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 struct session_sorted_results {
  99     const char *field;
 100     int increasing;
 101     struct session_sorted_results *next;
 102 };
 103
 104 /* session counting (1) , disable client counting (0) */
 105 static YAZ_MUTEX g_session_mutex = 0;
 106 static int no_sessions = 0;
 107 static int no_session_total = 0;
 108
 109 static int session_use(int delta)
 110 {
 111     int sessions;
 112     if (!g_session_mutex)
 113         yaz_mutex_create(&g_session_mutex);
 114     yaz_mutex_enter(g_session_mutex);
 115     no_sessions += delta;
 116     if (delta > 0)
 117         no_session_total += delta;
 118     sessions = no_sessions;
 119     yaz_mutex_leave(g_session_mutex);
 120     yaz_log(YLOG_DEBUG, "%s sessions=%d", delta == 0 ? "" : (delta > 0 ? "INC" : "DEC"), no_sessions);
 121     return sessions;
 122 }
 123
 124 int sessions_count(void)
 125 {
 126     return session_use(0);
 127 }
 128
 129 int session_count_total(void)
 130 {
 131     int total = 0;
 132     if (!g_session_mutex)
 133         return 0;
 134     yaz_mutex_enter(g_session_mutex);
 135     total = no_session_total;
 136     yaz_mutex_leave(g_session_mutex);
 137     return total;
 138 }
 139
 140 static void log_xml_doc(xmlDoc *doc)
 141 {
 142     FILE *lf = yaz_log_file();
 143     xmlChar *result = 0;
 144     int len = 0;
 145 #if LIBXML_VERSION >= 20600
 146     xmlDocDumpFormatMemory(doc, &result, &len, 1);
 147 #else
 148     xmlDocDumpMemory(doc, &result, &len);
 149 #endif
 150     if (lf && len)
 151     {
 152         (void) fwrite(result, 1, len, lf);
 153         fprintf(lf, "\n");
 154     }
 155     xmlFree(result);
 156 }
 157
 158 static void session_enter(struct session *s)
 159 {
 160     yaz_mutex_enter(s->session_mutex);
 161 }
 162
 163 static void session_leave(struct session *s)
 164 {
 165     yaz_mutex_leave(s->session_mutex);
 166 }
 167
 168 void add_facet(struct session *s, const char *type, const char *value, int count)
 169 {
 170     struct conf_service *service = s->service;
 171     pp2_charset_token_t prt;
 172     const char *facet_component;
 173     WRBUF facet_wrbuf = wrbuf_alloc();
 174     WRBUF display_wrbuf = wrbuf_alloc();
 175     int i;
 176     const char *icu_chain_id = 0;
 177
 178     for (i = 0; i < service->num_metadata; i++)
 179         if (!strcmp((service->metadata + i)->name, type))
 180             icu_chain_id = (service->metadata + i)->facetrule;
 181     if (!icu_chain_id)
 182         icu_chain_id = "facet";
 183     prt = pp2_charset_token_create(service->charsets, icu_chain_id);
 184     if (!prt)
 185     {
 186         yaz_log(YLOG_FATAL, "Unknown ICU chain '%s' for facet of type '%s'",
 187                 icu_chain_id, type);
 188         wrbuf_destroy(facet_wrbuf);
 189         wrbuf_destroy(display_wrbuf);
 190         return;
 191     }
 192     pp2_charset_token_first(prt, value, 0);
 193     while ((facet_component = pp2_charset_token_next(prt)))
 194     {
 195         const char *display_component;
 196         if (*facet_component)
 197         {
 198             if (wrbuf_len(facet_wrbuf))
 199                 wrbuf_puts(facet_wrbuf, " ");
 200             wrbuf_puts(facet_wrbuf, facet_component);
 201         }
 202         display_component = pp2_get_display(prt);
 203         if (display_component)
 204         {
 205             if (wrbuf_len(display_wrbuf))
 206                 wrbuf_puts(display_wrbuf, " ");
 207             wrbuf_puts(display_wrbuf, display_component);
 208         }
 209     }
 210     pp2_charset_token_destroy(prt);
 211
 212     if (wrbuf_len(facet_wrbuf))
 213     {
 214         int i;
 215         for (i = 0; i < s->num_termlists; i++)
 216             if (!strcmp(s->termlists[i].name, type))
 217                 break;
 218         if (i == s->num_termlists)
 219         {
 220             if (i == SESSION_MAX_TERMLISTS)
 221             {
 222                 session_log(s, YLOG_FATAL, "Too many termlists");
 223                 wrbuf_destroy(facet_wrbuf);
 224                 wrbuf_destroy(display_wrbuf);
 225                 return;
 226             }
 227
 228             s->termlists[i].name = nmem_strdup(s->nmem, type);
 229             s->termlists[i].termlist
 230                 = termlist_create(s->nmem, TERMLIST_HIGH_SCORE);
 231             s->num_termlists = i + 1;
 232         }
 233
 234 #if 0
 235         session_log(s, YLOG_DEBUG, "Facets for %s: %s norm:%s (%d)", type, value, wrbuf_cstr(facet_wrbuf), count);
 236 #endif
 237         termlist_insert(s->termlists[i].termlist, wrbuf_cstr(display_wrbuf),
 238                         wrbuf_cstr(facet_wrbuf), count);
 239     }
 240     wrbuf_destroy(facet_wrbuf);
 241     wrbuf_destroy(display_wrbuf);
 242 }
 243
 244 static xmlDoc *record_to_xml(struct session *se,
 245                              struct session_database *sdb, const char *rec)
 246 {
 247     struct database *db = sdb->database;
 248     xmlDoc *rdoc = 0;
 249
 250     rdoc = xmlParseMemory(rec, strlen(rec));
 251
 252     if (!rdoc)
 253     {
 254         session_log(se, YLOG_FATAL, "Non-wellformed XML received from %s",
 255                     db->id);
 256         return 0;
 257     }
 258
 259     if (global_parameters.dump_records)
 260     {
 261         session_log(se, YLOG_LOG, "Un-normalized record from %s", db->id);
 262         log_xml_doc(rdoc);
 263     }
 264
 265     return rdoc;
 266 }
 267
 268 #define MAX_XSLT_ARGS 16
 269
 270 // Add static values from session database settings if applicable
 271 static void insert_settings_parameters(struct session_database *sdb,
 272                                        struct conf_service *service,
 273                                        char **parms,
 274                                        NMEM nmem)
 275 {
 276     int i;
 277     int nparms = 0;
 278     int offset = 0;
 279
 280     for (i = 0; i < service->num_metadata; i++)
 281     {
 282         struct conf_metadata *md = &service->metadata[i];
 283         int setting;
 284
 285         if (md->setting == Metadata_setting_parameter &&
 286             (setting = settings_lookup_offset(service, md->name)) >= 0)
 287         {
 288             const char *val = session_setting_oneval(sdb, setting);
 289             if (val && nparms < MAX_XSLT_ARGS)
 290             {
 291                 char *buf;
 292                 int len = strlen(val);
 293                 buf = nmem_malloc(nmem, len + 3);
 294                 buf[0] = '\'';
 295                 strcpy(buf + 1, val);
 296                 buf[len+1] = '\'';
 297                 buf[len+2] = '\0';
 298                 parms[offset++] = md->name;
 299                 parms[offset++] = buf;
 300                 nparms++;
 301             }
 302         }
 303     }
 304     parms[offset] = 0;
 305 }
 306
 307 // Add static values from session database settings if applicable
 308 static void insert_settings_values(struct session_database *sdb, xmlDoc *doc,
 309     struct conf_service *service)
 310 {
 311     int i;
 312
 313     for (i = 0; i < service->num_metadata; i++)
 314     {
 315         struct conf_metadata *md = &service->metadata[i];
 316         int offset;
 317
 318         if (md->setting == Metadata_setting_postproc &&
 319             (offset = settings_lookup_offset(service, md->name)) >= 0)
 320         {
 321             const char *val = session_setting_oneval(sdb, offset);
 322             if (val)
 323             {
 324                 xmlNode *r = xmlDocGetRootElement(doc);
 325                 xmlNode *n = xmlNewTextChild(r, 0, (xmlChar *) "metadata",
 326                                              (xmlChar *) val);
 327                 xmlSetProp(n, (xmlChar *) "type", (xmlChar *) md->name);
 328             }
 329         }
 330     }
 331 }
 332
 333 static xmlDoc *normalize_record(struct session *se,
 334                                 struct session_database *sdb,
 335                                 struct conf_service *service,
 336                                 const char *rec, NMEM nmem)
 337 {
 338     xmlDoc *rdoc = record_to_xml(se, sdb, rec);
 339
 340     if (rdoc)
 341     {
 342         char *parms[MAX_XSLT_ARGS*2+1];
 343
 344         insert_settings_parameters(sdb, service, parms, nmem);
 345
 346         if (normalize_record_transform(sdb->map, &rdoc, (const char **)parms))
 347         {
 348             session_log(se, YLOG_WARN, "Normalize failed from %s",
 349                         sdb->database->id);
 350         }
 351         else
 352         {
 353             insert_settings_values(sdb, rdoc, service);
 354
 355             if (global_parameters.dump_records)
 356             {
 357                 session_log(se, YLOG_LOG, "Normalized record from %s",
 358                             sdb->database->id);
 359                 log_xml_doc(rdoc);
 360             }
 361         }
 362     }
 363     return rdoc;
 364 }
 365
 366 void session_settings_dump(struct session *se,
 367                            struct session_database *db,
 368                            WRBUF w)
 369 {
 370     if (db->settings)
 371     {
 372         int i, num = db->num_settings;
 373         for (i = 0; i < num; i++)
 374         {
 375             struct setting *s = db->settings[i];
 376             for (;s ; s = s->next)
 377             {
 378                 wrbuf_puts(w, "<set name=\"");
 379                 wrbuf_xmlputs(w, s->name);
 380                 wrbuf_puts(w, "\" value=\"");
 381                 wrbuf_xmlputs(w, s->value);
 382                 wrbuf_puts(w, "\"/>");
 383             }
 384             if (db->settings[i])
 385                 wrbuf_puts(w, "\n");
 386         }
 387     }
 388 }
 389
 390 // Retrieve first defined value for 'name' for given database.
 391 // Will be extended to take into account user associated with session
 392 const char *session_setting_oneval(struct session_database *db, int offset)
 393 {
 394     if (offset >= db->num_settings || !db->settings[offset])
 395         return "";
 396     return db->settings[offset]->value;
 397 }
 398
 399 // Prepare XSLT stylesheets for record normalization
 400 // Structures are allocated on the session_wide nmem to avoid having
 401 // to recompute this for every search. This would lead
 402 // to leaking if a single session was to repeatedly change the PZ_XSLT
 403 // setting. However, this is not a realistic use scenario.
 404 static int prepare_map(struct session *se, struct session_database *sdb)
 405 {
 406     const char *s;
 407
 408     if (!sdb->settings)
 409     {
 410         session_log(se, YLOG_WARN, "No settings on %s", sdb->database->id);
 411         return -1;
 412     }
 413     if ((s = session_setting_oneval(sdb, PZ_XSLT)))
 414     {
 415         char auto_stylesheet[256];
 416
 417         if (!strcmp(s, "auto"))
 418         {
 419             const char *request_syntax = session_setting_oneval(
 420                 sdb, PZ_REQUESTSYNTAX);
 421             if (request_syntax)
 422             {
 423                 char *cp;
 424                 yaz_snprintf(auto_stylesheet, sizeof(auto_stylesheet),
 425                              "%s.xsl", request_syntax);
 426                 for (cp = auto_stylesheet; *cp; cp++)
 427                 {
 428                     /* deliberately only consider ASCII */
 429                     if (*cp > 32 && *cp < 127)
 430                         *cp = tolower(*cp);
 431                 }
 432                 s = auto_stylesheet;
 433             }
 434             else
 435             {
 436                 session_log(se, YLOG_WARN,
 437                             "No pz:requestsyntax for auto stylesheet");
 438             }
 439         }
 440         sdb->map = normalize_cache_get(se->normalize_cache,
 441                                        se->service->server->config, s);
 442         if (!sdb->map)
 443             return -1;
 444     }
 445     return 0;
 446 }
 447
 448 // This analyzes settings and recomputes any supporting data structures
 449 // if necessary.
 450 static int prepare_session_database(struct session *se,
 451                                     struct session_database *sdb)
 452 {
 453     if (!sdb->settings)
 454     {
 455         session_log(se, YLOG_WARN,
 456                 "No settings associated with %s", sdb->database->id);
 457         return -1;
 458     }
 459     if (sdb->settings[PZ_XSLT] && !sdb->map)
 460     {
 461         if (prepare_map(se, sdb) < 0)
 462             return -1;
 463     }
 464     return 0;
 465 }
 466
 467 // called if watch should be removed because http_channel is to be destroyed
 468 static void session_watch_cancel(void *data, struct http_channel *c,
 469                                  void *data2)
 470 {
 471     struct session_watchentry *ent = data;
 472
 473     ent->fun = 0;
 474     ent->data = 0;
 475     ent->obs = 0;
 476 }
 477
 478 // set watch. Returns 0=OK, -1 if watch is already set
 479 int session_set_watch(struct session *s, int what,
 480                       session_watchfun fun, void *data,
 481                       struct http_channel *chan)
 482 {
 483     int ret;
 484     session_enter(s);
 485     if (s->watchlist[what].fun)
 486         ret = -1;
 487     else
 488     {
 489
 490         s->watchlist[what].fun = fun;
 491         s->watchlist[what].data = data;
 492         s->watchlist[what].obs = http_add_observer(chan, &s->watchlist[what],
 493                                                    session_watch_cancel);
 494         ret = 0;
 495     }
 496     session_leave(s);
 497     return ret;
 498 }
 499
 500 void session_alert_watch(struct session *s, int what)
 501 {
 502     assert(s);
 503     session_enter(s);
 504     if (s->watchlist[what].fun)
 505     {
 506         /* our watch is no longer associated with http_channel */
 507         void *data;
 508         session_watchfun fun;
 509
 510         http_remove_observer(s->watchlist[what].obs);
 511         fun  = s->watchlist[what].fun;
 512         data = s->watchlist[what].data;
 513
 514         /* reset watch before fun is invoked - in case fun wants to set
 515            it again */
 516         s->watchlist[what].fun = 0;
 517         s->watchlist[what].data = 0;
 518         s->watchlist[what].obs = 0;
 519
 520         session_leave(s);
 521         session_log(s, YLOG_DEBUG,
 522                     "Alert Watch: %d calling function: %p", what, fun);
 523         fun(data);
 524     }
 525     else
 526         session_leave(s);
 527 }
 528
 529 //callback for grep_databases
 530 static void select_targets_callback(struct session *se,
 531                                     struct session_database *db)
 532 {
 533     struct client *cl = client_create(db->database->id);
 534     struct client_list *l;
 535
 536     client_set_database(cl, db);
 537
 538     client_set_session(cl, se);
 539
 540     l = xmalloc(sizeof(*l));
 541     l->client = cl;
 542     l->next = se->clients;
 543     se->clients = l;
 544 }
 545
 546 static void session_remove_clients(struct session *se)
 547 {
 548     struct client_list *l;
 549
 550     session_enter(se);
 551     l = se->clients;
 552     se->clients = 0;
 553     session_leave(se);
 554
 555     while (l)
 556     {
 557         struct client_list *l_next = l->next;
 558         client_lock(l->client);
 559         client_set_session(l->client, 0);
 560         client_set_database(l->client, 0);
 561         client_unlock(l->client);
 562         client_destroy(l->client);
 563         xfree(l);
 564         l = l_next;
 565     }
 566 }
 567
 568 // Associates a set of clients with a session;
 569 // Note: Session-databases represent databases with per-session
 570 // setting overrides
 571 static int select_targets(struct session *se, const char *filter)
 572 {
 573     return session_grep_databases(se, filter, select_targets_callback);
 574 }
 575
 576 int session_active_clients(struct session *s)
 577 {
 578     struct client_list *l;
 579     int res = 0;
 580
 581     for (l = s->clients; l; l = l->next)
 582         if (client_is_active(l->client))
 583             res++;
 584
 585     return res;
 586 }
 587
 588 int session_is_preferred_clients_ready(struct session *s)
 589 {
 590     struct client_list *l;
 591     int res = 0;
 592
 593     for (l = s->clients; l; l = l->next)
 594         if (client_is_active_preferred(l->client))
 595             res++;
 596     session_log(s, YLOG_DEBUG, "Has %d active preferred clients.", res);
 597     return res == 0;
 598 }
 599
 600 static const char *get_strategy_plus_sort(struct client *l, const char *field)
 601 {
 602     struct session_database *sdb = client_get_database(l);
 603     struct setting *s;
 604
 605     const char *strategy_plus_sort = 0;
 606
 607     for (s = sdb->settings[PZ_SORTMAP]; s; s = s->next)
 608     {
 609         char *p = strchr(s->name + 3, ':');
 610         if (!p)
 611         {
 612             yaz_log(YLOG_WARN, "Malformed sortmap name: %s", s->name);
 613             continue;
 614         }
 615         p++;
 616         if (!strcmp(p, field))
 617         {
 618             strategy_plus_sort = s->value;
 619             break;
 620         }
 621     }
 622     return strategy_plus_sort;
 623 }
 624
 625 void session_sort(struct session *se, const char *field, int increasing)
 626 {
 627     struct session_sorted_results *sr;
 628     struct client_list *l;
 629
 630     session_enter(se);
 631
 632     /* see if we already have sorted for this critieria */
 633     for (sr = se->sorted_results; sr; sr = sr->next)
 634     {
 635         if (!strcmp(field, sr->field) && increasing == sr->increasing)
 636             break;
 637     }
 638     if (sr)
 639     {
 640         yaz_log(YLOG_LOG, "search_sort: field=%s increasing=%d already fetched",
 641                 field, increasing);
 642         session_leave(se);
 643         return;
 644     }
 645     yaz_log(YLOG_LOG, "search_sort: field=%s increasing=%d must fetch",
 646             field, increasing);
 647     sr = nmem_malloc(se->nmem, sizeof(*sr));
 648     sr->field = nmem_strdup(se->nmem, field);
 649     sr->increasing = increasing;
 650     sr->next = se->sorted_results;
 651     se->sorted_results = sr;
 652
 653     for (l = se->clients; l; l = l->next)
 654     {
 655         struct client *cl = l->client;
 656         const char *strategy_plus_sort = get_strategy_plus_sort(cl, field);
 657         if (strategy_plus_sort)
 658         {
 659             struct timeval tval;
 660             if (client_prep_connection(cl, se->service->z3950_operation_timeout,
 661                                        se->service->z3950_session_timeout,
 662                                        se->service->server->iochan_man,
 663                                        &tval))
 664                 client_start_search(cl, strategy_plus_sort, increasing);
 665         }
 666     }
 667     session_leave(se);
 668 }
 669
 670 enum pazpar2_error_code session_search(struct session *se,
 671                                        const char *query,
 672                                        const char *startrecs,
 673                                        const char *maxrecs,
 674                                        const char *filter,
 675                                        const char *limit,
 676                                        const char **addinfo,
 677                                        const char *sort_field, int increasing)
 678 {
 679     int live_channels = 0;
 680     int no_working = 0;
 681     int no_failed = 0;
 682     struct client_list *l;
 683     struct timeval tval;
 684     facet_limits_t facet_limits;
 685
 686     session_log(se, YLOG_DEBUG, "Search");
 687
 688     *addinfo = 0;
 689
 690     session_remove_clients(se);
 691
 692     session_enter(se);
 693     reclist_destroy(se->reclist);
 694     se->reclist = 0;
 695     relevance_destroy(&se->relevance);
 696     nmem_reset(se->nmem);
 697     se->total_records = se->total_merged = 0;
 698     se->num_termlists = 0;
 699
 700     /* reset list of sorted results and clear to relevance search */
 701     se->sorted_results = nmem_malloc(se->nmem, sizeof(*se->sorted_results));
 702     se->sorted_results->field = nmem_strdup(se->nmem, sort_field);
 703     se->sorted_results->increasing = increasing;
 704     se->sorted_results->next = 0;
 705
 706     live_channels = select_targets(se, filter);
 707     if (!live_channels)
 708     {
 709         session_leave(se);
 710         return PAZPAR2_NO_TARGETS;
 711     }
 712     se->reclist = reclist_create(se->nmem);
 713
 714     yaz_gettimeofday(&tval);
 715
 716     tval.tv_sec += 5;
 717
 718     facet_limits = facet_limits_create(limit);
 719     if (!facet_limits)
 720     {
 721         *addinfo = "limit";
 722         session_leave(se);
 723         return PAZPAR2_MALFORMED_PARAMETER_VALUE;
 724     }
 725     for (l = se->clients; l; l = l->next)
 726     {
 727         struct client *cl = l->client;
 728         const char *strategy_plus_sort = get_strategy_plus_sort(cl, sort_field);
 729
 730         if (maxrecs)
 731             client_set_maxrecs(cl, atoi(maxrecs));
 732         if (startrecs)
 733             client_set_startrecs(cl, atoi(startrecs));
 734         if (prepare_session_database(se, client_get_database(cl)) < 0)
 735             ;
 736         else if (client_parse_query(cl, query, facet_limits) < 0)
 737             no_failed++;
 738         else
 739         {
 740             no_working++;
 741             if (client_prep_connection(cl, se->service->z3950_operation_timeout,
 742                                        se->service->z3950_session_timeout,
 743                                        se->service->server->iochan_man,
 744                                        &tval))
 745                 client_start_search(cl, strategy_plus_sort, increasing);
 746         }
 747     }
 748     facet_limits_destroy(facet_limits);
 749     session_leave(se);
 750     if (no_working == 0)
 751     {
 752         if (no_failed > 0)
 753         {
 754             *addinfo = "query";
 755             return PAZPAR2_MALFORMED_PARAMETER_VALUE;
 756         }
 757         else
 758             return PAZPAR2_NO_TARGETS;
 759     }
 760     return PAZPAR2_NO_ERROR;
 761 }
 762
 763 // Creates a new session_database object for a database
 764 static void session_init_databases_fun(void *context, struct database *db)
 765 {
 766     struct session *se = (struct session *) context;
 767     struct session_database *new = nmem_malloc(se->session_nmem, sizeof(*new));
 768     int i;
 769
 770     new->database = db;
 771
 772     new->map = 0;
 773     assert(db->settings);
 774     new->settings = nmem_malloc(se->session_nmem,
 775                                 sizeof(struct settings *) * db->num_settings);
 776     new->num_settings = db->num_settings;
 777     for (i = 0; i < db->num_settings; i++)
 778     {
 779         struct setting *setting = db->settings[i];
 780         new->settings[i] = setting;
 781     }
 782     new->next = se->databases;
 783     se->databases = new;
 784 }
 785
 786 // Doesn't free memory associated with sdb -- nmem takes care of that
 787 static void session_database_destroy(struct session_database *sdb)
 788 {
 789     sdb->map = 0;
 790 }
 791
 792 // Initialize session_database list -- this represents this session's view
 793 // of the database list -- subject to modification by the settings ws command
 794 void session_init_databases(struct session *se)
 795 {
 796     se->databases = 0;
 797     predef_grep_databases(se, se->service, session_init_databases_fun);
 798 }
 799
 800 // Probably session_init_databases_fun should be refactored instead of
 801 // called here.
 802 static struct session_database *load_session_database(struct session *se,
 803                                                       char *id)
 804 {
 805     struct database *db = new_database(id, se->session_nmem);
 806
 807     session_init_databases_fun((void*) se, db);
 808
 809     // New sdb is head of se->databases list
 810     return se->databases;
 811 }
 812
 813 // Find an existing session database. If not found, load it
 814 static struct session_database *find_session_database(struct session *se,
 815                                                       char *id)
 816 {
 817     struct session_database *sdb;
 818
 819     for (sdb = se->databases; sdb; sdb = sdb->next)
 820         if (!strcmp(sdb->database->id, id))
 821             return sdb;
 822     return load_session_database(se, id);
 823 }
 824
 825 // Apply a session override to a database
 826 void session_apply_setting(struct session *se, char *dbname, char *setting,
 827                            char *value)
 828 {
 829     struct session_database *sdb = find_session_database(se, dbname);
 830     struct conf_service *service = se->service;
 831     struct setting *new = nmem_malloc(se->session_nmem, sizeof(*new));
 832     int offset = settings_create_offset(service, setting);
 833
 834     expand_settings_array(&sdb->settings, &sdb->num_settings, offset,
 835                           se->session_nmem);
 836     new->precedence = 0;
 837     new->target = dbname;
 838     new->name = setting;
 839     new->value = value;
 840     new->next = sdb->settings[offset];
 841     sdb->settings[offset] = new;
 842
 843     // Force later recompute of settings-driven data structures
 844     // (happens when a search starts and client connections are prepared)
 845     switch (offset)
 846     {
 847     case PZ_XSLT:
 848         if (sdb->map)
 849         {
 850             sdb->map = 0;
 851         }
 852         break;
 853     }
 854 }
 855
 856 void session_destroy(struct session *se) {
 857     struct session_database *sdb;
 858     session_log(se, YLOG_DEBUG, "Destroying");
 859     session_use(-1);
 860     session_remove_clients(se);
 861
 862     for (sdb = se->databases; sdb; sdb = sdb->next)
 863         session_database_destroy(sdb);
 864     normalize_cache_destroy(se->normalize_cache);
 865     relevance_destroy(&se->relevance);
 866     reclist_destroy(se->reclist);
 867     nmem_destroy(se->nmem);
 868     service_destroy(se->service);
 869     yaz_mutex_destroy(&se->session_mutex);
 870 }
 871
 872 /* Depreciated: use session_destroy */
 873 void destroy_session(struct session *se)
 874 {
 875     session_destroy(se);
 876 }
 877
 878 size_t session_get_memory_status(struct session *session) {
 879     size_t session_nmem;
 880     if (session == 0)
 881         return 0;
 882     session_enter(session);
 883     session_nmem = nmem_total(session->nmem);
 884     session_leave(session);
 885     return session_nmem;
 886 }
 887
 888
 889 struct session *new_session(NMEM nmem, struct conf_service *service,
 890                             unsigned session_id)
 891 {
 892     int i;
 893     struct session *session = nmem_malloc(nmem, sizeof(*session));
 894
 895     char tmp_str[50];
 896
 897     sprintf(tmp_str, "session#%u", session_id);
 898
 899     session->session_id = session_id;
 900     session_log(session, YLOG_DEBUG, "New");
 901     session->service = service;
 902     session->relevance = 0;
 903     session->total_records = 0;
 904     session->number_of_warnings_unknown_elements = 0;
 905     session->number_of_warnings_unknown_metadata = 0;
 906     session->num_termlists = 0;
 907     session->reclist = 0;
 908     session->clients = 0;
 909     session->session_nmem = nmem;
 910     session->nmem = nmem_create();
 911     session->databases = 0;
 912     for (i = 0; i <= SESSION_WATCH_MAX; i++)
 913     {
 914         session->watchlist[i].data = 0;
 915         session->watchlist[i].fun = 0;
 916     }
 917     session->normalize_cache = normalize_cache_create();
 918     session->session_mutex = 0;
 919     pazpar2_mutex_create(&session->session_mutex, tmp_str);
 920     session_use(1);
 921     return session;
 922 }
 923
 924 static struct hitsbytarget *hitsbytarget_nb(struct session *se,
 925                                             int *count, NMEM nmem)
 926 {
 927     struct hitsbytarget *res = 0;
 928     struct client_list *l;
 929     size_t sz = 0;
 930
 931     for (l = se->clients; l; l = l->next)
 932         sz++;
 933
 934     res = nmem_malloc(nmem, sizeof(*res) * sz);
 935     *count = 0;
 936     for (l = se->clients; l; l = l->next)
 937     {
 938         struct client *cl = l->client;
 939         WRBUF w = wrbuf_alloc();
 940         const char *name = session_setting_oneval(client_get_database(cl),
 941                                                   PZ_NAME);
 942
 943         res[*count].id = client_get_id(cl);
 944         res[*count].name = *name ? name : "Unknown";
 945         res[*count].hits = client_get_hits(cl);
 946         res[*count].records = client_get_num_records(cl);
 947         res[*count].diagnostic = client_get_diagnostic(cl);
 948         res[*count].state = client_get_state_str(cl);
 949         res[*count].connected  = client_get_connection(cl) ? 1 : 0;
 950         session_settings_dump(se, client_get_database(cl), w);
 951         res[*count].settings_xml = nmem_strdup(nmem, wrbuf_cstr(w));
 952         wrbuf_destroy(w);
 953         (*count)++;
 954     }
 955     return res;
 956 }
 957
 958 struct hitsbytarget *get_hitsbytarget(struct session *se, int *count, NMEM nmem)
 959 {
 960     struct hitsbytarget *p;
 961     session_enter(se);
 962     p = hitsbytarget_nb(se, count, nmem);
 963     session_leave(se);
 964     return p;
 965 }
 966
 967 struct termlist_score **get_termlist_score(struct session *se,
 968                                            const char *name, int *num)
 969 {
 970     int i;
 971     struct termlist_score **tl = 0;
 972
 973     session_enter(se);
 974     for (i = 0; i < se->num_termlists; i++)
 975         if (!strcmp((const char *) se->termlists[i].name, name))
 976         {
 977             tl = termlist_highscore(se->termlists[i].termlist, num);
 978             break;
 979         }
 980     session_leave(se);
 981     return tl;
 982 }
 983
 984 // Compares two hitsbytarget nodes by hitcount
 985 static int cmp_ht(const void *p1, const void *p2)
 986 {
 987     const struct hitsbytarget *h1 = p1;
 988     const struct hitsbytarget *h2 = p2;
 989     return h2->hits - h1->hits;
 990 }
 991
 992 static int targets_termlist_nb(WRBUF wrbuf, struct session *se, int num,
 993                                NMEM nmem)
 994 {
 995     struct hitsbytarget *ht;
 996     int count, i;
 997
 998     ht = hitsbytarget_nb(se, &count, nmem);
 999     qsort(ht, count, sizeof(struct hitsbytarget), cmp_ht);
1000     for (i = 0; i < count && i < num && ht[i].hits > 0; i++)
1001     {
1002
1003         // do only print terms which have display names
1004
1005         wrbuf_puts(wrbuf, "<term>\n");
1006
1007         wrbuf_puts(wrbuf, "<id>");
1008         wrbuf_xmlputs(wrbuf, ht[i].id);
1009         wrbuf_puts(wrbuf, "</id>\n");
1010
1011         wrbuf_puts(wrbuf, "<name>");
1012         if (!ht[i].name || !ht[i].name[0])
1013             wrbuf_xmlputs(wrbuf, "NO TARGET NAME");
1014         else
1015             wrbuf_xmlputs(wrbuf, ht[i].name);
1016         wrbuf_puts(wrbuf, "</name>\n");
1017
1018         wrbuf_printf(wrbuf, "<frequency>" ODR_INT_PRINTF "</frequency>\n",
1019                      ht[i].hits);
1020
1021         wrbuf_puts(wrbuf, "<state>");
1022         wrbuf_xmlputs(wrbuf, ht[i].state);
1023         wrbuf_puts(wrbuf, "</state>\n");
1024
1025         wrbuf_printf(wrbuf, "<diagnostic>%d</diagnostic>\n",
1026                      ht[i].diagnostic);
1027         wrbuf_puts(wrbuf, "</term>\n");
1028     }
1029     return count;
1030 }
1031
1032 void perform_termlist(struct http_channel *c, struct session *se,
1033                       const char *name, int num)
1034 {
1035     int i, j;
1036     NMEM nmem_tmp = nmem_create();
1037     char **names;
1038     int num_names = 0;
1039
1040     if (name)
1041         nmem_strsplit(nmem_tmp, ",", name, &names, &num_names);
1042
1043     session_enter(se);
1044
1045     for (j = 0; j < num_names; j++)
1046     {
1047         const char *tname;
1048         for (i = 0; i < se->num_termlists; i++)
1049         {
1050             tname = se->termlists[i].name;
1051             if (num_names > 0 && !strcmp(names[j], tname))
1052             {
1053                 struct termlist_score **p = 0;
1054                 int len;
1055                 p = termlist_highscore(se->termlists[i].termlist, &len);
1056                 if (p)
1057                 {
1058                     int i;
1059                     wrbuf_puts(c->wrbuf, "<list name=\"");
1060                     wrbuf_xmlputs(c->wrbuf, tname);
1061                     wrbuf_puts(c->wrbuf, "\">\n");
1062                     for (i = 0; i < len && i < num; i++)
1063                     {
1064                         // prevent sending empty term elements
1065                         if (!p[i]->display_term || !p[i]->display_term[0])
1066                             continue;
1067
1068                         wrbuf_puts(c->wrbuf, "<term>");
1069                         wrbuf_puts(c->wrbuf, "<name>");
1070                         wrbuf_xmlputs(c->wrbuf, p[i]->display_term);
1071                         wrbuf_puts(c->wrbuf, "</name>");
1072
1073                         wrbuf_printf(c->wrbuf,
1074                                      "<frequency>%d</frequency>",
1075                                      p[i]->frequency);
1076                         wrbuf_puts(c->wrbuf, "</term>\n");
1077                     }
1078                     wrbuf_puts(c->wrbuf, "</list>\n");
1079                 }
1080             }
1081         }
1082         tname = "xtargets";
1083         if (num_names > 0 && !strcmp(names[j], tname))
1084         {
1085             wrbuf_puts(c->wrbuf, "<list name=\"");
1086             wrbuf_xmlputs(c->wrbuf, tname);
1087             wrbuf_puts(c->wrbuf, "\">\n");
1088             targets_termlist_nb(c->wrbuf, se, num, c->nmem);
1089             wrbuf_puts(c->wrbuf, "</list>\n");
1090         }
1091     }
1092     session_leave(se);
1093     nmem_destroy(nmem_tmp);
1094 }
1095
1096 #ifdef MISSING_HEADERS
1097 void report_nmem_stats(void)
1098 {
1099     size_t in_use, is_free;
1100
1101     nmem_get_memory_in_use(&in_use);
1102     nmem_get_memory_free(&is_free);
1103
1104     yaz_log(YLOG_LOG, "nmem stat: use=%ld free=%ld",
1105             (long) in_use, (long) is_free);
1106 }
1107 #endif
1108
1109 struct record_cluster *show_single_start(struct session *se, const char *id,
1110                                          struct record_cluster **prev_r,
1111                                          struct record_cluster **next_r)
1112 {
1113     struct record_cluster *r = 0;
1114
1115     session_enter(se);
1116     *prev_r = 0;
1117     *next_r = 0;
1118     if (se->reclist)
1119     {
1120         reclist_enter(se->reclist);
1121         while ((r = reclist_read_record(se->reclist)))
1122         {
1123             if (!strcmp(r->recid, id))
1124             {
1125                 *next_r = reclist_read_record(se->reclist);
1126                 break;
1127             }
1128             *prev_r = r;
1129         }
1130         reclist_leave(se->reclist);
1131     }
1132     if (!r)
1133         session_leave(se);
1134     return r;
1135 }
1136
1137 void show_single_stop(struct session *se, struct record_cluster *rec)
1138 {
1139     session_leave(se);
1140 }
1141
1142 struct record_cluster **show_range_start(struct session *se,
1143                                          struct reclist_sortparms *sp,
1144                                          int start, int *num, int *total, Odr_int *sumhits)
1145 {
1146     struct record_cluster **recs;
1147     struct reclist_sortparms *spp;
1148     int i;
1149 #if USE_TIMING
1150     yaz_timing_t t = yaz_timing_create();
1151 #endif
1152     session_enter(se);
1153     recs = nmem_malloc(se->nmem, *num * sizeof(struct record_cluster *));
1154     if (!se->relevance)
1155     {
1156         *num = 0;
1157         *total = 0;
1158         *sumhits = 0;
1159         recs = 0;
1160     }
1161     else
1162     {
1163         struct client_list *l;
1164
1165         for (spp = sp; spp; spp = spp->next)
1166             if (spp->type == Metadata_sortkey_relevance)
1167             {
1168                 relevance_prepare_read(se->relevance, se->reclist);
1169                 break;
1170             }
1171         reclist_sort(se->reclist, sp);
1172
1173         reclist_enter(se->reclist);
1174         *total = reclist_get_num_records(se->reclist);
1175
1176         *sumhits = 0;
1177         for (l = se->clients; l; l = l->next)
1178             *sumhits += client_get_hits(l->client);
1179
1180         for (i = 0; i < start; i++)
1181             if (!reclist_read_record(se->reclist))
1182             {
1183                 *num = 0;
1184                 recs = 0;
1185                 break;
1186             }
1187
1188         for (i = 0; i < *num; i++)
1189         {
1190             struct record_cluster *r = reclist_read_record(se->reclist);
1191             if (!r)
1192             {
1193                 *num = i;
1194                 break;
1195             }
1196             recs[i] = r;
1197         }
1198         reclist_leave(se->reclist);
1199     }
1200 #if USE_TIMING
1201     yaz_timing_stop(t);
1202     yaz_log(YLOG_LOG, "show %6.5f %3.2f %3.2f",
1203             yaz_timing_get_real(t), yaz_timing_get_user(t),
1204             yaz_timing_get_sys(t));
1205     yaz_timing_destroy(&t);
1206 #endif
1207     return recs;
1208 }
1209
1210 void show_range_stop(struct session *se, struct record_cluster **recs)
1211 {
1212     session_leave(se);
1213 }
1214
1215 void statistics(struct session *se, struct statistics *stat)
1216 {
1217     struct client_list *l;
1218     int count = 0;
1219
1220     memset(stat, 0, sizeof(*stat));
1221     stat->num_hits = 0;
1222     for (l = se->clients; l; l = l->next)
1223     {
1224         struct client *cl = l->client;
1225         if (!client_get_connection(cl))
1226             stat->num_no_connection++;
1227         stat->num_hits += client_get_hits(cl);
1228         switch (client_get_state(cl))
1229         {
1230         case Client_Connecting: stat->num_connecting++; break;
1231         case Client_Working: stat->num_working++; break;
1232         case Client_Idle: stat->num_idle++; break;
1233         case Client_Failed: stat->num_failed++; break;
1234         case Client_Error: stat->num_error++; break;
1235         default: break;
1236         }
1237         count++;
1238     }
1239     stat->num_records = se->total_records;
1240
1241     stat->num_clients = count;
1242 }
1243
1244 static struct record_metadata *record_metadata_init(
1245     NMEM nmem, const char *value, enum conf_metadata_type type,
1246     struct _xmlAttr *attr)
1247 {
1248     struct record_metadata *rec_md = record_metadata_create(nmem);
1249     struct record_metadata_attr **attrp = &rec_md->attributes;
1250
1251     for (; attr; attr = attr->next)
1252     {
1253         if (attr->children && attr->children->content)
1254         {
1255             if (strcmp((const char *) attr->name, "type"))
1256             {  /* skip the "type" attribute.. Its value is already part of
1257                   the element in output (md-%s) and so repeating it here
1258                   is redundant */
1259                 *attrp = nmem_malloc(nmem, sizeof(**attrp));
1260                 (*attrp)->name =
1261                     nmem_strdup(nmem, (const char *) attr->name);
1262                 (*attrp)->value =
1263                     nmem_strdup(nmem, (const char *) attr->children->content);
1264                 attrp = &(*attrp)->next;
1265             }
1266         }
1267     }
1268     *attrp = 0;
1269
1270     if (type == Metadata_type_generic)
1271     {
1272         char *p = nmem_strdup(nmem, value);
1273
1274         p = normalize7bit_generic(p, " ,/.:([");
1275
1276         rec_md->data.text.disp = p;
1277         rec_md->data.text.sort = 0;
1278     }
1279     else if (type == Metadata_type_year || type == Metadata_type_date)
1280     {
1281         int first, last;
1282         int longdate = 0;
1283
1284         if (type == Metadata_type_date)
1285             longdate = 1;
1286         if (extract7bit_dates((char *) value, &first, &last, longdate) < 0)
1287             return 0;
1288
1289         rec_md->data.number.min = first;
1290         rec_md->data.number.max = last;
1291     }
1292     else
1293         return 0;
1294     return rec_md;
1295 }
1296
1297 static int get_mergekey_from_doc(xmlDoc *doc, xmlNode *root, const char *name,
1298                                  struct conf_service *service, WRBUF norm_wr)
1299 {
1300     xmlNode *n;
1301     int no_found = 0;
1302     for (n = root->children; n; n = n->next)
1303     {
1304         if (n->type != XML_ELEMENT_NODE)
1305             continue;
1306         if (!strcmp((const char *) n->name, "metadata"))
1307         {
1308             xmlChar *type = xmlGetProp(n, (xmlChar *) "type");
1309             if (type == NULL) {
1310                 yaz_log(YLOG_FATAL, "Missing type attribute on metadata element. Skipping!");
1311             }
1312             else if (!strcmp(name, (const char *) type))
1313             {
1314                 xmlChar *value = xmlNodeListGetString(doc, n->children, 1);
1315                 if (value)
1316                 {
1317                     const char *norm_str;
1318                     pp2_charset_token_t prt =
1319                         pp2_charset_token_create(service->charsets, "mergekey");
1320
1321                     pp2_charset_token_first(prt, (const char *) value, 0);
1322                     if (wrbuf_len(norm_wr) > 0)
1323                         wrbuf_puts(norm_wr, " ");
1324                     wrbuf_puts(norm_wr, name);
1325                     while ((norm_str =
1326                             pp2_charset_token_next(prt)))
1327                     {
1328                         if (*norm_str)
1329                         {
1330                             wrbuf_puts(norm_wr, " ");
1331                             wrbuf_puts(norm_wr, norm_str);
1332                         }
1333                     }
1334                     xmlFree(value);
1335                     pp2_charset_token_destroy(prt);
1336                     no_found++;
1337                 }
1338             }
1339             xmlFree(type);
1340         }
1341     }
1342     return no_found;
1343 }
1344
1345 static const char *get_mergekey(xmlDoc *doc, struct client *cl, int record_no,
1346                                 struct conf_service *service, NMEM nmem)
1347 {
1348     char *mergekey_norm = 0;
1349     xmlNode *root = xmlDocGetRootElement(doc);
1350     WRBUF norm_wr = wrbuf_alloc();
1351
1352     /* consider mergekey from XSL first */
1353     xmlChar *mergekey = xmlGetProp(root, (xmlChar *) "mergekey");
1354     if (mergekey)
1355     {
1356         const char *norm_str;
1357         pp2_charset_token_t prt =
1358             pp2_charset_token_create(service->charsets, "mergekey");
1359
1360         pp2_charset_token_first(prt, (const char *) mergekey, 0);
1361         while ((norm_str = pp2_charset_token_next(prt)))
1362         {
1363             if (*norm_str)
1364             {
1365                 if (wrbuf_len(norm_wr))
1366                     wrbuf_puts(norm_wr, " ");
1367                 wrbuf_puts(norm_wr, norm_str);
1368             }
1369         }
1370         pp2_charset_token_destroy(prt);
1371         xmlFree(mergekey);
1372     }
1373     else
1374     {
1375         /* no mergekey defined in XSL. Look for mergekey metadata instead */
1376         int field_id;
1377         for (field_id = 0; field_id < service->num_metadata; field_id++)
1378         {
1379             struct conf_metadata *ser_md = &service->metadata[field_id];
1380             if (ser_md->mergekey != Metadata_mergekey_no)
1381             {
1382                 int r = get_mergekey_from_doc(doc, root, ser_md->name,
1383                                               service, norm_wr);
1384                 if (r == 0 && ser_md->mergekey == Metadata_mergekey_required)
1385                 {
1386                     /* no mergekey on this one and it is required..
1387                        Generate unique key instead */
1388                     wrbuf_rewind(norm_wr);
1389                     break;
1390                 }
1391             }
1392         }
1393     }
1394
1395     /* generate unique key if none is not generated already or is empty */
1396     if (wrbuf_len(norm_wr) == 0)
1397     {
1398         wrbuf_printf(norm_wr, "%s-%d",
1399                      client_get_id(cl), record_no);
1400     }
1401     if (wrbuf_len(norm_wr) > 0)
1402         mergekey_norm = nmem_strdup(nmem, wrbuf_cstr(norm_wr));
1403     wrbuf_destroy(norm_wr);
1404     return mergekey_norm;
1405 }
1406
1407 /** \brief see if metadata for pz:recordfilter exists
1408     \param root xml root element of normalized record
1409     \param sdb session database for client
1410     \retval 0 if there is no metadata for pz:recordfilter
1411     \retval 1 if there is metadata for pz:recordfilter
1412
1413     If there is no pz:recordfilter defined, this function returns 1
1414     as well.
1415 */
1416
1417 static int check_record_filter(xmlNode *root, struct session_database *sdb)
1418 {
1419     int match = 0;
1420     xmlNode *n;
1421     const char *s;
1422     s = session_setting_oneval(sdb, PZ_RECORDFILTER);
1423
1424     if (!s || !*s)
1425         return 1;
1426
1427     for (n = root->children; n; n = n->next)
1428     {
1429         if (n->type != XML_ELEMENT_NODE)
1430             continue;
1431         if (!strcmp((const char *) n->name, "metadata"))
1432         {
1433             xmlChar *type = xmlGetProp(n, (xmlChar *) "type");
1434             if (type)
1435             {
1436                 size_t len;
1437                 int substring;
1438                 const char *eq;
1439
1440                 if ((eq = strchr(s, '=')))
1441                     substring = 0;
1442                 else if ((eq = strchr(s, '~')))
1443                     substring = 1;
1444                 if (eq)
1445                     len = eq - s;
1446                 else
1447                     len = strlen(s);
1448                 if (len == strlen((const char *)type) &&
1449                     !memcmp((const char *) type, s, len))
1450                 {
1451                     xmlChar *value = xmlNodeGetContent(n);
1452                     if (value && *value)
1453                     {
1454                         if (!eq ||
1455                             (substring && strstr((const char *) value, eq+1)) ||
1456                             (!substring && !strcmp((const char *) value, eq + 1)))
1457                             match = 1;
1458                     }
1459                     xmlFree(value);
1460                 }
1461                 xmlFree(type);
1462             }
1463         }
1464     }
1465     return match;
1466 }
1467
1468
1469 static int ingest_to_cluster(struct client *cl,
1470                              xmlDoc *xdoc,
1471                              xmlNode *root,
1472                              int record_no,
1473                              const char *mergekey_norm);
1474
1475 /** \brief ingest XML record
1476     \param cl client holds the result set for record
1477     \param rec record buffer (0 terminated)
1478     \param record_no record position (1, 2, ..)
1479     \param nmem working NMEM
1480     \retval 0 OK
1481     \retval -1 failure
1482     \retval -2 Filtered
1483 */
1484 int ingest_record(struct client *cl, const char *rec,
1485                   int record_no, NMEM nmem)
1486 {
1487     struct session *se = client_get_session(cl);
1488     int ret = 0;
1489     struct session_database *sdb = client_get_database(cl);
1490     struct conf_service *service = se->service;
1491     xmlDoc *xdoc = normalize_record(se, sdb, service, rec, nmem);
1492     xmlNode *root;
1493     const char *mergekey_norm;
1494
1495     if (!xdoc)
1496         return -1;
1497
1498     root = xmlDocGetRootElement(xdoc);
1499
1500     if (!check_record_filter(root, sdb))
1501     {
1502         session_log(se, YLOG_LOG, "Filtered out record no %d from %s",
1503                     record_no, sdb->database->id);
1504         xmlFreeDoc(xdoc);
1505         return -2;
1506     }
1507
1508     mergekey_norm = get_mergekey(xdoc, cl, record_no, service, nmem);
1509     if (!mergekey_norm)
1510     {
1511         session_log(se, YLOG_WARN, "Got no mergekey");
1512         xmlFreeDoc(xdoc);
1513         return -1;
1514     }
1515     session_enter(se);
1516     if (client_get_session(cl) == se)
1517         ingest_to_cluster(cl, xdoc, root, record_no, mergekey_norm);
1518     session_leave(se);
1519
1520     xmlFreeDoc(xdoc);
1521     return ret;
1522 }
1523
1524 static int ingest_to_cluster(struct client *cl,
1525                              xmlDoc *xdoc,
1526                              xmlNode *root,
1527                              int record_no,
1528                              const char *mergekey_norm)
1529 {
1530     xmlNode *n;
1531     xmlChar *type = 0;
1532     xmlChar *value = 0;
1533     struct session *se = client_get_session(cl);
1534     struct conf_service *service = se->service;
1535     int term_factor = 1;
1536     struct record_cluster *cluster;
1537     struct session_database *sdb = client_get_database(cl);
1538     struct record *record = record_create(se->nmem,
1539                                           service->num_metadata,
1540                                           service->num_sortkeys, cl,
1541                                           record_no);
1542
1543     for (n = root->children; n; n = n->next)
1544     {
1545         if (type)
1546             xmlFree(type);
1547         if (value)
1548             xmlFree(value);
1549         type = value = 0;
1550
1551         if (n->type != XML_ELEMENT_NODE)
1552             continue;
1553         if (!strcmp((const char *) n->name, "metadata"))
1554         {
1555             struct conf_metadata *ser_md = 0;
1556             struct record_metadata **wheretoput = 0;
1557             struct record_metadata *rec_md = 0;
1558             int md_field_id = -1;
1559
1560             type = xmlGetProp(n, (xmlChar *) "type");
1561             value = xmlNodeListGetString(xdoc, n->children, 1);
1562
1563             if (!type || !value || !*value)
1564                 continue;
1565
1566             md_field_id
1567                 = conf_service_metadata_field_id(service, (const char *) type);
1568             if (md_field_id < 0)
1569             {
1570                 if (se->number_of_warnings_unknown_metadata == 0)
1571                 {
1572                     session_log(se, YLOG_WARN,
1573                             "Ignoring unknown metadata element: %s", type);
1574                 }
1575                 se->number_of_warnings_unknown_metadata++;
1576                 continue;
1577             }
1578
1579             ser_md = &service->metadata[md_field_id];
1580
1581             // non-merged metadata
1582             rec_md = record_metadata_init(se->nmem, (const char *) value,
1583                                           ser_md->type, n->properties);
1584             if (!rec_md)
1585             {
1586                 session_log(se, YLOG_WARN, "bad metadata data '%s' "
1587                             "for element '%s'", value, type);
1588                 continue;
1589             }
1590             wheretoput = &record->metadata[md_field_id];
1591             while (*wheretoput)
1592                 wheretoput = &(*wheretoput)->next;
1593             *wheretoput = rec_md;
1594         }
1595     }
1596
1597     cluster = reclist_insert(se->reclist, service, record,
1598                              mergekey_norm, &se->total_merged);
1599     if (!cluster)
1600         return -1;
1601
1602     {
1603         const char *use_term_factor_str =
1604             session_setting_oneval(sdb, PZ_TERMLIST_TERM_FACTOR);
1605         if (use_term_factor_str && use_term_factor_str[0] == '1')
1606         {
1607             int maxrecs = client_get_maxrecs(cl);
1608             int hits = (int) client_get_hits(cl);
1609             term_factor = MAX(hits, maxrecs) /  MAX(1, maxrecs);
1610             assert(term_factor >= 1);
1611             yaz_log(YLOG_DEBUG, "Using term factor: %d (%d / %d)", term_factor, MAX(hits, maxrecs), MAX(1, maxrecs));
1612         }
1613     }
1614
1615     if (global_parameters.dump_records)
1616         session_log(se, YLOG_LOG, "Cluster id %s from %s (#%d)", cluster->recid,
1617                     sdb->database->id, record_no);
1618
1619
1620     relevance_newrec(se->relevance, cluster);
1621
1622     // now parsing XML record and adding data to cluster or record metadata
1623     for (n = root->children; n; n = n->next)
1624     {
1625         pp2_charset_token_t prt;
1626         if (type)
1627             xmlFree(type);
1628         if (value)
1629             xmlFree(value);
1630         type = value = 0;
1631
1632         if (n->type != XML_ELEMENT_NODE)
1633             continue;
1634         if (!strcmp((const char *) n->name, "metadata"))
1635         {
1636             struct conf_metadata *ser_md = 0;
1637             struct conf_sortkey *ser_sk = 0;
1638             struct record_metadata **wheretoput = 0;
1639             struct record_metadata *rec_md = 0;
1640             int md_field_id = -1;
1641             int sk_field_id = -1;
1642
1643             type = xmlGetProp(n, (xmlChar *) "type");
1644             value = xmlNodeListGetString(xdoc, n->children, 1);
1645
1646             if (!type || !value || !*value)
1647                 continue;
1648
1649             md_field_id
1650                 = conf_service_metadata_field_id(service, (const char *) type);
1651             if (md_field_id < 0)
1652                 continue;
1653
1654             ser_md = &service->metadata[md_field_id];
1655
1656             if (ser_md->sortkey_offset >= 0)
1657             {
1658                 sk_field_id = ser_md->sortkey_offset;
1659                 ser_sk = &service->sortkeys[sk_field_id];
1660             }
1661
1662             // merged metadata
1663             rec_md = record_metadata_init(se->nmem, (const char *) value,
1664                                           ser_md->type, 0);
1665             wheretoput = &cluster->metadata[md_field_id];
1666
1667             // and polulate with data:
1668             // assign cluster or record based on merge action
1669             if (ser_md->merge == Metadata_merge_unique)
1670             {
1671                 while (*wheretoput)
1672                 {
1673                     if (!strcmp((const char *) (*wheretoput)->data.text.disp,
1674                                 rec_md->data.text.disp))
1675                         break;
1676                     wheretoput = &(*wheretoput)->next;
1677                 }
1678                 if (!*wheretoput)
1679                     *wheretoput = rec_md;
1680             }
1681             else if (ser_md->merge == Metadata_merge_longest)
1682             {
1683                 if (!*wheretoput
1684                     || strlen(rec_md->data.text.disp)
1685                     > strlen((*wheretoput)->data.text.disp))
1686                 {
1687                     *wheretoput = rec_md;
1688                     if (ser_sk)
1689                     {
1690                         const char *sort_str = 0;
1691                         int skip_article =
1692                             ser_sk->type == Metadata_sortkey_skiparticle;
1693
1694                         if (!cluster->sortkeys[sk_field_id])
1695                             cluster->sortkeys[sk_field_id] =
1696                                 nmem_malloc(se->nmem,
1697                                             sizeof(union data_types));
1698
1699                         prt =
1700                             pp2_charset_token_create(service->charsets, "sort");
1701
1702                         pp2_charset_token_first(prt, rec_md->data.text.disp,
1703                                                 skip_article);
1704
1705                         pp2_charset_token_next(prt);
1706
1707                         sort_str = pp2_get_sort(prt);
1708
1709                         cluster->sortkeys[sk_field_id]->text.disp =
1710                             rec_md->data.text.disp;
1711                         if (!sort_str)
1712                         {
1713                             sort_str = rec_md->data.text.disp;
1714                             session_log(se, YLOG_WARN,
1715                                     "Could not make sortkey. Bug #1858");
1716                         }
1717                         cluster->sortkeys[sk_field_id]->text.sort =
1718                             nmem_strdup(se->nmem, sort_str);
1719                         pp2_charset_token_destroy(prt);
1720                     }
1721                 }
1722             }
1723             else if (ser_md->merge == Metadata_merge_all)
1724             {
1725                 while (*wheretoput)
1726                     wheretoput = &(*wheretoput)->next;
1727                 *wheretoput = rec_md;
1728             }
1729             else if (ser_md->merge == Metadata_merge_range)
1730             {
1731                 if (!*wheretoput)
1732                 {
1733                     *wheretoput = rec_md;
1734                     if (ser_sk)
1735                         cluster->sortkeys[sk_field_id]
1736                             = &rec_md->data;
1737                 }
1738                 else
1739                 {
1740                     int this_min = rec_md->data.number.min;
1741                     int this_max = rec_md->data.number.max;
1742                     if (this_min < (*wheretoput)->data.number.min)
1743                         (*wheretoput)->data.number.min = this_min;
1744                     if (this_max > (*wheretoput)->data.number.max)
1745                         (*wheretoput)->data.number.max = this_max;
1746                 }
1747             }
1748
1749
1750             // ranking of _all_ fields enabled ...
1751             if (ser_md->rank)
1752                 relevance_countwords(se->relevance, cluster,
1753                                      (char *) value, ser_md->rank,
1754                                      ser_md->name);
1755
1756             // construct facets ... unless the client already has reported them
1757             if (ser_md->termlist && !client_has_facet(cl, (char *) type))
1758             {
1759                 if (ser_md->type == Metadata_type_year)
1760                 {
1761                     char year[64];
1762                     sprintf(year, "%d", rec_md->data.number.max);
1763
1764                     add_facet(se, (char *) type, year, term_factor);
1765                     if (rec_md->data.number.max != rec_md->data.number.min)
1766                     {
1767                         sprintf(year, "%d", rec_md->data.number.min);
1768                         add_facet(se, (char *) type, year, term_factor);
1769                     }
1770                 }
1771                 else
1772                     add_facet(se, (char *) type, (char *) value, term_factor);
1773             }
1774
1775             // cleaning up
1776             xmlFree(type);
1777             xmlFree(value);
1778             type = value = 0;
1779         }
1780         else
1781         {
1782             if (se->number_of_warnings_unknown_elements == 0)
1783                 session_log(se, YLOG_WARN,
1784                         "Unexpected element in internal record: %s", n->name);
1785             se->number_of_warnings_unknown_elements++;
1786         }
1787     }
1788     if (type)
1789         xmlFree(type);
1790     if (value)
1791         xmlFree(value);
1792
1793     relevance_donerecord(se->relevance, cluster);
1794     se->total_records++;
1795
1796     return 0;
1797 }
1798
1799 void session_log(struct session *s, int level, const char *fmt, ...)
1800 {
1801     char buf[1024];
1802     va_list ap;
1803     va_start(ap, fmt);
1804
1805     yaz_vsnprintf(buf, sizeof(buf)-30, fmt, ap);
1806     yaz_log(level, "Session (%u): %s", s->session_id, buf);
1807
1808     va_end(ap);
1809 }
1810
1811 /*
1812  * Local variables:
1813  * c-basic-offset: 4
1814  * c-file-style: "Stroustrup"
1815  * indent-tabs-mode: nil
1816  * End:
1817  * vim: shiftwidth=4 tabstop=8 expandtab
1818  */
1819