src/session.c

   1 /* This file is part of Pazpar2.
   2    Copyright (C) 2006-2013 Index Data
   3
   4 Pazpar2 is free software; you can redistribute it and/or modify it under
   5 the terms of the GNU General Public License as published by the Free
   6 Software Foundation; either version 2, or (at your option) any later
   7 version.
   8
   9 Pazpar2 is distributed in the hope that it will be useful, but WITHOUT ANY
  10 WARRANTY; without even the implied warranty of MERCHANTABILITY or
  11 FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License
  12 for more details.
  13
  14 You should have received a copy of the GNU General Public License
  15 along with this program; if not, write to the Free Software
  16 Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
  17
  18 */
  19
  20 /** \file session.c
  21     \brief high-level logic; mostly user sessions and settings
  22 */
  23
  24 #if HAVE_CONFIG_H
  25 #include <config.h>
  26 #endif
  27
  28 #include <time.h>
  29 #include <stdlib.h>
  30 #include <stdio.h>
  31 #include <string.h>
  32 #if HAVE_SYS_TIME_H
  33 #include <sys/time.h>
  34 #endif
  35 #if HAVE_UNISTD_H
  36 #include <unistd.h>
  37 #endif
  38 #ifdef WIN32
  39 #include <windows.h>
  40 #endif
  41 #include <signal.h>
  42 #include <ctype.h>
  43 #include <assert.h>
  44 #include <math.h>
  45
  46 #include <yaz/marcdisp.h>
  47 #include <yaz/comstack.h>
  48 #include <yaz/tcpip.h>
  49 #include <yaz/proto.h>
  50 #include <yaz/readconf.h>
  51 #include <yaz/pquery.h>
  52 #include <yaz/otherinfo.h>
  53 #include <yaz/yaz-util.h>
  54 #include <yaz/nmem.h>
  55 #include <yaz/query-charset.h>
  56 #include <yaz/querytowrbuf.h>
  57 #include <yaz/oid_db.h>
  58 #include <yaz/snprintf.h>
  59
  60 #define USE_TIMING 0
  61 #if USE_TIMING
  62 #include <yaz/timing.h>
  63 #endif
  64
  65 #include "ppmutex.h"
  66 #include "parameters.h"
  67 #include "session.h"
  68 #include "eventl.h"
  69 #include "http.h"
  70 #include "termlists.h"
  71 #include "reclists.h"
  72 #include "relevance.h"
  73 #include "database.h"
  74 #include "client.h"
  75 #include "settings.h"
  76 #include "normalize7bit.h"
  77
  78 #include <libxml/tree.h>
  79
  80 #define MAX_CHUNK 15
  81
  82 #define MAX(a,b) ((a)>(b)?(a):(b))
  83
  84 // Note: Some things in this structure will eventually move to configuration
  85 struct parameters global_parameters =
  86 {
  87     0,   // dump_records
  88     0,   // debug_mode
  89     0,   // predictable sessions
  90 };
  91
  92 struct client_list {
  93     struct client *client;
  94     struct client_list *next;
  95 };
  96
  97 /* session counting (1) , disable client counting (0) */
  98 static YAZ_MUTEX g_session_mutex = 0;
  99 static int no_sessions = 0;
 100 static int no_session_total = 0;
 101
 102 static int session_use(int delta)
 103 {
 104     int sessions;
 105     if (!g_session_mutex)
 106         yaz_mutex_create(&g_session_mutex);
 107     yaz_mutex_enter(g_session_mutex);
 108     no_sessions += delta;
 109     if (delta > 0)
 110         no_session_total += delta;
 111     sessions = no_sessions;
 112     yaz_mutex_leave(g_session_mutex);
 113     yaz_log(YLOG_DEBUG, "%s sessions=%d", delta == 0 ? "" : (delta > 0 ? "INC" : "DEC"), no_sessions);
 114     return sessions;
 115 }
 116
 117 int sessions_count(void)
 118 {
 119     return session_use(0);
 120 }
 121
 122 int session_count_total(void)
 123 {
 124     int total = 0;
 125     if (!g_session_mutex)
 126         return 0;
 127     yaz_mutex_enter(g_session_mutex);
 128     total = no_session_total;
 129     yaz_mutex_leave(g_session_mutex);
 130     return total;
 131 }
 132
 133 static void log_xml_doc(xmlDoc *doc)
 134 {
 135     FILE *lf = yaz_log_file();
 136     xmlChar *result = 0;
 137     int len = 0;
 138 #if LIBXML_VERSION >= 20600
 139     xmlDocDumpFormatMemory(doc, &result, &len, 1);
 140 #else
 141     xmlDocDumpMemory(doc, &result, &len);
 142 #endif
 143     if (lf && len)
 144     {
 145         (void) fwrite(result, 1, len, lf);
 146         fprintf(lf, "\n");
 147     }
 148     xmlFree(result);
 149 }
 150
 151 static void session_enter(struct session *s, const char *caller)
 152 {
 153     if (caller)
 154         session_log(s, YLOG_DEBUG, "Session lock by %s", caller);
 155     yaz_mutex_enter(s->session_mutex);
 156 }
 157
 158 static void session_leave(struct session *s, const char *caller)
 159 {
 160     yaz_mutex_leave(s->session_mutex);
 161     if (caller)
 162         session_log(s, YLOG_DEBUG, "Session unlock by %s", caller);
 163 }
 164
 165 static void session_normalize_facet(struct session *s, const char *type,
 166                                     const char *value,
 167                                     WRBUF display_wrbuf,
 168                                     WRBUF facet_wrbuf)
 169 {
 170     struct conf_service *service = s->service;
 171     pp2_charset_token_t prt;
 172     const char *facet_component;
 173     int i;
 174     const char *icu_chain_id = 0;
 175
 176     for (i = 0; i < service->num_metadata; i++)
 177         if (!strcmp((service->metadata + i)->name, type))
 178             icu_chain_id = (service->metadata + i)->facetrule;
 179     if (!icu_chain_id)
 180         icu_chain_id = "facet";
 181     prt = pp2_charset_token_create(service->charsets, icu_chain_id);
 182     if (!prt)
 183     {
 184         yaz_log(YLOG_FATAL, "Unknown ICU chain '%s' for facet of type '%s'",
 185                 icu_chain_id, type);
 186         wrbuf_destroy(facet_wrbuf);
 187         wrbuf_destroy(display_wrbuf);
 188         return;
 189     }
 190     pp2_charset_token_first(prt, value, 0);
 191     while ((facet_component = pp2_charset_token_next(prt)))
 192     {
 193         const char *display_component;
 194         if (*facet_component)
 195         {
 196             if (wrbuf_len(facet_wrbuf))
 197                 wrbuf_puts(facet_wrbuf, " ");
 198             wrbuf_puts(facet_wrbuf, facet_component);
 199         }
 200         display_component = pp2_get_display(prt);
 201         if (display_component)
 202         {
 203             if (wrbuf_len(display_wrbuf))
 204                 wrbuf_puts(display_wrbuf, " ");
 205             wrbuf_puts(display_wrbuf, display_component);
 206         }
 207     }
 208     pp2_charset_token_destroy(prt);
 209 }
 210
 211 void add_facet(struct session *s, const char *type, const char *value, int count)
 212 {
 213     WRBUF facet_wrbuf = wrbuf_alloc();
 214     WRBUF display_wrbuf = wrbuf_alloc();
 215
 216     session_normalize_facet(s, type, value, display_wrbuf, facet_wrbuf);
 217
 218     if (wrbuf_len(facet_wrbuf))
 219     {
 220         int i;
 221         for (i = 0; i < s->num_termlists; i++)
 222             if (!strcmp(s->termlists[i].name, type))
 223                 break;
 224         if (i == s->num_termlists)
 225         {
 226             if (i == SESSION_MAX_TERMLISTS)
 227             {
 228                 session_log(s, YLOG_FATAL, "Too many termlists");
 229                 wrbuf_destroy(facet_wrbuf);
 230                 wrbuf_destroy(display_wrbuf);
 231                 return;
 232             }
 233
 234             s->termlists[i].name = nmem_strdup(s->nmem, type);
 235             s->termlists[i].termlist = termlist_create(s->nmem);
 236             s->num_termlists = i + 1;
 237         }
 238
 239 #if 0
 240         session_log(s, YLOG_LOG, "Facets for %s: %s norm:%s (%d)", type, value, wrbuf_cstr(facet_wrbuf), count);
 241 #endif
 242         termlist_insert(s->termlists[i].termlist, wrbuf_cstr(display_wrbuf),
 243                         wrbuf_cstr(facet_wrbuf), count);
 244     }
 245     wrbuf_destroy(facet_wrbuf);
 246     wrbuf_destroy(display_wrbuf);
 247 }
 248
 249 static xmlDoc *record_to_xml(struct session *se,
 250                              struct session_database *sdb, const char *rec)
 251 {
 252     struct database *db = sdb->database;
 253     xmlDoc *rdoc = 0;
 254
 255     rdoc = xmlParseMemory(rec, strlen(rec));
 256
 257     if (!rdoc)
 258     {
 259         session_log(se, YLOG_WARN, "Non-wellformed XML");
 260         return 0;
 261     }
 262
 263     if (global_parameters.dump_records)
 264     {
 265         session_log(se, YLOG_LOG, "Un-normalized record from %s", db->id);
 266         log_xml_doc(rdoc);
 267     }
 268
 269     return rdoc;
 270 }
 271
 272 #define MAX_XSLT_ARGS 16
 273
 274 // Add static values from session database settings if applicable
 275 static void insert_settings_parameters(struct session_database *sdb,
 276                                        struct conf_service *service,
 277                                        char **parms,
 278                                        NMEM nmem)
 279 {
 280     int i;
 281     int nparms = 0;
 282     int offset = 0;
 283
 284     for (i = 0; i < service->num_metadata; i++)
 285     {
 286         struct conf_metadata *md = &service->metadata[i];
 287         int setting;
 288
 289         if (md->setting == Metadata_setting_parameter &&
 290             (setting = settings_lookup_offset(service, md->name)) >= 0)
 291         {
 292             const char *val = session_setting_oneval(sdb, setting);
 293             if (val && nparms < MAX_XSLT_ARGS)
 294             {
 295                 char *buf;
 296                 int len = strlen(val);
 297                 buf = nmem_malloc(nmem, len + 3);
 298                 buf[0] = '\'';
 299                 strcpy(buf + 1, val);
 300                 buf[len+1] = '\'';
 301                 buf[len+2] = '\0';
 302                 parms[offset++] = md->name;
 303                 parms[offset++] = buf;
 304                 nparms++;
 305             }
 306         }
 307     }
 308     parms[offset] = 0;
 309 }
 310
 311 // Add static values from session database settings if applicable
 312 static void insert_settings_values(struct session_database *sdb, xmlDoc *doc,
 313     struct conf_service *service)
 314 {
 315     int i;
 316
 317     for (i = 0; i < service->num_metadata; i++)
 318     {
 319         struct conf_metadata *md = &service->metadata[i];
 320         int offset;
 321
 322         if (md->setting == Metadata_setting_postproc &&
 323             (offset = settings_lookup_offset(service, md->name)) >= 0)
 324         {
 325             const char *val = session_setting_oneval(sdb, offset);
 326             if (val)
 327             {
 328                 xmlNode *r = xmlDocGetRootElement(doc);
 329                 xmlNode *n = xmlNewTextChild(r, 0, (xmlChar *) "metadata",
 330                                              (xmlChar *) val);
 331                 xmlSetProp(n, (xmlChar *) "type", (xmlChar *) md->name);
 332             }
 333         }
 334     }
 335 }
 336
 337 static xmlDoc *normalize_record(struct session *se,
 338                                 struct session_database *sdb,
 339                                 struct conf_service *service,
 340                                 const char *rec, NMEM nmem)
 341 {
 342     xmlDoc *rdoc = record_to_xml(se, sdb, rec);
 343
 344     if (rdoc)
 345     {
 346         char *parms[MAX_XSLT_ARGS*2+1];
 347
 348         insert_settings_parameters(sdb, service, parms, nmem);
 349
 350         if (normalize_record_transform(sdb->map, &rdoc, (const char **)parms))
 351         {
 352             session_log(se, YLOG_WARN, "Normalize failed");
 353         }
 354         else
 355         {
 356             insert_settings_values(sdb, rdoc, service);
 357
 358             if (global_parameters.dump_records)
 359             {
 360                 session_log(se, YLOG_LOG, "Normalized record from %s",
 361                             sdb->database->id);
 362                 log_xml_doc(rdoc);
 363             }
 364         }
 365     }
 366     return rdoc;
 367 }
 368
 369 void session_settings_dump(struct session *se,
 370                            struct session_database *db,
 371                            WRBUF w)
 372 {
 373     if (db->settings)
 374     {
 375         int i, num = db->num_settings;
 376         for (i = 0; i < num; i++)
 377         {
 378             struct setting *s = db->settings[i];
 379             for (;s ; s = s->next)
 380             {
 381                 wrbuf_puts(w, "<set name=\"");
 382                 wrbuf_xmlputs(w, s->name);
 383                 wrbuf_puts(w, "\" value=\"");
 384                 wrbuf_xmlputs(w, s->value);
 385                 wrbuf_puts(w, "\"/>");
 386             }
 387             if (db->settings[i])
 388                 wrbuf_puts(w, "\n");
 389         }
 390     }
 391 }
 392
 393 // Retrieve first defined value for 'name' for given database.
 394 // Will be extended to take into account user associated with session
 395 const char *session_setting_oneval(struct session_database *db, int offset)
 396 {
 397     if (offset >= db->num_settings || !db->settings[offset])
 398         return "";
 399     return db->settings[offset]->value;
 400 }
 401
 402 // Prepare XSLT stylesheets for record normalization
 403 // Structures are allocated on the session_wide nmem to avoid having
 404 // to recompute this for every search. This would lead
 405 // to leaking if a single session was to repeatedly change the PZ_XSLT
 406 // setting. However, this is not a realistic use scenario.
 407 static int prepare_map(struct session *se, struct session_database *sdb)
 408 {
 409     if (sdb->settings && !sdb->map)
 410     {
 411         const char *s;
 412
 413         if (sdb->settings[PZ_XSLT] &&
 414             (s = session_setting_oneval(sdb, PZ_XSLT)))
 415         {
 416             char auto_stylesheet[256];
 417
 418             if (!strcmp(s, "auto"))
 419             {
 420                 const char *request_syntax = session_setting_oneval(
 421                     sdb, PZ_REQUESTSYNTAX);
 422                 if (request_syntax)
 423                 {
 424                     char *cp;
 425                     yaz_snprintf(auto_stylesheet, sizeof(auto_stylesheet),
 426                                  "%s.xsl", request_syntax);
 427                     for (cp = auto_stylesheet; *cp; cp++)
 428                     {
 429                         /* deliberately only consider ASCII */
 430                         if (*cp > 32 && *cp < 127)
 431                             *cp = tolower(*cp);
 432                     }
 433                     s = auto_stylesheet;
 434                 }
 435                 else
 436                 {
 437                     session_log(se, YLOG_WARN,
 438                                 "No pz:requestsyntax for auto stylesheet");
 439                 }
 440             }
 441             sdb->map = normalize_cache_get(se->normalize_cache,
 442                                            se->service, s);
 443             if (!sdb->map)
 444                 return -1;
 445         }
 446     }
 447     return 0;
 448 }
 449
 450 // called if watch should be removed because http_channel is to be destroyed
 451 static void session_watch_cancel(void *data, struct http_channel *c,
 452                                  void *data2)
 453 {
 454     struct session_watchentry *ent = data;
 455
 456     ent->fun = 0;
 457     ent->data = 0;
 458     ent->obs = 0;
 459 }
 460
 461 // set watch. Returns 0=OK, -1 if watch is already set
 462 int session_set_watch(struct session *s, int what,
 463                       session_watchfun fun, void *data,
 464                       struct http_channel *chan)
 465 {
 466     int ret;
 467     session_enter(s, "session_set_watch");
 468     if (s->watchlist[what].fun)
 469         ret = -1;
 470     else
 471     {
 472
 473         s->watchlist[what].fun = fun;
 474         s->watchlist[what].data = data;
 475         s->watchlist[what].obs = http_add_observer(chan, &s->watchlist[what],
 476                                                    session_watch_cancel);
 477         ret = 0;
 478     }
 479     session_leave(s, "session_set_watch");
 480     return ret;
 481 }
 482
 483 void session_alert_watch(struct session *s, int what)
 484 {
 485     assert(s);
 486     session_enter(s, "session_alert_watch");
 487     if (s->watchlist[what].fun)
 488     {
 489         /* our watch is no longer associated with http_channel */
 490         void *data;
 491         session_watchfun fun;
 492
 493         http_remove_observer(s->watchlist[what].obs);
 494         fun  = s->watchlist[what].fun;
 495         data = s->watchlist[what].data;
 496
 497         /* reset watch before fun is invoked - in case fun wants to set
 498            it again */
 499         s->watchlist[what].fun = 0;
 500         s->watchlist[what].data = 0;
 501         s->watchlist[what].obs = 0;
 502
 503         session_leave(s, "session_alert_watch");
 504         session_log(s, YLOG_DEBUG,
 505                     "Alert Watch: %d calling function: %p", what, fun);
 506         fun(data);
 507     }
 508     else
 509         session_leave(s,"session_alert_watch");
 510 }
 511
 512 //callback for grep_databases
 513 static void select_targets_callback(struct session *se,
 514                                     struct session_database *db)
 515 {
 516     struct client *cl;
 517     struct client_list *l;
 518
 519     for (l = se->clients_cached; l; l = l->next)
 520         if (client_get_database(l->client) == db)
 521             break;
 522
 523     if (l)
 524         cl = l->client;
 525     else
 526     {
 527         cl = client_create(db->database->id);
 528         client_set_database(cl, db);
 529
 530         l = xmalloc(sizeof(*l));
 531         l->client = cl;
 532         l->next = se->clients_cached;
 533         se->clients_cached = l;
 534     }
 535     /* set session always. If may be 0 if client is not active */
 536     client_set_session(cl, se);
 537
 538     l = xmalloc(sizeof(*l));
 539     l->client = cl;
 540     l->next = se->clients_active;
 541     se->clients_active = l;
 542 }
 543
 544 static void session_reset_active_clients(struct session *se,
 545                                          struct client_list *new_list)
 546 {
 547     struct client_list *l;
 548
 549     session_enter(se, "session_reset_active_clients");
 550     l = se->clients_active;
 551     se->clients_active = new_list;
 552     session_leave(se, "session_reset_active_clients");
 553
 554     while (l)
 555     {
 556         struct client_list *l_next = l->next;
 557
 558         client_lock(l->client);
 559         client_set_session(l->client, 0); /* mark client inactive */
 560         client_unlock(l->client);
 561
 562         xfree(l);
 563         l = l_next;
 564     }
 565 }
 566
 567 static void session_remove_cached_clients(struct session *se)
 568 {
 569     struct client_list *l;
 570
 571     session_reset_active_clients(se, 0);
 572
 573     session_enter(se, "session_remove_cached_clients");
 574     l = se->clients_cached;
 575     se->clients_cached = 0;
 576     session_leave(se, "session_remove_cached_clients");
 577
 578     while (l)
 579     {
 580         struct client_list *l_next = l->next;
 581         client_lock(l->client);
 582         client_set_session(l->client, 0);
 583         client_set_database(l->client, 0);
 584         client_unlock(l->client);
 585         client_destroy(l->client);
 586         xfree(l);
 587         l = l_next;
 588     }
 589 }
 590
 591 // Associates a set of clients with a session;
 592 // Note: Session-databases represent databases with per-session
 593 // setting overrides
 594 static int select_targets(struct session *se, const char *filter)
 595 {
 596     return session_grep_databases(se, filter, select_targets_callback);
 597 }
 598
 599 int session_active_clients(struct session *s)
 600 {
 601     struct client_list *l;
 602     int res = 0;
 603
 604     for (l = s->clients_active; l; l = l->next)
 605         if (client_is_active(l->client))
 606             res++;
 607
 608     return res;
 609 }
 610
 611 int session_is_preferred_clients_ready(struct session *s)
 612 {
 613     struct client_list *l;
 614     int res = 0;
 615
 616     for (l = s->clients_active; l; l = l->next)
 617         if (client_is_active_preferred(l->client))
 618             res++;
 619     session_log(s, YLOG_DEBUG, "Has %d active preferred clients.", res);
 620     return res == 0;
 621 }
 622
 623 static void session_clear_set(struct session *se, struct reclist_sortparms *sp)
 624 {
 625     reclist_destroy(se->reclist);
 626     se->reclist = 0;
 627     if (nmem_total(se->nmem))
 628         session_log(se, YLOG_DEBUG, "NMEN operation usage %zd",
 629                     nmem_total(se->nmem));
 630     nmem_reset(se->nmem);
 631     se->total_records = se->total_merged = 0;
 632     se->num_termlists = 0;
 633
 634     /* reset list of sorted results and clear to relevance search */
 635     se->sorted_results = nmem_malloc(se->nmem, sizeof(*se->sorted_results));
 636     se->sorted_results->name = nmem_strdup(se->nmem, sp->name);
 637     se->sorted_results->increasing = sp->increasing;
 638     se->sorted_results->type = sp->type;
 639     se->sorted_results->next = 0;
 640
 641     session_log(se, YLOG_DEBUG, "clear_set session_sort: field=%s increasing=%d type=%d configured",
 642             sp->name, sp->increasing, sp->type);
 643
 644     se->reclist = reclist_create(se->nmem);
 645 }
 646
 647 static void session_sort_unlocked(struct session *se, struct reclist_sortparms *sp)
 648 {
 649     struct reclist_sortparms *sr;
 650     struct client_list *l;
 651     const char *field = sp->name;
 652     int increasing = sp->increasing;
 653     int type  = sp->type;
 654     int clients_research = 0;
 655
 656     yaz_log(YLOG_LOG, "session_sort field=%s increasing=%d type=%d", field, increasing, type);
 657     /* see if we already have sorted for this criteria */
 658     for (sr = se->sorted_results; sr; sr = sr->next)
 659     {
 660         if (!reclist_sortparms_cmp(sr,sp))
 661             break;
 662     }
 663     if (sr)
 664     {
 665         session_log(se, YLOG_DEBUG, "search_sort: field=%s increasing=%d type=%d already fetched",
 666                     field, increasing, type);
 667         return;
 668     }
 669     session_log(se, YLOG_DEBUG, "search_sort: field=%s increasing=%d type=%d must fetch",
 670                     field, increasing, type);
 671
 672     // We need to reset reclist on every sort that changes the records, not just for position
 673     // So if just one client requires new searching, we need to clear set.
 674     // Ask each of the client if sorting requires re-search due to native sort
 675     // If it does it will require us to
 676     for (l = se->clients_active; l; l = l->next)
 677     {
 678         struct client *cl = l->client;
 679         // Assume no re-search is required.
 680         client_parse_init(cl, 1);
 681         clients_research += client_parse_sort(cl, sp);
 682     }
 683     if (clients_research) {
 684         yaz_log(YLOG_DEBUG, "Reset results due to %d clients researching", clients_research);
 685         session_clear_set(se, sp);
 686     }
 687     else {
 688         // A new sorting based on same record set
 689         sr = nmem_malloc(se->nmem, sizeof(*sr));
 690         sr->name = nmem_strdup(se->nmem, field);
 691         sr->increasing = increasing;
 692         sr->type = type;
 693         sr->next = se->sorted_results;
 694         se->sorted_results = sr;
 695         session_log(se, YLOG_DEBUG, "No research/ingesting done");
 696         return ;
 697     }
 698     session_log(se, YLOG_DEBUG, "Re- search/ingesting for clients due to change in sort order");
 699
 700     for (l = se->clients_active; l; l = l->next)
 701     {
 702         struct client *cl = l->client;
 703         if (client_get_state(cl) == Client_Connecting ||
 704             client_get_state(cl) == Client_Idle ||
 705             client_get_state(cl) == Client_Working) {
 706             client_start_search(cl);
 707         }
 708         else {
 709             yaz_log(YLOG_DEBUG, "Client %s: No re-start/ingest in show. Wrong client state: %d",
 710                         client_get_id(cl), client_get_state(cl));
 711         }
 712
 713     }
 714 }
 715
 716 void session_sort(struct session *se, struct reclist_sortparms *sp) {
 717     //session_enter(se, "session_sort");
 718     session_sort_unlocked(se, sp);
 719     //session_leave(se, "session_sort");
 720 }
 721
 722
 723 enum pazpar2_error_code session_search(struct session *se,
 724                                        const char *query,
 725                                        const char *startrecs,
 726                                        const char *maxrecs,
 727                                        const char *filter,
 728                                        const char *limit,
 729                                        const char **addinfo,
 730                                        struct reclist_sortparms *sp)
 731 {
 732     int live_channels = 0;
 733     int no_working = 0;
 734     int no_failed_query = 0;
 735     int no_failed_limit = 0;
 736     struct client_list *l, *l0;
 737
 738     session_log(se, YLOG_DEBUG, "Search");
 739
 740     *addinfo = 0;
 741
 742     if (se->settings_modified) {
 743         session_remove_cached_clients(se);
 744     }
 745     else
 746         session_reset_active_clients(se, 0);
 747
 748     session_enter(se, "session_search");
 749     se->settings_modified = 0;
 750
 751     session_clear_set(se, sp);
 752     relevance_destroy(&se->relevance);
 753
 754     live_channels = select_targets(se, filter);
 755     if (!live_channels)
 756     {
 757         session_leave(se, "session_search");
 758         return PAZPAR2_NO_TARGETS;
 759     }
 760
 761     facet_limits_destroy(se->facet_limits);
 762     se->facet_limits = facet_limits_create(limit);
 763     if (!se->facet_limits)
 764     {
 765         *addinfo = "limit";
 766         session_leave(se, "session_search");
 767         return PAZPAR2_MALFORMED_PARAMETER_VALUE;
 768     }
 769
 770     l0 = se->clients_active;
 771     se->clients_active = 0;
 772     session_leave(se, "session_search");
 773
 774     for (l = l0; l; l = l->next)
 775     {
 776         int parse_ret;
 777         struct client *cl = l->client;
 778         client_parse_init(cl, 1);
 779         if (prepare_map(se, client_get_database(cl)) < 0)
 780             continue;
 781
 782         parse_ret = client_parse_query(cl, query, se->facet_limits);
 783         if (parse_ret == -1)
 784             no_failed_query++;
 785         else if (parse_ret == -2)
 786             no_failed_limit++;
 787         else if (parse_ret < 0)
 788             no_working++; /* other error, such as bad CCL map */
 789         else
 790         {
 791             client_parse_range(cl, startrecs, maxrecs);
 792             client_parse_sort(cl, sp);
 793             client_start_search(cl);
 794             no_working++;
 795         }
 796     }
 797     session_reset_active_clients(se, l0);
 798
 799     if (no_working == 0)
 800     {
 801         if (no_failed_query > 0)
 802         {
 803             *addinfo = "query";
 804             return PAZPAR2_MALFORMED_PARAMETER_VALUE;
 805         }
 806         else if (no_failed_limit > 0)
 807         {
 808             *addinfo = "limit";
 809             return PAZPAR2_MALFORMED_PARAMETER_VALUE;
 810         }
 811         else
 812             return PAZPAR2_NO_TARGETS;
 813     }
 814     session_log(se, YLOG_LOG, "session_start_search done");
 815     return PAZPAR2_NO_ERROR;
 816 }
 817
 818 // Creates a new session_database object for a database
 819 static void session_init_databases_fun(void *context, struct database *db)
 820 {
 821     struct session *se = (struct session *) context;
 822     struct session_database *new = nmem_malloc(se->session_nmem, sizeof(*new));
 823     int i;
 824
 825     new->database = db;
 826
 827     new->map = 0;
 828     assert(db->settings);
 829     new->settings = nmem_malloc(se->session_nmem,
 830                                 sizeof(struct settings *) * db->num_settings);
 831     new->num_settings = db->num_settings;
 832     for (i = 0; i < db->num_settings; i++)
 833     {
 834         struct setting *setting = db->settings[i];
 835         new->settings[i] = setting;
 836     }
 837     new->next = se->databases;
 838     se->databases = new;
 839 }
 840
 841 // Doesn't free memory associated with sdb -- nmem takes care of that
 842 static void session_database_destroy(struct session_database *sdb)
 843 {
 844     sdb->map = 0;
 845 }
 846
 847 // Initialize session_database list -- this represents this session's view
 848 // of the database list -- subject to modification by the settings ws command
 849 void session_init_databases(struct session *se)
 850 {
 851     se->databases = 0;
 852     predef_grep_databases(se, se->service, session_init_databases_fun);
 853 }
 854
 855 // Probably session_init_databases_fun should be refactored instead of
 856 // called here.
 857 static struct session_database *load_session_database(struct session *se,
 858                                                       char *id)
 859 {
 860     struct database *db = new_database_inherit_settings(id, se->session_nmem, se->service->settings);
 861     session_init_databases_fun((void*) se, db);
 862
 863     // New sdb is head of se->databases list
 864     return se->databases;
 865 }
 866
 867 // Find an existing session database. If not found, load it
 868 static struct session_database *find_session_database(struct session *se,
 869                                                       char *id)
 870 {
 871     struct session_database *sdb;
 872
 873     for (sdb = se->databases; sdb; sdb = sdb->next)
 874         if (!strcmp(sdb->database->id, id))
 875             return sdb;
 876     return load_session_database(se, id);
 877 }
 878
 879 // Apply a session override to a database
 880 void session_apply_setting(struct session *se, char *dbname, char *setting,
 881                            char *value)
 882 {
 883     struct session_database *sdb = find_session_database(se, dbname);
 884     struct conf_service *service = se->service;
 885     struct setting *new = nmem_malloc(se->session_nmem, sizeof(*new));
 886     int offset = settings_create_offset(service, setting);
 887
 888     expand_settings_array(&sdb->settings, &sdb->num_settings, offset,
 889                           se->session_nmem);
 890     new->precedence = 0;
 891     new->target = dbname;
 892     new->name = setting;
 893     new->value = value;
 894     new->next = sdb->settings[offset];
 895     sdb->settings[offset] = new;
 896
 897     se->settings_modified = 1;
 898
 899     // Force later recompute of settings-driven data structures
 900     // (happens when a search starts and client connections are prepared)
 901     switch (offset)
 902     {
 903     case PZ_XSLT:
 904         if (sdb->map)
 905         {
 906             sdb->map = 0;
 907         }
 908         break;
 909     }
 910 }
 911
 912 void session_destroy(struct session *se)
 913 {
 914     struct session_database *sdb;
 915     session_log(se, YLOG_DEBUG, "Destroying");
 916     session_use(-1);
 917     session_remove_cached_clients(se);
 918
 919     for (sdb = se->databases; sdb; sdb = sdb->next)
 920         session_database_destroy(sdb);
 921     normalize_cache_destroy(se->normalize_cache);
 922     relevance_destroy(&se->relevance);
 923     reclist_destroy(se->reclist);
 924     if (nmem_total(se->nmem))
 925         session_log(se, YLOG_DEBUG, "NMEN operation usage %zd", nmem_total(se->nmem));
 926     if (nmem_total(se->session_nmem))
 927         session_log(se, YLOG_DEBUG, "NMEN session usage %zd", nmem_total(se->session_nmem));
 928     facet_limits_destroy(se->facet_limits);
 929     nmem_destroy(se->nmem);
 930     service_destroy(se->service);
 931     yaz_mutex_destroy(&se->session_mutex);
 932 }
 933
 934 size_t session_get_memory_status(struct session *session) {
 935     size_t session_nmem;
 936     if (session == 0)
 937         return 0;
 938     session_enter(session, "session_get_memory_status");
 939     session_nmem = nmem_total(session->nmem);
 940     session_leave(session, "session_get_memory_status");
 941     return session_nmem;
 942 }
 943
 944
 945 struct session *new_session(NMEM nmem, struct conf_service *service,
 946                             unsigned session_id)
 947 {
 948     int i;
 949     struct session *session = nmem_malloc(nmem, sizeof(*session));
 950
 951     char tmp_str[50];
 952
 953     sprintf(tmp_str, "session#%u", session_id);
 954
 955     session->session_id = session_id;
 956     session_log(session, YLOG_DEBUG, "New");
 957     session->service = service;
 958     session->relevance = 0;
 959     session->total_records = 0;
 960     session->number_of_warnings_unknown_elements = 0;
 961     session->number_of_warnings_unknown_metadata = 0;
 962     session->num_termlists = 0;
 963     session->reclist = 0;
 964     session->clients_active = 0;
 965     session->clients_cached = 0;
 966     session->settings_modified = 0;
 967     session->session_nmem = nmem;
 968     session->nmem = nmem_create();
 969     session->databases = 0;
 970     session->sorted_results = 0;
 971     session->facet_limits = 0;
 972
 973     for (i = 0; i <= SESSION_WATCH_MAX; i++)
 974     {
 975         session->watchlist[i].data = 0;
 976         session->watchlist[i].fun = 0;
 977     }
 978     session->normalize_cache = normalize_cache_create();
 979     session->session_mutex = 0;
 980     pazpar2_mutex_create(&session->session_mutex, tmp_str);
 981     session_use(1);
 982     return session;
 983 }
 984
 985 const char * client_get_suggestions_xml(struct client *cl, WRBUF wrbuf);
 986
 987 static struct hitsbytarget *hitsbytarget_nb(struct session *se,
 988                                             int *count, NMEM nmem)
 989 {
 990     struct hitsbytarget *res = 0;
 991     struct client_list *l;
 992     size_t sz = 0;
 993
 994     for (l = se->clients_active; l; l = l->next)
 995         sz++;
 996
 997     res = nmem_malloc(nmem, sizeof(*res) * sz);
 998     *count = 0;
 999     for (l = se->clients_active; l; l = l->next)
1000     {
1001         struct client *cl = l->client;
1002         WRBUF w = wrbuf_alloc();
1003         const char *name = session_setting_oneval(client_get_database(cl),
1004                                                   PZ_NAME);
1005
1006         res[*count].id = client_get_id(cl);
1007         res[*count].name = *name ? name : "Unknown";
1008         res[*count].hits = client_get_hits(cl);
1009         res[*count].approximation = client_get_approximation(cl);
1010         res[*count].records = client_get_num_records(cl);
1011         res[*count].filtered = client_get_num_records_filtered(cl);
1012         res[*count].diagnostic =
1013             client_get_diagnostic(cl, &res[*count].message,
1014                                   &res[*count].addinfo);
1015         res[*count].state = client_get_state_str(cl);
1016         res[*count].connected  = client_get_connection(cl) ? 1 : 0;
1017         session_settings_dump(se, client_get_database(cl), w);
1018         res[*count].settings_xml = nmem_strdup(nmem, wrbuf_cstr(w));
1019         wrbuf_rewind(w);
1020         wrbuf_puts(w, "");
1021         res[*count].suggestions_xml = nmem_strdup(nmem, client_get_suggestions_xml(cl, w));
1022         wrbuf_destroy(w);
1023         (*count)++;
1024     }
1025     return res;
1026 }
1027
1028 struct hitsbytarget *get_hitsbytarget(struct session *se, int *count, NMEM nmem)
1029 {
1030     struct hitsbytarget *p;
1031     session_enter(se, "get_hitsbytarget");
1032     p = hitsbytarget_nb(se, count, nmem);
1033     session_leave(se, "get_hitsbytarget");
1034     return p;
1035 }
1036
1037 // Compares two hitsbytarget nodes by hitcount
1038 static int cmp_ht(const void *p1, const void *p2)
1039 {
1040     const struct hitsbytarget *h1 = p1;
1041     const struct hitsbytarget *h2 = p2;
1042     return h2->hits - h1->hits;
1043 }
1044
1045 // Compares two hitsbytarget nodes by hitcount
1046 static int cmp_ht_approx(const void *p1, const void *p2)
1047 {
1048     const struct hitsbytarget *h1 = p1;
1049     const struct hitsbytarget *h2 = p2;
1050     return h2->approximation - h1->approximation;
1051 }
1052
1053 static int targets_termlist_nb(WRBUF wrbuf, struct session *se, int num,
1054                                NMEM nmem, int version)
1055 {
1056     struct hitsbytarget *ht;
1057     int count, i;
1058
1059     ht = hitsbytarget_nb(se, &count, nmem);
1060     if (version >= 2)
1061         qsort(ht, count, sizeof(struct hitsbytarget), cmp_ht_approx);
1062     else
1063         qsort(ht, count, sizeof(struct hitsbytarget), cmp_ht);
1064     for (i = 0; i < count && i < num && ht[i].hits > 0; i++)
1065     {
1066
1067         // do only print terms which have display names
1068
1069         wrbuf_puts(wrbuf, "<term>\n");
1070
1071         wrbuf_puts(wrbuf, "<id>");
1072         wrbuf_xmlputs(wrbuf, ht[i].id);
1073         wrbuf_puts(wrbuf, "</id>\n");
1074
1075         wrbuf_puts(wrbuf, "<name>");
1076         if (!ht[i].name || !ht[i].name[0])
1077             wrbuf_xmlputs(wrbuf, "NO TARGET NAME");
1078         else
1079             wrbuf_xmlputs(wrbuf, ht[i].name);
1080         wrbuf_puts(wrbuf, "</name>\n");
1081
1082         wrbuf_printf(wrbuf, "<frequency>" ODR_INT_PRINTF "</frequency>\n",
1083                      ht[i].hits);
1084
1085         if (version >= 2) {
1086             // Should not print if we know it isn't a approximation.
1087             wrbuf_printf(wrbuf, "<approximation>" ODR_INT_PRINTF "</approximation>\n", ht[i].approximation);
1088             wrbuf_printf(wrbuf, "<records>%d</records>\n", ht[i].records - ht[i].filtered);
1089             wrbuf_printf(wrbuf, "<filtered>%d</filtered>\n", ht[i].filtered);
1090         }
1091
1092         wrbuf_puts(wrbuf, "<state>");
1093         wrbuf_xmlputs(wrbuf, ht[i].state);
1094         wrbuf_puts(wrbuf, "</state>\n");
1095
1096         wrbuf_printf(wrbuf, "<diagnostic>%d</diagnostic>\n",
1097                      ht[i].diagnostic);
1098         wrbuf_puts(wrbuf, "</term>\n");
1099     }
1100     return count;
1101 }
1102
1103 void perform_termlist(struct http_channel *c, struct session *se,
1104                       const char *name, int num, int version)
1105 {
1106     int i, j;
1107     NMEM nmem_tmp = nmem_create();
1108     char **names;
1109     int num_names = 0;
1110
1111     if (!name)
1112         name = "*";
1113
1114     nmem_strsplit(nmem_tmp, ",", name, &names, &num_names);
1115
1116     session_enter(se, "perform_termlist");
1117
1118     for (j = 0; j < num_names; j++)
1119     {
1120         const char *tname;
1121         int must_generate_empty = 1; /* bug 5350 */
1122
1123         for (i = 0; i < se->num_termlists; i++)
1124         {
1125             tname = se->termlists[i].name;
1126             if (!strcmp(names[j], tname) || !strcmp(names[j], "*"))
1127             {
1128                 struct termlist_score **p = 0;
1129                 int len;
1130
1131                 wrbuf_puts(c->wrbuf, "<list name=\"");
1132                 wrbuf_xmlputs(c->wrbuf, tname);
1133                 wrbuf_puts(c->wrbuf, "\">\n");
1134                 must_generate_empty = 0;
1135
1136                 p = termlist_highscore(se->termlists[i].termlist, &len,
1137                                        nmem_tmp);
1138                 if (p)
1139                 {
1140                     int i;
1141                     for (i = 0; i < len && i < num; i++)
1142                     {
1143                         // prevent sending empty term elements
1144                         if (!p[i]->display_term || !p[i]->display_term[0])
1145                             continue;
1146
1147                         wrbuf_puts(c->wrbuf, "<term>");
1148                         wrbuf_puts(c->wrbuf, "<name>");
1149                         wrbuf_xmlputs(c->wrbuf, p[i]->display_term);
1150                         wrbuf_puts(c->wrbuf, "</name>");
1151
1152                         wrbuf_printf(c->wrbuf,
1153                                      "<frequency>%d</frequency>",
1154                                      p[i]->frequency);
1155                         wrbuf_puts(c->wrbuf, "</term>\n");
1156                     }
1157                 }
1158                 wrbuf_puts(c->wrbuf, "</list>\n");
1159             }
1160         }
1161         tname = "xtargets";
1162         if (!strcmp(names[j], tname) || !strcmp(names[j], "*"))
1163         {
1164             wrbuf_puts(c->wrbuf, "<list name=\"");
1165             wrbuf_xmlputs(c->wrbuf, tname);
1166             wrbuf_puts(c->wrbuf, "\">\n");
1167
1168             targets_termlist_nb(c->wrbuf, se, num, c->nmem, version);
1169             wrbuf_puts(c->wrbuf, "</list>\n");
1170             must_generate_empty = 0;
1171         }
1172         if (must_generate_empty)
1173         {
1174             wrbuf_puts(c->wrbuf, "<list name=\"");
1175             wrbuf_xmlputs(c->wrbuf, names[j]);
1176             wrbuf_puts(c->wrbuf, "\"/>\n");
1177         }
1178     }
1179     session_leave(se, "perform_termlist");
1180     nmem_destroy(nmem_tmp);
1181 }
1182
1183 #ifdef MISSING_HEADERS
1184 void report_nmem_stats(void)
1185 {
1186     size_t in_use, is_free;
1187
1188     nmem_get_memory_in_use(&in_use);
1189     nmem_get_memory_free(&is_free);
1190
1191     yaz_log(YLOG_LOG, "nmem stat: use=%ld free=%ld",
1192             (long) in_use, (long) is_free);
1193 }
1194 #endif
1195
1196 struct record_cluster *show_single_start(struct session *se, const char *id,
1197                                          struct record_cluster **prev_r,
1198                                          struct record_cluster **next_r)
1199 {
1200     struct record_cluster *r = 0;
1201
1202     session_enter(se, "show_single_start");
1203     *prev_r = 0;
1204     *next_r = 0;
1205     if (se->reclist)
1206     {
1207         reclist_limit(se->reclist, se);
1208
1209         reclist_enter(se->reclist);
1210         while ((r = reclist_read_record(se->reclist)))
1211         {
1212             if (!strcmp(r->recid, id))
1213             {
1214                 *next_r = reclist_read_record(se->reclist);
1215                 break;
1216             }
1217             *prev_r = r;
1218         }
1219         reclist_leave(se->reclist);
1220     }
1221     if (!r)
1222         session_leave(se, "show_single_start");
1223     return r;
1224 }
1225
1226 void show_single_stop(struct session *se, struct record_cluster *rec)
1227 {
1228     session_leave(se, "show_single_stop");
1229 }
1230
1231
1232 struct record_cluster **show_range_start(struct session *se,
1233                                          struct reclist_sortparms *sp,
1234                                          int start, int *num, int *total,
1235                                          Odr_int *sumhits, Odr_int *approx_hits)
1236 {
1237     struct record_cluster **recs = 0;
1238     struct reclist_sortparms *spp;
1239     struct client_list *l;
1240     int i;
1241 #if USE_TIMING
1242     yaz_timing_t t = yaz_timing_create();
1243 #endif
1244     session_enter(se, "show_range_start");
1245     *sumhits = 0;
1246     *approx_hits = 0;
1247     *total = 0;
1248     reclist_limit(se->reclist, se);
1249     if (se->relevance)
1250     {
1251         for (spp = sp; spp; spp = spp->next)
1252             if (spp->type == Metadata_sortkey_relevance)
1253             {
1254                 relevance_prepare_read(se->relevance, se->reclist);
1255                 break;
1256             }
1257         for (l = se->clients_active; l; l = l->next) {
1258             *sumhits += client_get_hits(l->client);
1259             *approx_hits += client_get_approximation(l->client);
1260         }
1261     }
1262     reclist_sort(se->reclist, sp);
1263
1264     reclist_enter(se->reclist);
1265     *total = reclist_get_num_records(se->reclist);
1266
1267     for (i = 0; i < start; i++)
1268         if (!reclist_read_record(se->reclist))
1269         {
1270             *num = 0;
1271             break;
1272         }
1273
1274     if (*num > 0)
1275         recs =
1276             nmem_malloc(se->nmem, *num * sizeof(struct record_cluster *));
1277     for (i = 0; i < *num; i++)
1278     {
1279         struct record_cluster *r = reclist_read_record(se->reclist);
1280         if (!r)
1281         {
1282             *num = i;
1283             break;
1284         }
1285         recs[i] = r;
1286     }
1287     reclist_leave(se->reclist);
1288 #if USE_TIMING
1289     yaz_timing_stop(t);
1290     yaz_log(YLOG_LOG, "show %6.5f %3.2f %3.2f",
1291             yaz_timing_get_real(t), yaz_timing_get_user(t),
1292             yaz_timing_get_sys(t));
1293     yaz_timing_destroy(&t);
1294 #endif
1295     return recs;
1296 }
1297
1298 void show_range_stop(struct session *se, struct record_cluster **recs)
1299 {
1300     session_leave(se, "show_range_stop");
1301 }
1302
1303 void statistics(struct session *se, struct statistics *stat)
1304 {
1305     struct client_list *l;
1306     int count = 0;
1307
1308     memset(stat, 0, sizeof(*stat));
1309     stat->num_hits = 0;
1310     for (l = se->clients_active; l; l = l->next)
1311     {
1312         struct client *cl = l->client;
1313         if (!client_get_connection(cl))
1314             stat->num_no_connection++;
1315         stat->num_hits += client_get_hits(cl);
1316         switch (client_get_state(cl))
1317         {
1318         case Client_Connecting: stat->num_connecting++; break;
1319         case Client_Working: stat->num_working++; break;
1320         case Client_Idle: stat->num_idle++; break;
1321         case Client_Failed: stat->num_failed++; break;
1322         case Client_Error: stat->num_error++; break;
1323         default: break;
1324         }
1325         count++;
1326     }
1327     stat->num_records = se->total_records;
1328
1329     stat->num_clients = count;
1330 }
1331
1332 static struct record_metadata *record_metadata_init(
1333     NMEM nmem, const char *value, enum conf_metadata_type type,
1334     struct _xmlAttr *attr)
1335 {
1336     struct record_metadata *rec_md = record_metadata_create(nmem);
1337     struct record_metadata_attr **attrp = &rec_md->attributes;
1338
1339     for (; attr; attr = attr->next)
1340     {
1341         if (attr->children && attr->children->content)
1342         {
1343             if (strcmp((const char *) attr->name, "type"))
1344             {  /* skip the "type" attribute.. Its value is already part of
1345                   the element in output (md-%s) and so repeating it here
1346                   is redundant */
1347                 *attrp = nmem_malloc(nmem, sizeof(**attrp));
1348                 (*attrp)->name =
1349                     nmem_strdup(nmem, (const char *) attr->name);
1350                 (*attrp)->value =
1351                     nmem_strdup(nmem, (const char *) attr->children->content);
1352                 attrp = &(*attrp)->next;
1353             }
1354         }
1355     }
1356     *attrp = 0;
1357
1358     if (type == Metadata_type_generic)
1359     {
1360         char *p = nmem_strdup(nmem, value);
1361
1362         p = normalize7bit_generic(p, " ,/.:([");
1363
1364         rec_md->data.text.disp = p;
1365         rec_md->data.text.sort = 0;
1366     }
1367     else if (type == Metadata_type_year || type == Metadata_type_date)
1368     {
1369         int first, last;
1370         int longdate = 0;
1371
1372         if (type == Metadata_type_date)
1373             longdate = 1;
1374         if (extract7bit_dates((char *) value, &first, &last, longdate) < 0)
1375             return 0;
1376
1377         rec_md->data.number.min = first;
1378         rec_md->data.number.max = last;
1379     }
1380     else
1381         return 0;
1382     return rec_md;
1383 }
1384
1385 static int get_mergekey_from_doc(xmlDoc *doc, xmlNode *root, const char *name,
1386                                  struct conf_service *service, WRBUF norm_wr)
1387 {
1388     xmlNode *n;
1389     int no_found = 0;
1390     for (n = root->children; n; n = n->next)
1391     {
1392         if (n->type != XML_ELEMENT_NODE)
1393             continue;
1394         if (!strcmp((const char *) n->name, "metadata"))
1395         {
1396             xmlChar *type = xmlGetProp(n, (xmlChar *) "type");
1397             if (type == NULL) {
1398                 yaz_log(YLOG_FATAL, "Missing type attribute on metadata element. Skipping!");
1399             }
1400             else if (!strcmp(name, (const char *) type))
1401             {
1402                 xmlChar *value = xmlNodeListGetString(doc, n->children, 1);
1403                 if (value)
1404                 {
1405                     const char *norm_str;
1406                     pp2_charset_token_t prt =
1407                         pp2_charset_token_create(service->charsets, "mergekey");
1408
1409                     pp2_charset_token_first(prt, (const char *) value, 0);
1410                     if (wrbuf_len(norm_wr) > 0)
1411                         wrbuf_puts(norm_wr, " ");
1412                     wrbuf_puts(norm_wr, name);
1413                     while ((norm_str =
1414                             pp2_charset_token_next(prt)))
1415                     {
1416                         if (*norm_str)
1417                         {
1418                             wrbuf_puts(norm_wr, " ");
1419                             wrbuf_puts(norm_wr, norm_str);
1420                         }
1421                     }
1422                     xmlFree(value);
1423                     pp2_charset_token_destroy(prt);
1424                     no_found++;
1425                 }
1426             }
1427             xmlFree(type);
1428         }
1429     }
1430     return no_found;
1431 }
1432
1433 static const char *get_mergekey(xmlDoc *doc, struct client *cl, int record_no,
1434                                 struct conf_service *service, NMEM nmem)
1435 {
1436     char *mergekey_norm = 0;
1437     xmlNode *root = xmlDocGetRootElement(doc);
1438     WRBUF norm_wr = wrbuf_alloc();
1439
1440     /* consider mergekey from XSL first */
1441     xmlChar *mergekey = xmlGetProp(root, (xmlChar *) "mergekey");
1442     if (mergekey)
1443     {
1444         const char *norm_str;
1445         pp2_charset_token_t prt =
1446             pp2_charset_token_create(service->charsets, "mergekey");
1447
1448         pp2_charset_token_first(prt, (const char *) mergekey, 0);
1449         while ((norm_str = pp2_charset_token_next(prt)))
1450         {
1451             if (*norm_str)
1452             {
1453                 if (wrbuf_len(norm_wr))
1454                     wrbuf_puts(norm_wr, " ");
1455                 wrbuf_puts(norm_wr, norm_str);
1456             }
1457         }
1458         pp2_charset_token_destroy(prt);
1459         xmlFree(mergekey);
1460     }
1461     else
1462     {
1463         /* no mergekey defined in XSL. Look for mergekey metadata instead */
1464         int field_id;
1465         for (field_id = 0; field_id < service->num_metadata; field_id++)
1466         {
1467             struct conf_metadata *ser_md = &service->metadata[field_id];
1468             if (ser_md->mergekey != Metadata_mergekey_no)
1469             {
1470                 int r = get_mergekey_from_doc(doc, root, ser_md->name,
1471                                               service, norm_wr);
1472                 if (r == 0 && ser_md->mergekey == Metadata_mergekey_required)
1473                 {
1474                     /* no mergekey on this one and it is required..
1475                        Generate unique key instead */
1476                     wrbuf_rewind(norm_wr);
1477                     break;
1478                 }
1479             }
1480         }
1481     }
1482
1483     /* generate unique key if none is not generated already or is empty */
1484     if (wrbuf_len(norm_wr) == 0)
1485     {
1486         wrbuf_printf(norm_wr, "position: %s-%d",
1487                      client_get_id(cl), record_no);
1488     }
1489     else
1490     {
1491         const char *lead = "content: ";
1492         wrbuf_insert(norm_wr, 0, lead, strlen(lead));
1493     }
1494     if (wrbuf_len(norm_wr) > 0)
1495         mergekey_norm = nmem_strdup(nmem, wrbuf_cstr(norm_wr));
1496     wrbuf_destroy(norm_wr);
1497     return mergekey_norm;
1498 }
1499
1500 /** \brief see if metadata for pz:recordfilter exists
1501     \param root xml root element of normalized record
1502     \param sdb session database for client
1503     \retval 0 if there is no metadata for pz:recordfilter
1504     \retval 1 if there is metadata for pz:recordfilter
1505
1506     If there is no pz:recordfilter defined, this function returns 1
1507     as well.
1508 */
1509
1510 static int check_record_filter(xmlNode *root, struct session_database *sdb)
1511 {
1512     int match = 0;
1513     xmlNode *n;
1514     const char *s;
1515     s = session_setting_oneval(sdb, PZ_RECORDFILTER);
1516
1517     if (!s || !*s)
1518         return 1;
1519
1520     for (n = root->children; n; n = n->next)
1521     {
1522         if (n->type != XML_ELEMENT_NODE)
1523             continue;
1524         if (!strcmp((const char *) n->name, "metadata"))
1525         {
1526             xmlChar *type = xmlGetProp(n, (xmlChar *) "type");
1527             if (type)
1528             {
1529                 size_t len;
1530                 int substring;
1531                 const char *eq;
1532
1533                 if ((eq = strchr(s, '=')))
1534                     substring = 0;
1535                 else if ((eq = strchr(s, '~')))
1536                     substring = 1;
1537                 if (eq)
1538                     len = eq - s;
1539                 else
1540                     len = strlen(s);
1541                 if (len == strlen((const char *)type) &&
1542                     !memcmp((const char *) type, s, len))
1543                 {
1544                     xmlChar *value = xmlNodeGetContent(n);
1545                     if (value && *value)
1546                     {
1547                         if (!eq ||
1548                             (substring && strstr((const char *) value, eq+1)) ||
1549                             (!substring && !strcmp((const char *) value, eq + 1)))
1550                             match = 1;
1551                     }
1552                     xmlFree(value);
1553                 }
1554                 xmlFree(type);
1555             }
1556         }
1557     }
1558     return match;
1559 }
1560
1561
1562 static int ingest_to_cluster(struct client *cl,
1563                              xmlDoc *xdoc,
1564                              xmlNode *root,
1565                              int record_no,
1566                              const char *mergekey_norm);
1567
1568 /** \brief ingest XML record
1569     \param cl client holds the result set for record
1570     \param rec record buffer (0 terminated)
1571     \param record_no record position (1, 2, ..)
1572     \param nmem working NMEM
1573     \retval 0 OK
1574     \retval -1 failure
1575     \retval -2 Filtered
1576 */
1577 int ingest_record(struct client *cl, const char *rec,
1578                   int record_no, NMEM nmem)
1579 {
1580     struct session *se = client_get_session(cl);
1581     int ret = 0;
1582     struct session_database *sdb = client_get_database(cl);
1583     struct conf_service *service = se->service;
1584     xmlDoc *xdoc = normalize_record(se, sdb, service, rec, nmem);
1585     xmlNode *root;
1586     const char *mergekey_norm;
1587
1588     if (!xdoc)
1589         return -1;
1590
1591     root = xmlDocGetRootElement(xdoc);
1592
1593     if (!check_record_filter(root, sdb))
1594     {
1595         session_log(se, YLOG_LOG, "Filtered out record no %d from %s", record_no, sdb->database->id);
1596         xmlFreeDoc(xdoc);
1597         return -2;
1598     }
1599
1600     mergekey_norm = get_mergekey(xdoc, cl, record_no, service, nmem);
1601     if (!mergekey_norm)
1602     {
1603         session_log(se, YLOG_WARN, "Got no mergekey");
1604         xmlFreeDoc(xdoc);
1605         return -1;
1606     }
1607     session_enter(se, "ingest_record");
1608     if (client_get_session(cl) == se)
1609         ret = ingest_to_cluster(cl, xdoc, root, record_no, mergekey_norm);
1610     session_leave(se, "ingest_record");
1611
1612     xmlFreeDoc(xdoc);
1613     return ret;
1614 }
1615
1616 //    struct conf_metadata *ser_md = &service->metadata[md_field_id];
1617 //    struct record_metadata *rec_md = record->metadata[md_field_id];
1618 static int match_metadata_local(struct conf_metadata *ser_md,
1619                                 struct record_metadata *rec_md0,
1620                                 char **values, int num_v)
1621 {
1622     int i;
1623     struct record_metadata *rec_md = rec_md0;
1624     for (i = 0; i < num_v; )
1625     {
1626         if (rec_md)
1627         {
1628             if (ser_md->type == Metadata_type_year
1629                 || ser_md->type == Metadata_type_date)
1630             {
1631                 int y = atoi(values[i]);
1632                 if (y >= rec_md->data.number.min
1633                     && y <= rec_md->data.number.max)
1634                     break;
1635             }
1636             else
1637             {
1638                 yaz_log(YLOG_DEBUG, "cmp: '%s' '%s'", rec_md->data.text.disp, values[i]);
1639                 if (!strcmp(rec_md->data.text.disp, values[i]))
1640                 {
1641                     // Value equals, should not be filtered.
1642                     break;
1643                 }
1644             }
1645             rec_md = rec_md->next;
1646         }
1647         else
1648         {
1649             rec_md = rec_md0;
1650             i++;
1651         }
1652     }
1653     return i < num_v ? 1 : 0;
1654 }
1655
1656 int session_check_cluster_limit(struct session *se, struct record_cluster *rec)
1657 {
1658     int i;
1659     struct conf_service *service = se->service;
1660     int ret = 1;
1661     const char *name;
1662     const char *value;
1663     NMEM nmem_tmp = nmem_create();
1664
1665     for (i = 0; (name = facet_limits_get(se->facet_limits, i, &value)); i++)
1666     {
1667         int j;
1668         for (j = 0; j < service->num_metadata; j++)
1669         {
1670             struct conf_metadata *md = service->metadata + j;
1671             if (!strcmp(md->name, name) && md->limitcluster)
1672             {
1673                 char **values = 0;
1674                 int num = 0;
1675                 int md_field_id =
1676                     conf_service_metadata_field_id(service,
1677                                                    md->limitcluster);
1678
1679                 if (md_field_id < 0)
1680                 {
1681                     ret = 0;
1682                     break;
1683                 }
1684
1685                 nmem_strsplit_escape2(nmem_tmp, "|", value, &values,
1686                                       &num, 1, '\\', 1);
1687
1688                 if (!match_metadata_local(&service->metadata[md_field_id],
1689                                           rec->metadata[md_field_id],
1690                                           values, num))
1691                 {
1692                     ret = 0;
1693                     break;
1694                 }
1695             }
1696         }
1697     }
1698     nmem_destroy(nmem_tmp);
1699     return ret;
1700 }
1701
1702 // Skip record on non-zero
1703 static int check_limit_local(struct client *cl,
1704                              struct record *record,
1705                              int record_no)
1706 {
1707     int skip_record = 0;
1708     struct session *se = client_get_session(cl);
1709     struct conf_service *service = se->service;
1710     NMEM nmem_tmp = nmem_create();
1711     struct session_database *sdb = client_get_database(cl);
1712     int l = 0;
1713     while (!skip_record)
1714     {
1715         int md_field_id;
1716         char **values = 0;
1717         int num_v = 0;
1718         const char *name =
1719             client_get_facet_limit_local(cl, sdb, &l, nmem_tmp,
1720                                          &num_v, &values);
1721         if (!name)
1722             break;
1723
1724         if (!strcmp(name, "*"))
1725         {
1726             for (md_field_id = 0; md_field_id < service->num_metadata;
1727                  md_field_id++)
1728             {
1729                 if (match_metadata_local(
1730                         &service->metadata[md_field_id],
1731                         record->metadata[md_field_id],
1732                         values, num_v))
1733                     break;
1734             }
1735             if (md_field_id == service->num_metadata)
1736                 skip_record = 1;
1737         }
1738         else
1739         {
1740             md_field_id = conf_service_metadata_field_id(service, name);
1741             if (md_field_id < 0)
1742             {
1743                 skip_record = 1;
1744                 break;
1745             }
1746             if (!match_metadata_local(
1747                     &service->metadata[md_field_id],
1748                     record->metadata[md_field_id],
1749                     values, num_v))
1750             {
1751                 skip_record = 1;
1752             }
1753         }
1754     }
1755     nmem_destroy(nmem_tmp);
1756     return skip_record;
1757 }
1758
1759 static int ingest_to_cluster(struct client *cl,
1760                              xmlDoc *xdoc,
1761                              xmlNode *root,
1762                              int record_no,
1763                              const char *mergekey_norm)
1764 {
1765     xmlNode *n;
1766     xmlChar *type = 0;
1767     xmlChar *value = 0;
1768     struct session *se = client_get_session(cl);
1769     struct conf_service *service = se->service;
1770     int term_factor = 1;
1771     struct record_cluster *cluster;
1772     struct record_metadata **metadata0;
1773     struct session_database *sdb = client_get_database(cl);
1774     struct record *record = record_create(se->nmem,
1775                                           service->num_metadata,
1776                                           service->num_sortkeys, cl,
1777                                           record_no);
1778
1779     for (n = root->children; n; n = n->next)
1780     {
1781         if (type)
1782             xmlFree(type);
1783         if (value)
1784             xmlFree(value);
1785         type = value = 0;
1786
1787         if (n->type != XML_ELEMENT_NODE)
1788             continue;
1789         if (!strcmp((const char *) n->name, "metadata"))
1790         {
1791             struct conf_metadata *ser_md = 0;
1792             struct record_metadata **wheretoput = 0;
1793             struct record_metadata *rec_md = 0;
1794             int md_field_id = -1;
1795
1796             type = xmlGetProp(n, (xmlChar *) "type");
1797             value = xmlNodeListGetString(xdoc, n->children, 1);
1798
1799             if (!type || !value || !*value)
1800                 continue;
1801
1802             md_field_id
1803                 = conf_service_metadata_field_id(service, (const char *) type);
1804             if (md_field_id < 0)
1805             {
1806                 if (se->number_of_warnings_unknown_metadata == 0)
1807                 {
1808                     session_log(se, YLOG_WARN,
1809                             "Ignoring unknown metadata element: %s", type);
1810                 }
1811                 se->number_of_warnings_unknown_metadata++;
1812                 continue;
1813             }
1814
1815             ser_md = &service->metadata[md_field_id];
1816
1817             // non-merged metadata
1818             rec_md = record_metadata_init(se->nmem, (const char *) value,
1819                                           ser_md->type, n->properties);
1820             if (!rec_md)
1821             {
1822                 session_log(se, YLOG_WARN, "bad metadata data '%s' "
1823                             "for element '%s'", value, type);
1824                 continue;
1825             }
1826             wheretoput = &record->metadata[md_field_id];
1827             while (*wheretoput)
1828                 wheretoput = &(*wheretoput)->next;
1829             *wheretoput = rec_md;
1830         }
1831     }
1832
1833     if (check_limit_local(cl, record, record_no))
1834     {
1835         session_log(se, YLOG_LOG, "Facet filtered out record no %d from %s",
1836                     record_no, sdb->database->id);
1837         if (type)
1838             xmlFree(type);
1839         if (value)
1840             xmlFree(value);
1841         return -2;
1842     }
1843     cluster = reclist_insert(se->reclist, service, record,
1844                              mergekey_norm, &se->total_merged);
1845     if (!cluster)
1846         return -1;
1847
1848     {
1849         const char *use_term_factor_str =
1850             session_setting_oneval(sdb, PZ_TERMLIST_TERM_FACTOR);
1851         if (use_term_factor_str && use_term_factor_str[0] == '1')
1852         {
1853             int maxrecs = client_get_maxrecs(cl);
1854             int hits = (int) client_get_hits(cl);
1855             term_factor = MAX(hits, maxrecs) /  MAX(1, maxrecs);
1856             assert(term_factor >= 1);
1857             yaz_log(YLOG_DEBUG, "Using term factor: %d (%d / %d)", term_factor, MAX(hits, maxrecs), MAX(1, maxrecs));
1858         }
1859     }
1860
1861     if (global_parameters.dump_records)
1862         session_log(se, YLOG_LOG, "Cluster id %s from %s (#%d)", cluster->recid,
1863                     sdb->database->id, record_no);
1864
1865
1866     relevance_newrec(se->relevance, cluster);
1867
1868     // original metadata, to check if first existence of a field
1869     metadata0 = xmalloc(sizeof(*metadata0) * service->num_metadata);
1870     memcpy(metadata0, cluster->metadata,
1871            sizeof(*metadata0) * service->num_metadata);
1872
1873     // now parsing XML record and adding data to cluster or record metadata
1874     for (n = root->children; n; n = n->next)
1875     {
1876         pp2_charset_token_t prt;
1877         if (type)
1878             xmlFree(type);
1879         if (value)
1880             xmlFree(value);
1881         type = value = 0;
1882
1883         if (n->type != XML_ELEMENT_NODE)
1884             continue;
1885         if (!strcmp((const char *) n->name, "metadata"))
1886         {
1887             struct conf_metadata *ser_md = 0;
1888             struct conf_sortkey *ser_sk = 0;
1889             struct record_metadata **wheretoput = 0;
1890             struct record_metadata *rec_md = 0;
1891             int md_field_id = -1;
1892             int sk_field_id = -1;
1893             const char *rank;
1894             xmlChar *xml_rank;
1895
1896             type = xmlGetProp(n, (xmlChar *) "type");
1897             value = xmlNodeListGetString(xdoc, n->children, 1);
1898
1899             if (!type || !value || !*value)
1900                 continue;
1901
1902             md_field_id
1903                 = conf_service_metadata_field_id(service, (const char *) type);
1904             if (md_field_id < 0)
1905                 continue;
1906
1907             ser_md = &service->metadata[md_field_id];
1908
1909             if (ser_md->sortkey_offset >= 0)
1910             {
1911                 sk_field_id = ser_md->sortkey_offset;
1912                 ser_sk = &service->sortkeys[sk_field_id];
1913             }
1914
1915             // merged metadata
1916             rec_md = record_metadata_init(se->nmem, (const char *) value,
1917                                           ser_md->type, 0);
1918
1919             // see if the field was not in cluster already (from beginning)
1920
1921             if (!rec_md)
1922                 continue;
1923
1924             xml_rank = xmlGetProp(n, (xmlChar *) "rank");
1925             rank = xml_rank ? (const char *) xml_rank : ser_md->rank;
1926
1927             wheretoput = &cluster->metadata[md_field_id];
1928
1929             if (ser_md->merge == Metadata_merge_first)
1930             {
1931                 if (!metadata0[md_field_id])
1932                 {
1933                     while (*wheretoput)
1934                         wheretoput = &(*wheretoput)->next;
1935                     *wheretoput = rec_md;
1936                 }
1937             }
1938             else if (ser_md->merge == Metadata_merge_unique)
1939             {
1940                 while (*wheretoput)
1941                 {
1942                     if (!strcmp((const char *) (*wheretoput)->data.text.disp,
1943                                 rec_md->data.text.disp))
1944                         break;
1945                     wheretoput = &(*wheretoput)->next;
1946                 }
1947                 if (!*wheretoput)
1948                     *wheretoput = rec_md;
1949             }
1950             else if (ser_md->merge == Metadata_merge_longest)
1951             {
1952                 if (!*wheretoput
1953                     || strlen(rec_md->data.text.disp)
1954                     > strlen((*wheretoput)->data.text.disp))
1955                 {
1956                     *wheretoput = rec_md;
1957                     if (ser_sk)
1958                     {
1959                         const char *sort_str = 0;
1960                         int skip_article =
1961                             ser_sk->type == Metadata_sortkey_skiparticle;
1962
1963                         if (!cluster->sortkeys[sk_field_id])
1964                             cluster->sortkeys[sk_field_id] =
1965                                 nmem_malloc(se->nmem,
1966                                             sizeof(union data_types));
1967
1968                         prt =
1969                             pp2_charset_token_create(service->charsets, "sort");
1970
1971                         pp2_charset_token_first(prt, rec_md->data.text.disp,
1972                                                 skip_article);
1973
1974                         pp2_charset_token_next(prt);
1975
1976                         sort_str = pp2_get_sort(prt);
1977
1978                         cluster->sortkeys[sk_field_id]->text.disp =
1979                             rec_md->data.text.disp;
1980                         if (!sort_str)
1981                         {
1982                             sort_str = rec_md->data.text.disp;
1983                             session_log(se, YLOG_WARN,
1984                                     "Could not make sortkey. Bug #1858");
1985                         }
1986                         cluster->sortkeys[sk_field_id]->text.sort =
1987                             nmem_strdup(se->nmem, sort_str);
1988                         pp2_charset_token_destroy(prt);
1989                     }
1990                 }
1991             }
1992             else if (ser_md->merge == Metadata_merge_all)
1993             {
1994                 while (*wheretoput)
1995                     wheretoput = &(*wheretoput)->next;
1996                 *wheretoput = rec_md;
1997             }
1998             else if (ser_md->merge == Metadata_merge_range)
1999             {
2000                 if (!*wheretoput)
2001                 {
2002                     *wheretoput = rec_md;
2003                     if (ser_sk)
2004                         cluster->sortkeys[sk_field_id]
2005                             = &rec_md->data;
2006                 }
2007                 else
2008                 {
2009                     int this_min = rec_md->data.number.min;
2010                     int this_max = rec_md->data.number.max;
2011                     if (this_min < (*wheretoput)->data.number.min)
2012                         (*wheretoput)->data.number.min = this_min;
2013                     if (this_max > (*wheretoput)->data.number.max)
2014                         (*wheretoput)->data.number.max = this_max;
2015                 }
2016             }
2017
2018             // ranking of _all_ fields enabled ...
2019             if (rank)
2020             {
2021                 relevance_countwords(se->relevance, cluster,
2022                                      (char *) value, rank, ser_md->name);
2023             }
2024
2025             // construct facets ... unless the client already has reported them
2026             if (ser_md->termlist && !client_has_facet(cl, (char *) type))
2027             {
2028                 if (ser_md->type == Metadata_type_year)
2029                 {
2030                     char year[64];
2031                     sprintf(year, "%d", rec_md->data.number.max);
2032
2033                     add_facet(se, (char *) type, year, term_factor);
2034                     if (rec_md->data.number.max != rec_md->data.number.min)
2035                     {
2036                         sprintf(year, "%d", rec_md->data.number.min);
2037                         add_facet(se, (char *) type, year, term_factor);
2038                     }
2039                 }
2040                 else
2041                     add_facet(se, (char *) type, (char *) value, term_factor);
2042             }
2043
2044             // cleaning up
2045             if (xml_rank)
2046                 xmlFree(xml_rank);
2047             xmlFree(type);
2048             xmlFree(value);
2049             type = value = 0;
2050         }
2051         else
2052         {
2053             if (se->number_of_warnings_unknown_elements == 0)
2054                 session_log(se, YLOG_WARN,
2055                         "Unexpected element in internal record: %s", n->name);
2056             se->number_of_warnings_unknown_elements++;
2057         }
2058     }
2059     if (type)
2060         xmlFree(type);
2061     if (value)
2062         xmlFree(value);
2063
2064     xfree(metadata0);
2065     relevance_donerecord(se->relevance, cluster);
2066     se->total_records++;
2067
2068     return 0;
2069 }
2070
2071 void session_log(struct session *s, int level, const char *fmt, ...)
2072 {
2073     char buf[1024];
2074     va_list ap;
2075     va_start(ap, fmt);
2076
2077     yaz_vsnprintf(buf, sizeof(buf)-30, fmt, ap);
2078     yaz_log(level, "Session %u: %s", s->session_id, buf);
2079
2080     va_end(ap);
2081 }
2082
2083 /*
2084  * Local variables:
2085  * c-basic-offset: 4
2086  * c-file-style: "Stroustrup"
2087  * indent-tabs-mode: nil
2088  * End:
2089  * vim: shiftwidth=4 tabstop=8 expandtab
2090  */
2091