed
 448 static void session_watch_cancel(void *data, struct http_channel *c,
 449                                  void *data2)
 450 {
 451     struct session_watchentry *ent = data;
 452
 453     ent->fun = 0;
 454     ent->data = 0;
 455     ent->obs = 0;
 456 }
 457
 458 // set watch. Returns 0=OK, -1 if watch is already set
 459 int session_set_watch(struct session *s, int what,
 460                       session_watchfun fun, void *data,
 461                       struct http_channel *chan)
 462 {
 463     int ret;
 464     session_enter(s);
 465     if (s->watchlist[what].fun)
 466         ret = -1;
 467     else
 468     {
 469
 470         s->watchlist[what].fun = fun;
 471         s->watchlist[what].data = data;
 472         s->watchlist[what].obs = http_add_observer(chan, &s->watchlist[what],
 473                                                    session_watch_cancel);
 474         ret = 0;
 475     }
 476     session_leave(s);
 477     return ret;
 478 }
 479
 480 void session_alert_watch(struct session *s, int what)
 481 {
 482     assert(s);
 483     session_enter(s);
 484     if (s->watchlist[what].fun)
 485     {
 486         /* our watch is no longer associated with http_channel */
 487         void *data;
 488         session_watchfun fun;
 489
 490         http_remove_observer(s->watchlist[what].obs);
 491         fun  = s->watchlist[what].fun;
 492         data = s->watchlist[what].data;
 493
 494         /* reset watch before fun is invoked - in case fun wants to set
 495            it again */
 496         s->watchlist[what].fun = 0;
 497         s->watchlist[what].data = 0;
 498         s->watchlist[what].obs = 0;
 499
 500         session_leave(s);
 501         session_log(s, YLOG_DEBUG,
 502                     "Alert Watch: %d calling function: %p", what, fun);
 503         fun(data);
 504     }
 505     else
 506         session_leave(s);
 507 }
 508
 509 //callback for grep_databases
 510 static void select_targets_callback(struct session *se,
 511                                     struct session_database *db)
 512 {
 513     struct client *cl;
 514     struct client_list *l;
 515
 516     for (l = se->clients_cached; l; l = l->next)
 517         if (client_get_database(l->client) == db)
 518             break;
 519
 520     if (l)
 521         cl = l->client;
 522     else
 523     {
 524         cl = client_create(db->database->id);
 525         client_set_database(cl, db);
 526
 527         l = xmalloc(sizeof(*l));
 528         l->client = cl;
 529         l->next = se->clients_cached;
 530         se->clients_cached = l;
 531     }
 532     /* set session always. If may be 0 if client is not active */
 533     client_set_session(cl, se);
 534
 535     l = xmalloc(sizeof(*l));
 536     l->client = cl;
 537     l->next = se->clients_active;
 538     se->clients_active = l;
 539 }
 540
 541 static void session_reset_active_clients(struct session *se,
 542                                          struct client_list *new_list)
 543 {
 544     struct client_list *l;
 545
 546     session_enter(se);
 547     l = se->clients_active;
 548     se->clients_active = new_list;
 549     session_leave(se);
 550
 551     while (l)
 552     {
 553         struct client_list *l_next = l->next;
 554
 555         client_lock(l->client);
 556         client_set_session(l->client, 0); /* mark client inactive */
 557         client_unlock(l->client);
 558
 559         xfree(l);
 560         l = l_next;
 561     }
 562 }
 563
 564 static void session_remove_cached_clients(struct session *se)
 565 {
 566     struct client_list *l;
 567
 568     session_reset_active_clients(se, 0);
 569
 570     session_enter(se);
 571     l = se->clients_cached;
 572     se->clients_cached = 0;
 573     session_leave(se);
 574
 575     while (l)
 576     {
 577         struct client_list *l_next = l->next;
 578         client_lock(l->client);
 579         client_set_session(l->client, 0);
 580         client_set_database(l->client, 0);
 581         client_unlock(l->client);
 582         client_destroy(l->client);
 583         xfree(l);
 584         l = l_next;
 585     }
 586 }
 587
 588 // Associates a set of clients with a session;
 589 // Note: Session-databases represent databases with per-session
 590 // setting overrides
 591 static int select_targets(struct session *se, const char *filter)
 592 {
 593     return session_grep_databases(se, filter, select_targets_callback);
 594 }
 595
 596 int session_active_clients(struct session *s)
 597 {
 598     struct client_list *l;
 599     int res = 0;
 600
 601     for (l = s->clients_active; l; l = l->next)
 602         if (client_is_active(l->client))
 603             res++;
 604
 605     return res;
 606 }
 607
 608 int session_is_preferred_clients_ready(struct session *s)
 609 {
 610     struct client_list *l;
 611     int res = 0;
 612
 613     for (l = s->clients_active; l; l = l->next)
 614         if (client_is_active_preferred(l->client))
 615             res++;
 616     session_log(s, YLOG_DEBUG, "Has %d active preferred clients.", res);
 617     return res == 0;
 618 }
 619
 620 void session_sort(struct session *se, const char *field, int increasing)
 621 {
 622     struct session_sorted_results *sr;
 623     struct client_list *l;
 624
 625     session_enter(se);
 626
 627     /* see if we already have sorted for this critieria */
 628     for (sr = se->sorted_results; sr; sr = sr->next)
 629     {
 630         if (!strcmp(field, sr->field) && increasing == sr->increasing)
 631             break;
 632     }
 633     if (sr)
 634     {
 635         yaz_log(YLOG_LOG, "search_sort: field=%s increasing=%d already fetched",
 636                 field, increasing);
 637         session_leave(se);
 638         return;
 639     }
 640     yaz_log(YLOG_LOG, "search_sort: field=%s increasing=%d must fetch",
 641             field, increasing);
 642     sr = nmem_malloc(se->nmem, sizeof(*sr));
 643     sr->field = nmem_strdup(se->nmem, field);
 644     sr->increasing = increasing;
 645     sr->next = se->sorted_results;
 646     se->sorted_results = sr;
 647
 648     for (l = se->clients_active; l; l = l->next)
 649     {
 650         struct client *cl = l->client;
 651         struct timeval tval;
 652         if (client_prep_connection(cl, se->service->z3950_operation_timeout,
 653                                    se->service->z3950_session_timeout,
 654                                    se->service->server->iochan_man,
 655                                    &tval))
 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
 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].diagnostic =
 968             client_get_diagnostic(cl, &res[*count].addinfo);
 969         res[*count].state = client_get_state_str(cl);
 970         res[*count].connected  = client_get_connection(cl) ? 1 : 0;
 971         session_settings_dump(se, client_get_database(cl), w);
 972         res[*count].settings_xml = nmem_strdup(nmem, wrbuf_cstr(w));
 973         wrbuf_rewind(w);
 974         wrbuf_puts(w, "");
 975         res[*count].suggestions_xml = nmem_strdup(nmem, client_get_suggestions_xml(cl, w));
 976         wrbuf_destroy(w);
 977         (*count)++;
 978     }
 979     return res;
 980 }
 981
 982 struct hitsbytarget *get_hitsbytarget(struct session *se, int *count, NMEM nmem)
 983 {
 984     struct hitsbytarget *p;
 985     session_enter(se);
 986     p = hitsbytarget_nb(se, count, nmem);
 987     session_leave(se);
 988     return p;
 989 }
 990
 991 struct termlist_score **get_termlist_score(struct session *se,
 992                                            const char *name, int *num)
 993 {
 994     int i;
 995     struct termlist_score **tl = 0;
 996
 997     session_enter(se);
 998     for (i = 0; i < se->num_termlists; i++)
 999         if (!strcmp((const char *) se->termlists[i].name, name))
1000         {
1001             tl = termlist_highscore(se->termlists[i].termlist, num);
1002             break;
1003         }
1004     session_leave(se);
1005     return tl;
1006 }
1007
1008 // Compares two hitsbytarget nodes by hitcount
1009 static int cmp_ht(const void *p1, const void *p2)
1010 {
1011     const struct hitsbytarget *h1 = p1;
1012     const struct hitsbytarget *h2 = p2;
1013     return h2->hits - h1->hits;
1014 }
1015
1016 static int targets_termlist_nb(WRBUF wrbuf, struct session *se, int num,
1017                                NMEM nmem)
1018 {
1019     struct hitsbytarget *ht;
1020     int count, i;
1021
1022     ht = hitsbytarget_nb(se, &count, nmem);
1023     qsort(ht, count, sizeof(struct hitsbytarget), cmp_ht);
1024     for (i = 0; i < count && i < num && ht[i].hits > 0; i++)
1025     {
1026
1027         // do only print terms which have display names
1028
1029         wrbuf_puts(wrbuf, "<term>\n");
1030
1031         wrbuf_puts(wrbuf, "<id>");
1032         wrbuf_xmlputs(wrbuf, ht[i].id);
1033         wrbuf_puts(wrbuf, "</id>\n");
1034
1035         wrbuf_puts(wrbuf, "<name>");
1036         if (!ht[i].name || !ht[i].name[0])
1037             wrbuf_xmlputs(wrbuf, "NO TARGET NAME");
1038         else
1039             wrbuf_xmlputs(wrbuf, ht[i].name);
1040         wrbuf_puts(wrbuf, "</name>\n");
1041
1042         wrbuf_printf(wrbuf, "<frequency>" ODR_INT_PRINTF "</frequency>\n",
1043                      ht[i].hits);
1044
1045         wrbuf_puts(wrbuf, "<state>");
1046         wrbuf_xmlputs(wrbuf, ht[i].state);
1047         wrbuf_puts(wrbuf, "</state>\n");
1048
1049         wrbuf_printf(wrbuf, "<diagnostic>%d</diagnostic>\n",
1050                      ht[i].diagnostic);
1051         wrbuf_puts(wrbuf, "</term>\n");
1052     }
1053     return count;
1054 }
1055
1056 void perform_termlist(struct http_channel *c, struct session *se,
1057                       const char *name, int num)
1058 {
1059     int i, j;
1060     NMEM nmem_tmp = nmem_create();
1061     char **names;
1062     int num_names = 0;
1063
1064     if (!name)
1065         name = "*";
1066
1067     nmem_strsplit(nmem_tmp, ",", name, &names, &num_names);
1068
1069     session_enter(se);
1070
1071     for (j = 0; j < num_names; j++)
1072     {
1073         const char *tname;
1074
1075         for (i = 0; i < se->num_termlists; i++)
1076         {
1077             tname = se->termlists[i].name;
1078             if (!strcmp(names[j], tname) || !strcmp(names[j], "*"))
1079             {
1080                 struct termlist_score **p = 0;
1081                 int len;
1082
1083                 wrbuf_puts(c->wrbuf, "<list name=\"");
1084                 wrbuf_xmlputs(c->wrbuf, tname);
1085                 wrbuf_puts(c->wrbuf, "\">\n");
1086
1087                 p = termlist_highscore(se->termlists[i].termlist, &len);
1088                 if (p)
1089                 {
1090                     int i;
1091                     for (i = 0; i < len && i < num; i++)
1092                     {
1093                         // prevent sending empty term elements
1094                         if (!p[i]->display_term || !p[i]->display_term[0])
1095                             continue;
1096
1097                         wrbuf_puts(c->wrbuf, "<term>");
1098                         wrbuf_puts(c->wrbuf, "<name>");
1099                         wrbuf_xmlputs(c->wrbuf, p[i]->display_term);
1100                         wrbuf_puts(c->wrbuf, "</name>");
1101
1102                         wrbuf_printf(c->wrbuf,
1103                                      "<frequency>%d</frequency>",
1104                                      p[i]->frequency);
1105                         wrbuf_puts(c->wrbuf, "</term>\n");
1106                     }
1107                 }
1108                 wrbuf_puts(c->wrbuf, "</list>\n");
1109             }
1110         }
1111         tname = "xtargets";
1112         if (!strcmp(names[j], tname) || !strcmp(names[j], "*"))
1113         {
1114             wrbuf_puts(c->wrbuf, "<list name=\"");
1115             wrbuf_xmlputs(c->wrbuf, tname);
1116             wrbuf_puts(c->wrbuf, "\">\n");
1117
1118             targets_termlist_nb(c->wrbuf, se, num, c->nmem);
1119             wrbuf_puts(c->wrbuf, "</list>\n");
1120         }
1121     }
1122     session_leave(se);
1123     nmem_destroy(nmem_tmp);
1124 }
1125
1126 #ifdef MISSING_HEADERS
1127 void report_nmem_stats(void)
1128 {
1129     size_t in_use, is_free;
1130
1131     nmem_get_memory_in_use(&in_use);
1132     nmem_get_memory_free(&is_free);
1133
1134     yaz_log(YLOG_LOG, "nmem stat: use=%ld free=%ld",
1135             (long) in_use, (long) is_free);
1136 }
1137 #endif
1138
1139 struct record_cluster *show_single_start(struct session *se, const char *id,
1140                                          struct record_cluster **prev_r,
1141                                          struct record_cluster **next_r)
1142 {
1143     struct record_cluster *r = 0;
1144
1145     session_enter(se);
1146     *prev_r = 0;
1147     *next_r = 0;
1148     if (se->reclist)
1149     {
1150         reclist_enter(se->reclist);
1151         while ((r = reclist_read_record(se->reclist)))
1152         {
1153             if (!strcmp(r->recid, id))
1154             {
1155                 *next_r = reclist_read_record(se->reclist);
1156                 break;
1157             }
1158             *prev_r = r;
1159         }
1160         reclist_leave(se->reclist);
1161     }
1162     if (!r)
1163         session_leave(se);
1164     return r;
1165 }
1166
1167 void show_single_stop(struct session *se, struct record_cluster *rec)
1168 {
1169     session_leave(se);
1170 }
1171
1172 struct record_cluster **show_range_start(struct session *se,
1173                                          struct reclist_sortparms *sp,
1174                                          int start, int *num, int *total, Odr_int *sumhits)
1175 {
1176     struct record_cluster **recs;
1177     struct reclist_sortparms *spp;
1178     int i;
1179 #if USE_TIMING
1180     yaz_timing_t t = yaz_timing_create();
1181 #endif
1182     session_enter(se);
1183     recs = nmem_malloc(se->nmem, *num * sizeof(struct record_cluster *));
1184     if (!se->relevance)
1185     {
1186         *num = 0;
1187         *total = 0;
1188         *sumhits = 0;
1189         recs = 0;
1190     }
1191     else
1192     {
1193         struct client_list *l;
1194
1195         for (spp = sp; spp; spp = spp->next)
1196             if (spp->type == Metadata_sortkey_relevance)
1197             {
1198                 relevance_prepare_read(se->relevance, se->reclist);
1199                 break;
1200             }
1201         reclist_sort(se->reclist, sp);
1202
1203         reclist_enter(se->reclist);
1204         *total = reclist_get_num_records(se->reclist);
1205
1206         *sumhits = 0;
1207         for (l = se->clients_active; l; l = l->next)
1208             *sumhits += client_get_hits(l->client);
1209
1210         for (i = 0; i < start; i++)
1211             if (!reclist_read_record(se->reclist))
1212             {
1213                 *num = 0;
1214                 recs = 0;
1215                 break;
1216             }
1217
1218         for (i = 0; i < *num; i++)
1219         {
1220             struct record_cluster *r = reclist_read_record(se->reclist);
1221             if (!r)
1222             {
1223                 *num = i;
1224                 break;
1225             }
1226             recs[i] = r;
1227         }
1228         reclist_leave(se->reclist);
1229     }
1230 #if USE_TIMING
1231     yaz_timing_stop(t);
1232     yaz_log(YLOG_LOG, "show %6.5f %3.2f %3.2f",
1233             yaz_timing_get_real(t), yaz_timing_get_user(t),
1234             yaz_timing_get_sys(t));
1235     yaz_timing_destroy(&t);
1236 #endif
1237     return recs;
1238 }
1239
1240 void show_range_stop(struct session *se, struct record_cluster **recs)
1241 {
1242     session_leave(se);
1243 }
1244
1245 void statistics(struct session *se, struct statistics *stat)
1246 {
1247     struct client_list *l;
1248     int count = 0;
1249
1250     memset(stat, 0, sizeof(*stat));
1251     stat->num_hits = 0;
1252     for (l = se->clients_active; l; l = l->next)
1253     {
1254         struct client *cl = l->client;
1255         if (!client_get_connection(cl))
1256             stat->num_no_connection++;
1257         stat->num_hits += client_get_hits(cl);
1258         switch (client_get_state(cl))
1259         {
1260         case Client_Connecting: stat->num_connecting++; break;
1261         case Client_Working: stat->num_working++; break;
1262         case Client_Idle: stat->num_idle++; break;
1263         case Client_Failed: stat->num_failed++; break;
1264         case Client_Error: stat->num_error++; break;
1265         default: break;
1266         }
1267         count++;
1268     }
1269     stat->num_records = se->total_records;
1270
1271     stat->num_clients = count;
1272 }
1273
1274 static struct record_metadata *record_metadata_init(
1275     NMEM nmem, const char *value, enum conf_metadata_type type,
1276     struct _xmlAttr *attr)
1277 {
1278     struct record_metadata *rec_md = record_metadata_create(nmem);
1279     struct record_metadata_attr **attrp = &rec_md->attributes;
1280
1281     for (; attr; attr = attr->next)
1282     {
1283         if (attr->children && attr->children->content)
1284         {
1285             if (strcmp((const char *) attr->name, "type"))
1286             {  /* skip the "type" attribute.. Its value is already part of
1287                   the element in output (md-%s) and so repeating it here
1288                   is redundant */
1289                 *attrp = nmem_malloc(nmem, sizeof(**attrp));
1290                 (*attrp)->name =
1291                     nmem_strdup(nmem, (const char *) attr->name);
1292                 (*attrp)->value =
1293                     nmem_strdup(nmem, (const char *) attr->children->content);
1294                 attrp = &(*attrp)->next;
1295             }
1296         }
1297     }
1298     *attrp = 0;
1299
1300     if (type == Metadata_type_generic)
1301     {
1302         char *p = nmem_strdup(nmem, value);
1303
1304         p = normalize7bit_generic(p, " ,/.:([");
1305
1306         rec_md->data.text.disp = p;
1307         rec_md->data.text.sort = 0;
1308     }
1309     else if (type == Metadata_type_year || type == Metadata_type_date)
1310     {
1311         int first, last;
1312         int longdate = 0;
1313
1314         if (type == Metadata_type_date)
1315             longdate = 1;
1316         if (extract7bit_dates((char *) value, &first, &last, longdate) < 0)
1317             return 0;
1318
1319         rec_md->data.number.min = first;
1320         rec_md->data.number.max = last;
1321     }
1322     else
1323         return 0;
1324     return rec_md;
1325 }
1326
1327 static int get_mergekey_from_doc(xmlDoc *doc, xmlNode *root, const char *name,
1328                                  struct conf_service *service, WRBUF norm_wr)
1329 {
1330     xmlNode *n;
1331     int no_found = 0;
1332     for (n = root->children; n; n = n->next)
1333     {
1334         if (n->type != XML_ELEMENT_NODE)
1335             continue;
1336         if (!strcmp((const char *) n->name, "metadata"))
1337         {
1338             xmlChar *type = xmlGetProp(n, (xmlChar *) "type");
1339             if (type == NULL) {
1340                 yaz_log(YLOG_FATAL, "Missing type attribute on metadata element. Skipping!");
1341             }
1342             else if (!strcmp(name, (const char *) type))
1343             {
1344                 xmlChar *value = xmlNodeListGetString(doc, n->children, 1);
1345                 if (value)
1346                 {
1347                     const char *norm_str;
1348                     pp2_charset_token_t prt =
1349                         pp2_charset_token_create(service->charsets, "mergekey");
1350
1351                     pp2_charset_token_first(prt, (const char *) value, 0);
1352                     if (wrbuf_len(norm_wr) > 0)
1353                         wrbuf_puts(norm_wr, " ");
1354                     wrbuf_puts(norm_wr, name);
1355                     while ((norm_str =
1356                             pp2_charset_token_next(prt)))
1357                     {
1358                         if (*norm_str)
1359                         {
1360                             wrbuf_puts(norm_wr, " ");
1361                             wrbuf_puts(norm_wr, norm_str);
1362                         }
1363                     }
1364                     xmlFree(value);
1365                     pp2_charset_token_destroy(prt);
1366                     no_found++;
1367                 }
1368             }
1369             xmlFree(type);
1370         }
1371     }
1372     return no_found;
1373 }
1374
1375 static const char *get_mergekey(xmlDoc *doc, struct client *cl, int record_no,
1376                                 struct conf_service *service, NMEM nmem)
1377 {
1378     char *mergekey_norm = 0;
1379     xmlNode *root = xmlDocGetRootElement(doc);
1380     WRBUF norm_wr = wrbuf_alloc();
1381
1382     /* consider mergekey from XSL first */
1383     xmlChar *mergekey = xmlGetProp(root, (xmlChar *) "mergekey");
1384     if (mergekey)
1385     {
1386         const char *norm_str;
1387         pp2_charset_token_t prt =
1388             pp2_charset_token_create(service->charsets, "mergekey");
1389
1390         pp2_charset_token_first(prt, (const char *) mergekey, 0);
1391         while ((norm_str = pp2_charset_token_next(prt)))
1392         {
1393             if (*norm_str)
1394             {
1395                 if (wrbuf_len(norm_wr))
1396                     wrbuf_puts(norm_wr, " ");
1397                 wrbuf_puts(norm_wr, norm_str);
1398             }
1399         }
1400         pp2_charset_token_destroy(prt);
1401         xmlFree(mergekey);
1402     }
1403     else
1404     {
1405         /* no mergekey defined in XSL. Look for mergekey metadata instead */
1406         int field_id;
1407         for (field_id = 0; field_id < service->num_metadata; field_id++)
1408         {
1409             struct conf_metadata *ser_md = &service->metadata[field_id];
1410             if (ser_md->mergekey != Metadata_mergekey_no)
1411             {
1412                 int r = get_mergekey_from_doc(doc, root, ser_md->name,
1413                                               service, norm_wr);
1414                 if (r == 0 && ser_md->mergekey == Metadata_mergekey_required)
1415                 {
1416                     /* no mergekey on this one and it is required..
1417                        Generate unique key instead */
1418                     wrbuf_rewind(norm_wr);
1419                     break;
1420                 }
1421             }
1422         }
1423     }
1424
1425     /* generate unique key if none is not generated already or is empty */
1426     if (wrbuf_len(norm_wr) == 0)
1427     {
1428         wrbuf_printf(norm_wr, "position: %s-%d",
1429                      client_get_id(cl), record_no);
1430     }
1431     else
1432     {
1433         const char *lead = "content: ";
1434         wrbuf_insert(norm_wr, 0, lead, strlen(lead));
1435     }
1436     if (wrbuf_len(norm_wr) > 0)
1437         mergekey_norm = nmem_strdup(nmem, wrbuf_cstr(norm_wr));
1438     wrbuf_destroy(norm_wr);
1439     return mergekey_norm;
1440 }
1441
1442 /** \brief see if metadata for pz:recordfilter exists
1443     \param root xml root element of normalized record
1444     \param sdb session database for client
1445     \retval 0 if there is no metadata for pz:recordfilter
1446     \retval 1 if there is metadata for pz:recordfilter
1447
1448     If there is no pz:recordfilter defined, this function returns 1
1449     as well.
1450 */
1451
1452 static int check_record_filter(xmlNode *root, struct session_database *sdb)
1453 {
1454     int match = 0;
1455     xmlNode *n;
1456     const char *s;
1457     s = session_setting_oneval(sdb, PZ_RECORDFILTER);
1458
1459     if (!s || !*s)
1460         return 1;
1461
1462     for (n = root->children; n; n = n->next)
1463     {
1464         if (n->type != XML_ELEMENT_NODE)
1465             continue;
1466         if (!strcmp((const char *) n->name, "metadata"))
1467         {
1468             xmlChar *type = xmlGetProp(n, (xmlChar *) "type");
1469             if (type)
1470             {
1471                 size_t len;
1472                 int substring;
1473                 const char *eq;
1474
1475                 if ((eq = strchr(s, '=')))
1476                     substring = 0;
1477                 else if ((eq = strchr(s, '~')))
1478                     substring = 1;
1479                 if (eq)
1480                     len = eq - s;
1481                 else
1482                     len = strlen(s);
1483                 if (len == strlen((const char *)type) &&
1484                     !memcmp((const char *) type, s, len))
1485                 {
1486                     xmlChar *value = xmlNodeGetContent(n);
1487                     if (value && *value)
1488                     {
1489                         if (!eq ||
1490                             (substring && strstr((const char *) value, eq+1)) ||
1491                             (!substring && !strcmp((const char *) value, eq + 1)))
1492                             match = 1;
1493                     }
1494                     xmlFree(value);
1495                 }
1496                 xmlFree(type);
1497             }
1498         }
1499     }
1500     return match;
1501 }
1502
1503
1504 static int ingest_to_cluster(struct client *cl,
1505                              xmlDoc *xdoc,
1506                              xmlNode *root,
1507                              int record_no,
1508                              const char *mergekey_norm);
1509
1510 /** \brief ingest XML record
1511     \param cl client holds the result set for record
1512     \param rec record buffer (0 terminated)
1513     \param record_no record position (1, 2, ..)
1514     \param nmem working NMEM
1515     \retval 0 OK
1516     \retval -1 failure
1517     \retval -2 Filtered
1518 */
1519 int ingest_record(struct client *cl, const char *rec,
1520                   int record_no, NMEM nmem)
1521 {
1522     struct session *se = client_get_session(cl);
1523     int ret = 0;
1524     struct session_database *sdb = client_get_database(cl);
1525     struct conf_service *service = se->service;
1526     xmlDoc *xdoc = normalize_record(se, sdb, service, rec, nmem);
1527     xmlNode *root;
1528     const char *mergekey_norm;
1529
1530     if (!xdoc)
1531         return -1;
1532
1533     root = xmlDocGetRootElement(xdoc);
1534
1535     if (!check_record_filter(root, sdb))
1536     {
1537         session_log(se, YLOG_LOG, "Filtered out record no %d from %s",
1538                     record_no, sdb->database->id);
1539         xmlFreeDoc(xdoc);
1540         return -2;
1541     }
1542
1543     mergekey_norm = get_mergekey(xdoc, cl, record_no, service, nmem);
1544     if (!mergekey_norm)
1545     {
1546         session_log(se, YLOG_WARN, "Got no mergekey");
1547         xmlFreeDoc(xdoc);
1548         return -1;
1549     }
1550     session_enter(se);
1551     if (client_get_session(cl) == se)
1552         ret = ingest_to_cluster(cl, xdoc, root, record_no, mergekey_norm);
1553     session_leave(se);
1554
1555     xmlFreeDoc(xdoc);
1556     return ret;
1557 }
1558
1559 static int check_limit_local(struct client *cl,
1560                              struct record *record,
1561                              int record_no)
1562 {
1563     int skip_record = 0;
1564     struct session *se = client_get_session(cl);
1565     struct conf_service *service = se->service;
1566     NMEM nmem_tmp = nmem_create();
1567     struct session_database *sdb = client_get_database(cl);
1568     int l = 0;
1569     while (!skip_record)
1570     {
1571         struct conf_metadata *ser_md = 0;
1572         struct record_metadata *rec_md = 0;
1573         int md_field_id;
1574         char **values = 0;
1575         int i, num_v = 0;
1576
1577         const char *name =
1578             client_get_facet_limit_local(cl, sdb, &l, nmem_tmp, &num_v,
1579                                          &values);
1580         if (!name)
1581             break;
1582
1583         md_field_id = conf_service_metadata_field_id(service, name);
1584         if (md_field_id < 0)
1585         {
1586             skip_record = 1;
1587             break;
1588         }
1589         ser_md = &service->metadata[md_field_id];
1590         rec_md = record->metadata[md_field_id];
1591         yaz_log(YLOG_LOG, "check limit local %s", name);
1592         for (i = 0; i < num_v; )
1593         {
1594             if (rec_md)
1595             {
1596                 if (ser_md->type == Metadata_type_year
1597                     || ser_md->type == Metadata_type_date)
1598                 {
1599                     int y = atoi(values[i]);
1600                     if (y >= rec_md->data.number.min
1601                         && y <= rec_md->data.number.max)
1602                         break;
1603                 }
1604                 else
1605                 {
1606                     yaz_log(YLOG_LOG, "cmp: '%s' '%s'",
1607                             rec_md->data.text.disp, values[i]);
1608                     if (!strcmp(rec_md->data.text.disp, values[i]))
1609                     {
1610                         break;
1611                     }
1612                 }
1613                 rec_md = rec_md->next;
1614             }
1615             else
1616             {
1617                 rec_md = record->metadata[md_field_id];
1618                 i++;
1619             }
1620         }
1621         if (i == num_v)
1622         {
1623             skip_record = 1;
1624             break;
1625         }
1626     }
1627     nmem_destroy(nmem_tmp);
1628     return skip_record;
1629 }
1630
1631 static int ingest_to_cluster(struct client *cl,
1632                              xmlDoc *xdoc,
1633                              xmlNode *root,
1634                              int record_no,
1635                              const char *mergekey_norm)
1636 {
1637     xmlNode *n;
1638     xmlChar *type = 0;
1639     xmlChar *value = 0;
1640     struct session *se = client_get_session(cl);
1641     struct conf_service *service = se->service;
1642     int term_factor = 1;
1643     struct record_cluster *cluster;
1644     struct session_database *sdb = client_get_database(cl);
1645     struct record *record = record_create(se->nmem,
1646                                           service->num_metadata,
1647                                           service->num_sortkeys, cl,
1648                                           record_no);
1649
1650     for (n = root->children; n; n = n->next)
1651     {
1652         if (type)
1653             xmlFree(type);
1654         if (value)
1655             xmlFree(value);
1656         type = value = 0;
1657
1658         if (n->type != XML_ELEMENT_NODE)
1659             continue;
1660         if (!strcmp((const char *) n->name, "metadata"))
1661         {
1662             struct conf_metadata *ser_md = 0;
1663             struct record_metadata **wheretoput = 0;
1664             struct record_metadata *rec_md = 0;
1665             int md_field_id = -1;
1666
1667             type = xmlGetProp(n, (xmlChar *) "type");
1668             value = xmlNodeListGetString(xdoc, n->children, 1);
1669
1670             if (!type || !value || !*value)
1671                 continue;
1672
1673             md_field_id
1674                 = conf_service_metadata_field_id(service, (const char *) type);
1675             if (md_field_id < 0)
1676             {
1677                 if (se->number_of_warnings_unknown_metadata == 0)
1678                 {
1679                     session_log(se, YLOG_WARN,
1680                             "Ignoring unknown metadata element: %s", type);
1681                 }
1682                 se->number_of_warnings_unknown_metadata++;
1683                 continue;
1684             }
1685
1686             ser_md = &service->metadata[md_field_id];
1687
1688             // non-merged metadata
1689             rec_md = record_metadata_init(se->nmem, (const char *) value,
1690                                           ser_md->type, n->properties);
1691             if (!rec_md)
1692             {
1693                 session_log(se, YLOG_WARN, "bad metadata data '%s' "
1694                             "for element '%s'", value, type);
1695                 continue;
1696             }
1697             wheretoput = &record->metadata[md_field_id];
1698             while (*wheretoput)
1699                 wheretoput = &(*wheretoput)->next;
1700             *wheretoput = rec_md;
1701         }
1702     }
1703
1704     if (check_limit_local(cl, record, record_no))
1705     {
1706         session_log(se, YLOG_LOG, "Facet filtered out record no %d from %s",
1707                     record_no, sdb->database->id);
1708         if (type)
1709             xmlFree(type);
1710         if (value)
1711             xmlFree(value);
1712         return -2;
1713     }
1714     cluster = reclist_insert(se->reclist, service, record,
1715                              mergekey_norm, &se->total_merged);
1716     if (!cluster)
1717         return -1;
1718
1719     {
1720         const char *use_term_factor_str =
1721             session_setting_oneval(sdb, PZ_TERMLIST_TERM_FACTOR);
1722         if (use_term_factor_str && use_term_factor_str[0] == '1')
1723         {
1724             int maxrecs = client_get_maxrecs(cl);
1725             int hits = (int) client_get_hits(cl);
1726             term_factor = MAX(hits, maxrecs) /  MAX(1, maxrecs);
1727             assert(term_factor >= 1);
1728             yaz_log(YLOG_DEBUG, "Using term factor: %d (%d / %d)", term_factor, MAX(hits, maxrecs), MAX(1, maxrecs));
1729         }
1730     }
1731
1732     if (global_parameters.dump_records)
1733         session_log(se, YLOG_LOG, "Cluster id %s from %s (#%d)", cluster->recid,
1734                     sdb->database->id, record_no);
1735
1736
1737     relevance_newrec(se->relevance, cluster);
1738
1739     // now parsing XML record and adding data to cluster or record metadata
1740     for (n = root->children; n; n = n->next)
1741     {
1742         pp2_charset_token_t prt;
1743         if (type)
1744             xmlFree(type);
1745         if (value)
1746             xmlFree(value);
1747         type = value = 0;
1748
1749         if (n->type != XML_ELEMENT_NODE)
1750             continue;
1751         if (!strcmp((const char *) n->name, "metadata"))
1752         {
1753             struct conf_metadata *ser_md = 0;
1754             struct conf_sortkey *ser_sk = 0;
1755             struct record_metadata **wheretoput = 0;
1756             struct record_metadata *rec_md = 0;
1757             int md_field_id = -1;
1758             int sk_field_id = -1;
1759
1760             type = xmlGetProp(n, (xmlChar *) "type");
1761             value = xmlNodeListGetString(xdoc, n->children, 1);
1762
1763             if (!type || !value || !*value)
1764                 continue;
1765
1766             md_field_id
1767                 = conf_service_metadata_field_id(service, (const char *) type);
1768             if (md_field_id < 0)
1769                 continue;
1770
1771             ser_md = &service->metadata[md_field_id];
1772
1773             if (ser_md->sortkey_offset >= 0)
1774             {
1775                 sk_field_id = ser_md->sortkey_offset;
1776                 ser_sk = &service->sortkeys[sk_field_id];
1777             }
1778
1779             // merged metadata
1780             rec_md = record_metadata_init(se->nmem, (const char *) value,
1781                                           ser_md->type, 0);
1782             if (!rec_md)
1783                 continue;
1784
1785             wheretoput = &cluster->metadata[md_field_id];
1786
1787             // and polulate with data:
1788             // assign cluster or record based on merge action
1789             if (ser_md->merge == Metadata_merge_unique)
1790             {
1791                 while (*wheretoput)
1792                 {
1793                     if (!strcmp((const char *) (*wheretoput)->data.text.disp,
1794                                 rec_md->data.text.disp))
1795                         break;
1796                     wheretoput = &(*wheretoput)->next;
1797                 }
1798                 if (!*wheretoput)
1799                     *wheretoput = rec_md;
1800             }
1801             else if (ser_md->merge == Metadata_merge_longest)
1802             {
1803                 if (!*wheretoput
1804                     || strlen(rec_md->data.text.disp)
1805                     > strlen((*wheretoput)->data.text.disp))
1806                 {
1807                     *wheretoput = rec_md;
1808                     if (ser_sk)
1809                     {
1810                         const char *sort_str = 0;
1811                         int skip_article =
1812                             ser_sk->type == Metadata_sortkey_skiparticle;
1813
1814                         if (!cluster->sortkeys[sk_field_id])
1815                             cluster->sortkeys[sk_field_id] =
1816                                 nmem_malloc(se->nmem,
1817                                             sizeof(union data_types));
1818
1819                         prt =
1820                             pp2_charset_token_create(service->charsets, "sort");
1821
1822                         pp2_charset_token_first(prt, rec_md->data.text.disp,
1823                                                 skip_article);
1824
1825                         pp2_charset_token_next(prt);
1826
1827                         sort_str = pp2_get_sort(prt);
1828
1829                         cluster->sortkeys[sk_field_id]->text.disp =
1830                             rec_md->data.text.disp;
1831                         if (!sort_str)
1832                         {
1833                             sort_str = rec_md->data.text.disp;
1834                             session_log(se, YLOG_WARN,
1835                                     "Could not make sortkey. Bug #1858");
1836                         }
1837                         cluster->sortkeys[sk_field_id]->text.sort =
1838                             nmem_strdup(se->nmem, sort_str);
1839                         pp2_charset_token_destroy(prt);
1840                     }
1841                 }
1842             }
1843             else if (ser_md->merge == Metadata_merge_all)
1844             {
1845                 while (*wheretoput)
1846                     wheretoput = &(*wheretoput)->next;
1847                 *wheretoput = rec_md;
1848             }
1849             else if (ser_md->merge == Metadata_merge_range)
1850             {
1851                 if (!*wheretoput)
1852                 {
1853                     *wheretoput = rec_md;
1854                     if (ser_sk)
1855                         cluster->sortkeys[sk_field_id]
1856                             = &rec_md->data;
1857                 }
1858                 else
1859                 {
1860                     int this_min = rec_md->data.number.min;
1861                     int this_max = rec_md->data.number.max;
1862                     if (this_min < (*wheretoput)->data.number.min)
1863                         (*wheretoput)->data.number.min = this_min;
1864                     if (this_max > (*wheretoput)->data.number.max)
1865                         (*wheretoput)->data.number.max = this_max;
1866                 }
1867             }
1868
1869
1870             // ranking of _all_ fields enabled ...
1871             if (ser_md->rank)
1872                 relevance_countwords(se->relevance, cluster,
1873                                      (char *) value, ser_md->rank,
1874                                      ser_md->name);
1875
1876             // construct facets ... unless the client already has reported them
1877             if (ser_md->termlist && !client_has_facet(cl, (char *) type))
1878             {
1879                 if (ser_md->type == Metadata_type_year)
1880                 {
1881                     char year[64];
1882                     sprintf(year, "%d", rec_md->data.number.max);
1883
1884                     add_facet(se, (char *) type, year, term_factor);
1885                     if (rec_md->data.number.max != rec_md->data.number.min)
1886                     {
1887                         sprintf(year, "%d", rec_md->data.number.min);
1888                         add_facet(se, (char *) type, year, term_factor);
1889                     }
1890                 }
1891                 else
1892                     add_facet(se, (char *) type, (char *) value, term_factor);
1893             }
1894
1895             // cleaning up
1896             xmlFree(type);
1897             xmlFree(value);
1898             type = value = 0;
1899         }
1900         else
1901         {
1902             if (se->number_of_warnings_unknown_elements == 0)
1903                 session_log(se, YLOG_WARN,
1904                         "Unexpected element in internal record: %s", n->name);
1905             se->number_of_warnings_unknown_elements++;
1906         }
1907     }
1908     if (type)
1909         xmlFree(type);
1910     if (value)
1911         xmlFree(value);
1912
1913     relevance_donerecord(se->relevance, cluster);
1914     se->total_records++;
1915
1916     return 0;
1917 }
1918
1919 void session_log(struct session *s, int level, const char *fmt, ...)
1920 {
1921     char buf[1024];
1922     va_list ap;
1923     va_start(ap, fmt);
1924
1925     yaz_vsnprintf(buf, sizeof(buf)-30, fmt, ap);
1926     yaz_log(level, "Session %u: %s", s->session_id, buf);
1927
1928     va_end(ap);
1929 }
1930
1931 /*
1932  * Local variables:
1933  * c-basic-offset: 4
1934  * c-file-style: "Stroustrup"
1935  * indent-tabs-mode: nil
1936  * End:
1937  * vim: shiftwidth=4 tabstop=8 expandtab
1938  */
1939