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 void session_enter_ro(struct session *s, const char *caller)
 152 {
 153     assert(s);
 154     if (caller)
 155         session_log(s, YLOG_LOG, "Session read lock by %s", caller);
 156     pazpar2_lock_rdwr_rlock(&s->lock);
 157 }
 158
 159 void session_enter_rw(struct session *s, const char *caller)
 160 {
 161     assert(s);
 162     if (caller)
 163         session_log(s, YLOG_LOG, "Session write lock by %s", caller);
 164     pazpar2_lock_rdwr_wlock(&s->lock);
 165 }
 166
 167 void session_upgrade(struct session *s, const char *caller)
 168 {
 169     assert(s);
 170     if (caller)
 171         session_log(s, YLOG_LOG, "Session upgrade lock by %s", caller);
 172     pazpar2_lock_rdwr_upgrade(&s->lock);
 173 }
 174
 175 void session_leave_ro(struct session *s, const char *caller)
 176 {
 177     assert(s);
 178     if (caller)
 179         session_log(s, YLOG_LOG, "Session read unlock by %s", caller);
 180     pazpar2_lock_rdwr_runlock(&s->lock);
 181 }
 182
 183 void session_leave_rw(struct session *s, const char *caller)
 184 {
 185     assert(s);
 186     if (caller)
 187         session_log(s, YLOG_LOG, "Session write unlock by %s", caller);
 188     pazpar2_lock_rdwr_wunlock(&s->lock);
 189 }
 190
 191 void session_downgrade(struct session *s, const char *caller)
 192 {
 193     assert(s);
 194     if (caller)
 195         session_log(s, YLOG_LOG, "Session write unlock by %s", caller);
 196     pazpar2_lock_rdwr_downgrade(&s->lock);
 197 }
 198
 199 static void session_normalize_facet(struct session *s, const char *type,
 200                                     const char *value,
 201                                     WRBUF display_wrbuf,
 202                                     WRBUF facet_wrbuf)
 203 {
 204     struct conf_service *service = s->service;
 205     pp2_charset_token_t prt;
 206     const char *facet_component;
 207     int i;
 208     const char *icu_chain_id = 0;
 209
 210     for (i = 0; i < service->num_metadata; i++)
 211         if (!strcmp((service->metadata + i)->name, type))
 212             icu_chain_id = (service->metadata + i)->facetrule;
 213     if (!icu_chain_id)
 214         icu_chain_id = "facet";
 215     prt = pp2_charset_token_create(service->charsets, icu_chain_id);
 216     if (!prt)
 217     {
 218         yaz_log(YLOG_FATAL, "Unknown ICU chain '%s' for facet of type '%s'",
 219                 icu_chain_id, type);
 220         wrbuf_destroy(facet_wrbuf);
 221         wrbuf_destroy(display_wrbuf);
 222         return;
 223     }
 224     pp2_charset_token_first(prt, value, 0);
 225     while ((facet_component = pp2_charset_token_next(prt)))
 226     {
 227         const char *display_component;
 228         if (*facet_component)
 229         {
 230             if (wrbuf_len(facet_wrbuf))
 231                 wrbuf_puts(facet_wrbuf, " ");
 232             wrbuf_puts(facet_wrbuf, facet_component);
 233         }
 234         display_component = pp2_get_display(prt);
 235         if (display_component)
 236         {
 237             if (wrbuf_len(display_wrbuf))
 238                 wrbuf_puts(display_wrbuf, " ");
 239             wrbuf_puts(display_wrbuf, display_component);
 240         }
 241     }
 242     pp2_charset_token_destroy(prt);
 243 }
 244
 245 void add_facet(struct session *s, const char *type, const char *value, int count)
 246 {
 247     WRBUF facet_wrbuf = wrbuf_alloc();
 248     WRBUF display_wrbuf = wrbuf_alloc();
 249
 250     session_normalize_facet(s, type, value, display_wrbuf, facet_wrbuf);
 251
 252     if (wrbuf_len(facet_wrbuf))
 253     {
 254         int i;
 255         for (i = 0; i < s->num_termlists; i++)
 256             if (!strcmp(s->termlists[i].name, type))
 257                 break;
 258         if (i == s->num_termlists)
 259         {
 260             if (i == SESSION_MAX_TERMLISTS)
 261             {
 262                 session_log(s, YLOG_FATAL, "Too many termlists");
 263                 wrbuf_destroy(facet_wrbuf);
 264                 wrbuf_destroy(display_wrbuf);
 265                 return;
 266             }
 267
 268             s->termlists[i].name = nmem_strdup(s->nmem, type);
 269             s->termlists[i].termlist = termlist_create(s->nmem);
 270             s->num_termlists = i + 1;
 271         }
 272
 273 #if 0
 274         session_log(s, YLOG_LOG, "Facets for %s: %s norm:%s (%d)", type, value, wrbuf_cstr(facet_wrbuf), count);
 275 #endif
 276         termlist_insert(s->termlists[i].termlist, wrbuf_cstr(display_wrbuf),
 277                         wrbuf_cstr(facet_wrbuf), count);
 278     }
 279     wrbuf_destroy(facet_wrbuf);
 280     wrbuf_destroy(display_wrbuf);
 281 }
 282
 283 static xmlDoc *record_to_xml(struct session *se,
 284                              struct session_database *sdb, const char *rec)
 285 {
 286     struct database *db = sdb->database;
 287     xmlDoc *rdoc = 0;
 288
 289     rdoc = xmlParseMemory(rec, strlen(rec));
 290
 291     if (!rdoc)
 292     {
 293         session_log(se, YLOG_WARN, "Non-wellformed XML");
 294         return 0;
 295     }
 296
 297     if (global_parameters.dump_records)
 298     {
 299         session_log(se, YLOG_LOG, "Un-normalized record from %s", db->id);
 300         log_xml_doc(rdoc);
 301     }
 302
 303     return rdoc;
 304 }
 305
 306 #define MAX_XSLT_ARGS 16
 307
 308 // Add static values from session database settings if applicable
 309 static void insert_settings_parameters(struct session_database *sdb,
 310                                        struct conf_service *service,
 311                                        char **parms,
 312                                        NMEM nmem)
 313 {
 314     int i;
 315     int nparms = 0;
 316     int offset = 0;
 317
 318     for (i = 0; i < service->num_metadata; i++)
 319     {
 320         struct conf_metadata *md = &service->metadata[i];
 321         int setting;
 322
 323         if (md->setting == Metadata_setting_parameter &&
 324             (setting = settings_lookup_offset(service, md->name)) >= 0)
 325         {
 326             const char *val = session_setting_oneval(sdb, setting);
 327             if (val && nparms < MAX_XSLT_ARGS)
 328             {
 329                 char *buf;
 330                 int len = strlen(val);
 331                 buf = nmem_malloc(nmem, len + 3);
 332                 buf[0] = '\'';
 333                 strcpy(buf + 1, val);
 334                 buf[len+1] = '\'';
 335                 buf[len+2] = '\0';
 336                 parms[offset++] = md->name;
 337                 parms[offset++] = buf;
 338                 nparms++;
 339             }
 340         }
 341     }
 342     parms[offset] = 0;
 343 }
 344
 345 // Add static values from session database settings if applicable
 346 static void insert_settings_values(struct session_database *sdb, xmlDoc *doc,
 347     struct conf_service *service)
 348 {
 349     int i;
 350
 351     for (i = 0; i < service->num_metadata; i++)
 352     {
 353         struct conf_metadata *md = &service->metadata[i];
 354         int offset;
 355
 356         if (md->setting == Metadata_setting_postproc &&
 357             (offset = settings_lookup_offset(service, md->name)) >= 0)
 358         {
 359             const char *val = session_setting_oneval(sdb, offset);
 360             if (val)
 361             {
 362                 xmlNode *r = xmlDocGetRootElement(doc);
 363                 xmlNode *n = xmlNewTextChild(r, 0, (xmlChar *) "metadata",
 364                                              (xmlChar *) val);
 365                 xmlSetProp(n, (xmlChar *) "type", (xmlChar *) md->name);
 366             }
 367         }
 368     }
 369 }
 370
 371 static xmlDoc *normalize_record(struct session *se,
 372                                 struct session_database *sdb,
 373                                 struct conf_service *service,
 374                                 const char *rec, NMEM nmem)
 375 {
 376     xmlDoc *rdoc = record_to_xml(se, sdb, rec);
 377
 378     if (rdoc)
 379     {
 380         char *parms[MAX_XSLT_ARGS*2+1];
 381
 382         insert_settings_parameters(sdb, service, parms, nmem);
 383
 384         if (normalize_record_transform(sdb->map, &rdoc, (const char **)parms))
 385         {
 386             session_log(se, YLOG_WARN, "Normalize failed");
 387         }
 388         else
 389         {
 390             insert_settings_values(sdb, rdoc, service);
 391
 392             if (global_parameters.dump_records)
 393             {
 394                 session_log(se, YLOG_LOG, "Normalized record from %s",
 395                             sdb->database->id);
 396                 log_xml_doc(rdoc);
 397             }
 398         }
 399     }
 400     return rdoc;
 401 }
 402
 403 void session_settings_dump(struct session *se,
 404                            struct session_database *db,
 405                            WRBUF w)
 406 {
 407     if (db->settings)
 408     {
 409         int i, num = db->num_settings;
 410         for (i = 0; i < num; i++)
 411         {
 412             struct setting *s = db->settings[i];
 413             for (;s ; s = s->next)
 414             {
 415                 wrbuf_puts(w, "<set name=\"");
 416                 wrbuf_xmlputs(w, s->name);
 417                 wrbuf_puts(w, "\" value=\"");
 418                 wrbuf_xmlputs(w, s->value);
 419                 wrbuf_puts(w, "\"/>");
 420             }
 421             if (db->settings[i])
 422                 wrbuf_puts(w, "\n");
 423         }
 424     }
 425 }
 426
 427 // Retrieve first defined value for 'name' for given database.
 428 // Will be extended to take into account user associated with session
 429 const char *session_setting_oneval(struct session_database *db, int offset)
 430 {
 431     if (offset >= db->num_settings || !db->settings[offset])
 432         return "";
 433     return db->settings[offset]->value;
 434 }
 435
 436 // Prepare XSLT stylesheets for record normalization
 437 // Structures are allocated on the session_wide nmem to avoid having
 438 // to recompute this for every search. This would lead
 439 // to leaking if a single session was to repeatedly change the PZ_XSLT
 440 // setting. However, this is not a realistic use scenario.
 441 static int prepare_map(struct session *se, struct session_database *sdb)
 442 {
 443     if (sdb->settings && !sdb->map)
 444     {
 445         const char *s;
 446
 447         if (sdb->settings[PZ_XSLT] &&
 448             (s = session_setting_oneval(sdb, PZ_XSLT)))
 449         {
 450             char auto_stylesheet[256];
 451
 452             if (!strcmp(s, "auto"))
 453             {
 454                 const char *request_syntax = session_setting_oneval(
 455                     sdb, PZ_REQUESTSYNTAX);
 456                 if (request_syntax)
 457                 {
 458                     char *cp;
 459                     yaz_snprintf(auto_stylesheet, sizeof(auto_stylesheet),
 460                                  "%s.xsl", request_syntax);
 461                     for (cp = auto_stylesheet; *cp; cp++)
 462                     {
 463                         /* deliberately only consider ASCII */
 464                         if (*cp > 32 && *cp < 127)
 465                             *cp = tolower(*cp);
 466                     }
 467                     s = auto_stylesheet;
 468                 }
 469                 else
 470                 {
 471                     session_log(se, YLOG_WARN,
 472                                 "No pz:requestsyntax for auto stylesheet");
 473                 }
 474             }
 475             sdb->map = normalize_cache_get(se->normalize_cache,
 476                                            se->service, s);
 477             if (!sdb->map)
 478                 return -1;
 479         }
 480     }
 481     return 0;
 482 }
 483
 484 // called if watch should be removed because http_channel is to be destroyed
 485 static void session_watch_cancel(void *data, struct http_channel *c,
 486                                  void *data2)
 487 {
 488     struct session_watchentry *ent = data;
 489
 490     ent->fun = 0;
 491     ent->data = 0;
 492     ent->obs = 0;
 493 }
 494
 495 // set watch. Returns 0=OK, -1 if watch is already set
 496 int session_set_watch(struct session *s, int what,
 497                       session_watchfun fun, void *data,
 498                       struct http_channel *chan)
 499 {
 500     int ret;
 501     if (s->watchlist[what].fun)
 502         ret = -1;
 503     else
 504     {
 505
 506         s->watchlist[what].fun = fun;
 507         s->watchlist[what].data = data;
 508         s->watchlist[what].obs = http_add_observer(chan, &s->watchlist[what],
 509                                                    session_watch_cancel);
 510         ret = 0;
 511     }
 512     return ret;
 513 }
 514
 515 void session_alert_watch(struct session *s, int what)
 516 {
 517     assert(s);
 518     if (s->watchlist[what].fun)
 519     {
 520         /* our watch is no longer associated with http_channel */
 521         void *data;
 522         session_watchfun fun;
 523
 524         http_remove_observer(s->watchlist[what].obs);
 525         fun  = s->watchlist[what].fun;
 526         data = s->watchlist[what].data;
 527
 528         /* reset watch before fun is invoked - in case fun wants to set
 529            it again */
 530         s->watchlist[what].fun = 0;
 531         s->watchlist[what].data = 0;
 532         s->watchlist[what].obs = 0;
 533
 534         session_log(s, YLOG_DEBUG,
 535                     "Alert Watch: %d calling function: %p", what, fun);
 536         fun(data);
 537     }
 538 }
 539
 540 //callback for grep_databases
 541 static void select_targets_callback(struct session *se,
 542                                     struct session_database *db)
 543 {
 544     struct client *cl;
 545     struct client_list *l;
 546
 547     for (l = se->clients_cached; l; l = l->next)
 548         if (client_get_database(l->client) == db)
 549             break;
 550
 551     if (l)
 552         cl = l->client;
 553     else
 554     {
 555         cl = client_create(db->database->id);
 556         client_set_database(cl, db);
 557
 558         l = xmalloc(sizeof(*l));
 559         l->client = cl;
 560         l->next = se->clients_cached;
 561         se->clients_cached = l;
 562     }
 563     /* set session always. If may be 0 if client is not active */
 564     client_set_session(cl, se);
 565
 566     l = xmalloc(sizeof(*l));
 567     l->client = cl;
 568     l->next = se->clients_active;
 569     se->clients_active = l;
 570 }
 571
 572 static void session_reset_active_clients(struct session *se,
 573                                          struct client_list *new_list)
 574 {
 575     struct client_list *l;
 576
 577     l = se->clients_active;
 578     se->clients_active = new_list;
 579
 580     while (l)
 581     {
 582         struct client_list *l_next = l->next;
 583
 584         client_set_session(l->client, 0); /* mark client inactive */
 585
 586         xfree(l);
 587         l = l_next;
 588     }
 589 }
 590
 591 static void session_remove_cached_clients(struct session *se)
 592 {
 593     struct client_list *l;
 594
 595     session_reset_active_clients(se, 0);
 596
 597     l = se->clients_cached;
 598     se->clients_cached = 0;
 599
 600     while (l)
 601     {
 602         struct client_list *l_next = l->next;
 603         client_set_session(l->client, 0);
 604         client_set_database(l->client, 0);
 605         client_destroy(l->client);
 606         xfree(l);
 607         l = l_next;
 608     }
 609 }
 610
 611 // Associates a set of clients with a session;
 612 // Note: Session-databases represent databases with per-session
 613 // setting overrides
 614 static int select_targets(struct session *se, const char *filter)
 615 {
 616     return session_grep_databases(se, filter, select_targets_callback);
 617 }
 618
 619 int session_active_clients(struct session *s)
 620 {
 621     struct client_list *l;
 622     int res = 0;
 623
 624     for (l = s->clients_active; l; l = l->next)
 625         if (client_is_active(l->client))
 626             res++;
 627
 628     return res;
 629 }
 630
 631 int session_is_preferred_clients_ready(struct session *s)
 632 {
 633     struct client_list *l;
 634     int res = 0;
 635
 636     for (l = s->clients_active; l; l = l->next)
 637         if (client_is_active_preferred(l->client))
 638             res++;
 639     session_log(s, YLOG_DEBUG, "Has %d active preferred clients.", res);
 640     return res == 0;
 641 }
 642
 643 static void session_clear_set(struct session *se, struct reclist_sortparms *sp)
 644 {
 645     reclist_destroy(se->reclist);
 646     se->reclist = 0;
 647     if (nmem_total(se->nmem))
 648         session_log(se, YLOG_DEBUG, "NMEN operation usage %zd",
 649                     nmem_total(se->nmem));
 650     nmem_reset(se->nmem);
 651     se->total_records = se->total_merged = 0;
 652     se->num_termlists = 0;
 653
 654     /* reset list of sorted results and clear to relevance search */
 655     se->sorted_results = nmem_malloc(se->nmem, sizeof(*se->sorted_results));
 656     se->sorted_results->name = nmem_strdup(se->nmem, sp->name);
 657     se->sorted_results->increasing = sp->increasing;
 658     se->sorted_results->type = sp->type;
 659     se->sorted_results->next = 0;
 660
 661     session_log(se, YLOG_DEBUG, "clear_set session_sort: field=%s increasing=%d type=%d configured",
 662             sp->name, sp->increasing, sp->type);
 663
 664     se->reclist = reclist_create(se->nmem);
 665 }
 666
 667 static void session_sort_unlocked(struct session *se,
 668                                   struct reclist_sortparms *sp)
 669 {
 670     struct reclist_sortparms *sr;
 671     struct client_list *l;
 672     const char *field = sp->name;
 673     int increasing = sp->increasing;
 674     int type  = sp->type;
 675     int clients_research = 0;
 676
 677     yaz_log(YLOG_DEBUG, "session_sort field=%s increasing=%d type=%d", field, increasing, type);
 678     /* see if we already have sorted for this criteria */
 679     for (sr = se->sorted_results; sr; sr = sr->next)
 680     {
 681         if (!reclist_sortparms_cmp(sr,sp))
 682             break;
 683     }
 684     if (sr)
 685     {
 686         session_log(se, YLOG_DEBUG, "search_sort: field=%s increasing=%d type=%d already fetched",
 687                     field, increasing, type);
 688         return;
 689     }
 690     session_log(se, YLOG_DEBUG, "search_sort: field=%s increasing=%d type=%d must fetch",
 691                     field, increasing, type);
 692
 693     // We need to reset reclist on every sort that changes the records, not just for position
 694     // So if just one client requires new searching, we need to clear set.
 695     // Ask each of the client if sorting requires re-search due to native sort
 696     // If it does it will require us to
 697     for (l = se->clients_active; l; l = l->next)
 698     {
 699         struct client *cl = l->client;
 700         // Assume no re-search is required.
 701         client_parse_init(cl, 1);
 702         clients_research += client_parse_sort(cl, sp);
 703     }
 704     if (clients_research) {
 705         yaz_log(YLOG_DEBUG, "Reset results due to %d clients researching", clients_research);
 706         session_clear_set(se, sp);
 707     }
 708     else {
 709         // A new sorting based on same record set
 710         sr = nmem_malloc(se->nmem, sizeof(*sr));
 711         sr->name = nmem_strdup(se->nmem, field);
 712         sr->increasing = increasing;
 713         sr->type = type;
 714         sr->next = se->sorted_results;
 715         se->sorted_results = sr;
 716         session_log(se, YLOG_DEBUG, "No research/ingesting done");
 717         return ;
 718     }
 719     session_log(se, YLOG_DEBUG, "Re- search/ingesting for clients due to change in sort order");
 720
 721     for (l = se->clients_active; l; l = l->next)
 722     {
 723         struct client *cl = l->client;
 724         if (client_get_state(cl) == Client_Connecting ||
 725             client_get_state(cl) == Client_Idle ||
 726             client_get_state(cl) == Client_Working) {
 727             client_start_search(cl);
 728         }
 729         else {
 730             yaz_log(YLOG_DEBUG, "Client %s: No re-start/ingest in show. Wrong client state: %d",
 731                         client_get_id(cl), client_get_state(cl));
 732         }
 733
 734     }
 735 }
 736
 737 void session_sort(struct session *se, struct reclist_sortparms *sp)
 738 {
 739     session_enter_rw(se, "session_sort");
 740     session_sort_unlocked(se, sp);
 741     session_leave_rw(se, "session_sort");
 742 }
 743
 744 static
 745 enum pazpar2_error_code session_search_unlocked(struct session *se,
 746                                                 const char *query,
 747                                                 const char *startrecs,
 748                                                 const char *maxrecs,
 749                                                 const char *filter,
 750                                                 const char *limit,
 751                                                 const char **addinfo,
 752                                                 struct reclist_sortparms *sp)
 753 {
 754     int live_channels = 0;
 755     int no_working = 0;
 756     int no_failed_query = 0;
 757     int no_failed_limit = 0;
 758     struct client_list *l, *l0;
 759
 760     session_log(se, YLOG_DEBUG, "Search");
 761
 762     *addinfo = 0;
 763
 764     if (se->settings_modified)
 765         session_remove_cached_clients(se);
 766     else
 767         session_reset_active_clients(se, 0);
 768
 769     se->settings_modified = 0;
 770
 771     session_clear_set(se, sp);
 772     relevance_destroy(&se->relevance);
 773
 774     live_channels = select_targets(se, filter);
 775     if (!live_channels)
 776         return PAZPAR2_NO_TARGETS;
 777
 778     facet_limits_destroy(se->facet_limits);
 779     se->facet_limits = facet_limits_create(limit);
 780     if (!se->facet_limits)
 781     {
 782         *addinfo = "limit";
 783         return PAZPAR2_MALFORMED_PARAMETER_VALUE;
 784     }
 785
 786     l0 = se->clients_active;
 787     se->clients_active = 0;
 788
 789     for (l = l0; l; l = l->next)
 790     {
 791         int parse_ret;
 792         struct client *cl = l->client;
 793         client_parse_init(cl, 1);
 794         if (prepare_map(se, client_get_database(cl)) < 0)
 795             continue;
 796
 797         parse_ret = client_parse_query(cl, query, se->facet_limits);
 798         if (parse_ret == -1)
 799             no_failed_query++;
 800         else if (parse_ret == -2)
 801             no_failed_limit++;
 802         else if (parse_ret < 0)
 803             no_working++; /* other error, such as bad CCL map */
 804         else
 805         {
 806             client_parse_range(cl, startrecs, maxrecs);
 807             client_parse_sort(cl, sp);
 808             client_start_search(cl);
 809             no_working++;
 810         }
 811     }
 812     session_reset_active_clients(se, l0);
 813
 814     if (no_working == 0)
 815     {
 816         if (no_failed_query > 0)
 817         {
 818             *addinfo = "query";
 819             return PAZPAR2_MALFORMED_PARAMETER_VALUE;
 820         }
 821         else if (no_failed_limit > 0)
 822         {
 823             *addinfo = "limit";
 824             return PAZPAR2_MALFORMED_PARAMETER_VALUE;
 825         }
 826         else
 827             return PAZPAR2_NO_TARGETS;
 828     }
 829     return PAZPAR2_NO_ERROR;
 830 }
 831
 832 enum pazpar2_error_code session_search(struct session *se,
 833                                        const char *query,
 834                                        const char *startrecs,
 835                                        const char *maxrecs,
 836                                        const char *filter,
 837                                        const char *limit,
 838                                        const char **addinfo,
 839                                        struct reclist_sortparms *sp)
 840 {
 841     enum pazpar2_error_code c;
 842     session_enter_rw(se, "session_search");
 843     c = session_search_unlocked(se, query, startrecs, maxrecs, filter,
 844                                 limit, addinfo, sp);
 845     session_leave_rw(se, "session_search");
 846     return c;
 847 }
 848
 849 // Creates a new session_database object for a database
 850 static void session_init_databases_fun(void *context, struct database *db)
 851 {
 852     struct session *se = (struct session *) context;
 853     struct session_database *new = nmem_malloc(se->session_nmem, sizeof(*new));
 854     int i;
 855
 856     new->database = db;
 857
 858     new->map = 0;
 859     assert(db->settings);
 860     new->settings = nmem_malloc(se->session_nmem,
 861                                 sizeof(struct settings *) * db->num_settings);
 862     new->num_settings = db->num_settings;
 863     for (i = 0; i < db->num_settings; i++)
 864     {
 865         struct setting *setting = db->settings[i];
 866         new->settings[i] = setting;
 867     }
 868     new->next = se->databases;
 869     se->databases = new;
 870 }
 871
 872 // Doesn't free memory associated with sdb -- nmem takes care of that
 873 static void session_database_destroy(struct session_database *sdb)
 874 {
 875     sdb->map = 0;
 876 }
 877
 878 // Initialize session_database list -- this represents this session's view
 879 // of the database list -- subject to modification by the settings ws command
 880 void session_init_databases(struct session *se)
 881 {
 882     se->databases = 0;
 883     predef_grep_databases(se, se->service, session_init_databases_fun);
 884 }
 885
 886 // Probably session_init_databases_fun should be refactored instead of
 887 // called here.
 888 static struct session_database *load_session_database(struct session *se,
 889                                                       char *id)
 890 {
 891     struct database *db = new_database_inherit_settings(id, se->session_nmem, se->service->settings);
 892     session_init_databases_fun((void*) se, db);
 893
 894     // New sdb is head of se->databases list
 895     return se->databases;
 896 }
 897
 898 // Find an existing session database. If not found, load it
 899 static struct session_database *find_session_database(struct session *se,
 900                                                       char *id)
 901 {
 902     struct session_database *sdb;
 903
 904     for (sdb = se->databases; sdb; sdb = sdb->next)
 905         if (!strcmp(sdb->database->id, id))
 906             return sdb;
 907     return load_session_database(se, id);
 908 }
 909
 910 // Apply a session override to a database
 911 void session_apply_setting(struct session *se, char *dbname, char *setting,
 912                            char *value)
 913 {
 914     struct session_database *sdb = find_session_database(se, dbname);
 915     struct conf_service *service = se->service;
 916     struct setting *new = nmem_malloc(se->session_nmem, sizeof(*new));
 917     int offset = settings_create_offset(service, setting);
 918
 919     expand_settings_array(&sdb->settings, &sdb->num_settings, offset,
 920                           se->session_nmem);
 921     new->precedence = 0;
 922     new->target = dbname;
 923     new->name = setting;
 924     new->value = value;
 925     new->next = sdb->settings[offset];
 926     sdb->settings[offset] = new;
 927
 928     se->settings_modified = 1;
 929
 930     // Force later recompute of settings-driven data structures
 931     // (happens when a search starts and client connections are prepared)
 932     switch (offset)
 933     {
 934     case PZ_XSLT:
 935         if (sdb->map)
 936         {
 937             sdb->map = 0;
 938         }
 939         break;
 940     }
 941 }
 942
 943 void session_destroy(struct session *se)
 944 {
 945     struct session_database *sdb;
 946
 947     session_enter_rw(se, "session_destroy");
 948     session_log(se, YLOG_DEBUG, "Destroying");
 949     session_use(-1);
 950     session_remove_cached_clients(se);
 951
 952     for (sdb = se->databases; sdb; sdb = sdb->next)
 953         session_database_destroy(sdb);
 954     normalize_cache_destroy(se->normalize_cache);
 955     relevance_destroy(&se->relevance);
 956     reclist_destroy(se->reclist);
 957     if (nmem_total(se->nmem))
 958         session_log(se, YLOG_DEBUG, "NMEN operation usage %zd", nmem_total(se->nmem));
 959     if (nmem_total(se->session_nmem))
 960         session_log(se, YLOG_DEBUG, "NMEN session usage %zd", nmem_total(se->session_nmem));
 961     facet_limits_destroy(se->facet_limits);
 962     nmem_destroy(se->nmem);
 963     service_destroy(se->service);
 964
 965     session_leave_rw(se, "session_destroy");
 966     pazpar2_lock_rdwr_destroy(&se->lock);
 967 }
 968
 969 size_t session_get_memory_status(struct session *session) {
 970     size_t session_nmem;
 971     if (session == 0)
 972         return 0;
 973     session_enter_ro(session, "session_get_memory_status");
 974     session_nmem = nmem_total(session->nmem);
 975     session_leave_ro(session, "session_get_memory_status");
 976     return session_nmem;
 977 }
 978
 979
 980 struct session *session_create(NMEM nmem, struct conf_service *service,
 981                                unsigned session_id)
 982 {
 983     int i;
 984     struct session *session = nmem_malloc(nmem, sizeof(*session));
 985
 986     char tmp_str[50];
 987
 988     sprintf(tmp_str, "session#%u", session_id);
 989
 990     session->session_id = session_id;
 991     session_log(session, YLOG_DEBUG, "New");
 992     session->service = service;
 993     session->relevance = 0;
 994     session->total_records = 0;
 995     session->number_of_warnings_unknown_elements = 0;
 996     session->number_of_warnings_unknown_metadata = 0;
 997     session->num_termlists = 0;
 998     session->reclist = 0;
 999     session->clients_active = 0;
1000     session->clients_cached = 0;
1001     session->settings_modified = 0;
1002     session->session_nmem = nmem;
1003     session->nmem = nmem_create();
1004     session->databases = 0;
1005     session->sorted_results = 0;
1006     session->facet_limits = 0;
1007
1008     for (i = 0; i <= SESSION_WATCH_MAX; i++)
1009     {
1010         session->watchlist[i].data = 0;
1011         session->watchlist[i].fun = 0;
1012     }
1013     session->normalize_cache = normalize_cache_create();
1014
1015     pazpar2_lock_rdwr_init(&session->lock);
1016
1017     session_use(1);
1018     return session;
1019 }
1020
1021 const char * client_get_suggestions_xml(struct client *cl, WRBUF wrbuf);
1022
1023 static struct hitsbytarget *hitsbytarget_nb(struct session *se,
1024                                             int *count, NMEM nmem)
1025 {
1026     struct hitsbytarget *res = 0;
1027     struct client_list *l;
1028     size_t sz = 0;
1029
1030     for (l = se->clients_active; l; l = l->next)
1031         sz++;
1032
1033     res = nmem_malloc(nmem, sizeof(*res) * sz);
1034     *count = 0;
1035     for (l = se->clients_active; l; l = l->next)
1036     {
1037         struct client *cl = l->client;
1038         WRBUF w = wrbuf_alloc();
1039         const char *name = session_setting_oneval(client_get_database(cl),
1040                                                   PZ_NAME);
1041
1042         res[*count].id = client_get_id(cl);
1043         res[*count].name = *name ? name : "Unknown";
1044         res[*count].hits = client_get_hits(cl);
1045         res[*count].approximation = client_get_approximation(cl);
1046         res[*count].records = client_get_num_records(cl);
1047         res[*count].filtered = client_get_num_records_filtered(cl);
1048         res[*count].diagnostic =
1049             client_get_diagnostic(cl, &res[*count].message,
1050                                   &res[*count].addinfo);
1051         res[*count].state = client_get_state_str(cl);
1052         res[*count].connected  = client_get_connection(cl) ? 1 : 0;
1053         session_settings_dump(se, client_get_database(cl), w);
1054         res[*count].settings_xml = nmem_strdup(nmem, wrbuf_cstr(w));
1055         wrbuf_rewind(w);
1056         wrbuf_puts(w, "");
1057         res[*count].suggestions_xml = nmem_strdup(nmem, client_get_suggestions_xml(cl, w));
1058         wrbuf_destroy(w);
1059         (*count)++;
1060     }
1061     return res;
1062 }
1063
1064 struct hitsbytarget *get_hitsbytarget(struct session *se, int *count, NMEM nmem)
1065 {
1066     struct hitsbytarget *p;
1067     p = hitsbytarget_nb(se, count, nmem);
1068     return p;
1069 }
1070
1071 // Compares two hitsbytarget nodes by hitcount
1072 static int cmp_ht(const void *p1, const void *p2)
1073 {
1074     const struct hitsbytarget *h1 = p1;
1075     const struct hitsbytarget *h2 = p2;
1076     return h2->hits - h1->hits;
1077 }
1078
1079 // Compares two hitsbytarget nodes by hitcount
1080 static int cmp_ht_approx(const void *p1, const void *p2)
1081 {
1082     const struct hitsbytarget *h1 = p1;
1083     const struct hitsbytarget *h2 = p2;
1084     return h2->approximation - h1->approximation;
1085 }
1086
1087 static int targets_termlist_nb(WRBUF wrbuf, struct session *se, int num,
1088                                NMEM nmem, int version)
1089 {
1090     struct hitsbytarget *ht;
1091     int count, i;
1092
1093     ht = hitsbytarget_nb(se, &count, nmem);
1094     if (version >= 2)
1095         qsort(ht, count, sizeof(struct hitsbytarget), cmp_ht_approx);
1096     else
1097         qsort(ht, count, sizeof(struct hitsbytarget), cmp_ht);
1098     for (i = 0; i < count && i < num && ht[i].hits > 0; i++)
1099     {
1100
1101         // do only print terms which have display names
1102
1103         wrbuf_puts(wrbuf, "<term>\n");
1104
1105         wrbuf_puts(wrbuf, "<id>");
1106         wrbuf_xmlputs(wrbuf, ht[i].id);
1107         wrbuf_puts(wrbuf, "</id>\n");
1108
1109         wrbuf_puts(wrbuf, "<name>");
1110         if (!ht[i].name || !ht[i].name[0])
1111             wrbuf_xmlputs(wrbuf, "NO TARGET NAME");
1112         else
1113             wrbuf_xmlputs(wrbuf, ht[i].name);
1114         wrbuf_puts(wrbuf, "</name>\n");
1115
1116         wrbuf_printf(wrbuf, "<frequency>" ODR_INT_PRINTF "</frequency>\n",
1117                      ht[i].hits);
1118
1119         if (version >= 2) {
1120             // Should not print if we know it isn't a approximation.
1121             wrbuf_printf(wrbuf, "<approximation>" ODR_INT_PRINTF "</approximation>\n", ht[i].approximation);
1122             wrbuf_printf(wrbuf, "<records>%d</records>\n", ht[i].records - ht[i].filtered);
1123             wrbuf_printf(wrbuf, "<filtered>%d</filtered>\n", ht[i].filtered);
1124         }
1125
1126         wrbuf_puts(wrbuf, "<state>");
1127         wrbuf_xmlputs(wrbuf, ht[i].state);
1128         wrbuf_puts(wrbuf, "</state>\n");
1129
1130         wrbuf_printf(wrbuf, "<diagnostic>%d</diagnostic>\n",
1131                      ht[i].diagnostic);
1132         wrbuf_puts(wrbuf, "</term>\n");
1133     }
1134     return count;
1135 }
1136
1137 void perform_termlist(struct http_channel *c, struct session *se,
1138                       const char *name, int num, int version)
1139 {
1140     int i, j;
1141     NMEM nmem_tmp = nmem_create();
1142     char **names;
1143     int num_names = 0;
1144
1145     if (!name)
1146         name = "*";
1147
1148     nmem_strsplit(nmem_tmp, ",", name, &names, &num_names);
1149
1150     session_enter_ro(se, "perform_termlist");
1151
1152     for (j = 0; j < num_names; j++)
1153     {
1154         const char *tname;
1155         int must_generate_empty = 1; /* bug 5350 */
1156
1157         for (i = 0; i < se->num_termlists; i++)
1158         {
1159             tname = se->termlists[i].name;
1160             if (!strcmp(names[j], tname) || !strcmp(names[j], "*"))
1161             {
1162                 struct termlist_score **p = 0;
1163                 int len;
1164
1165                 wrbuf_puts(c->wrbuf, "<list name=\"");
1166                 wrbuf_xmlputs(c->wrbuf, tname);
1167                 wrbuf_puts(c->wrbuf, "\">\n");
1168                 must_generate_empty = 0;
1169
1170                 p = termlist_highscore(se->termlists[i].termlist, &len,
1171                                        nmem_tmp);
1172                 if (p)
1173                 {
1174                     int i;
1175                     for (i = 0; i < len && i < num; i++)
1176                     {
1177                         // prevent sending empty term elements
1178                         if (!p[i]->display_term || !p[i]->display_term[0])
1179                             continue;
1180
1181                         wrbuf_puts(c->wrbuf, "<term>");
1182                         wrbuf_puts(c->wrbuf, "<name>");
1183                         wrbuf_xmlputs(c->wrbuf, p[i]->display_term);
1184                         wrbuf_puts(c->wrbuf, "</name>");
1185
1186                         wrbuf_printf(c->wrbuf,
1187                                      "<frequency>%d</frequency>",
1188                                      p[i]->frequency);
1189                         wrbuf_puts(c->wrbuf, "</term>\n");
1190                     }
1191                 }
1192                 wrbuf_puts(c->wrbuf, "</list>\n");
1193             }
1194         }
1195         tname = "xtargets";
1196         if (!strcmp(names[j], tname) || !strcmp(names[j], "*"))
1197         {
1198             wrbuf_puts(c->wrbuf, "<list name=\"");
1199             wrbuf_xmlputs(c->wrbuf, tname);
1200             wrbuf_puts(c->wrbuf, "\">\n");
1201
1202             targets_termlist_nb(c->wrbuf, se, num, c->nmem, version);
1203             wrbuf_puts(c->wrbuf, "</list>\n");
1204             must_generate_empty = 0;
1205         }
1206         if (must_generate_empty)
1207         {
1208             wrbuf_puts(c->wrbuf, "<list name=\"");
1209             wrbuf_xmlputs(c->wrbuf, names[j]);
1210             wrbuf_puts(c->wrbuf, "\"/>\n");
1211         }
1212     }
1213     session_leave_ro(se, "perform_termlist");
1214     nmem_destroy(nmem_tmp);
1215 }
1216
1217 #ifdef MISSING_HEADERS
1218 void report_nmem_stats(void)
1219 {
1220     size_t in_use, is_free;
1221
1222     nmem_get_memory_in_use(&in_use);
1223     nmem_get_memory_free(&is_free);
1224
1225     yaz_log(YLOG_LOG, "nmem stat: use=%ld free=%ld",
1226             (long) in_use, (long) is_free);
1227 }
1228 #endif
1229
1230 struct record_cluster *show_single_start(struct session *se, const char *id,
1231                                          struct record_cluster **prev_r,
1232                                          struct record_cluster **next_r)
1233 {
1234     struct record_cluster *r = 0;
1235
1236     session_enter_ro(se, "show_single_start");
1237     *prev_r = 0;
1238     *next_r = 0;
1239     if (se->reclist)
1240     {
1241         reclist_limit(se->reclist, se);
1242
1243         reclist_enter(se->reclist);
1244         while ((r = reclist_read_record(se->reclist)))
1245         {
1246             if (!strcmp(r->recid, id))
1247             {
1248                 *next_r = reclist_read_record(se->reclist);
1249                 break;
1250             }
1251             *prev_r = r;
1252         }
1253         reclist_leave(se->reclist);
1254     }
1255     if (!r)
1256         session_leave_ro(se, "show_single_start");
1257     return r;
1258 }
1259
1260 void show_single_stop(struct session *se, struct record_cluster *rec)
1261 {
1262     session_leave_ro(se, "show_single_stop");
1263 }
1264
1265
1266 struct record_cluster **show_range_start(struct session *se,
1267                                          struct reclist_sortparms *sp,
1268                                          int start, int *num, int *total,
1269                                          Odr_int *sumhits, Odr_int *approx_hits)
1270 {
1271     struct record_cluster **recs = 0;
1272     struct reclist_sortparms *spp;
1273     struct client_list *l;
1274     int i;
1275 #if USE_TIMING
1276     yaz_timing_t t = yaz_timing_create();
1277 #endif
1278     *sumhits = 0;
1279     *approx_hits = 0;
1280     *total = 0;
1281     reclist_limit(se->reclist, se);
1282     if (se->relevance)
1283     {
1284         for (spp = sp; spp; spp = spp->next)
1285             if (spp->type == Metadata_sortkey_relevance)
1286             {
1287                 relevance_prepare_read(se->relevance, se->reclist);
1288                 break;
1289             }
1290         for (l = se->clients_active; l; l = l->next) {
1291             *sumhits += client_get_hits(l->client);
1292             *approx_hits += client_get_approximation(l->client);
1293         }
1294     }
1295     reclist_sort(se->reclist, sp);
1296
1297     reclist_enter(se->reclist);
1298     *total = reclist_get_num_records(se->reclist);
1299
1300     for (i = 0; i < start; i++)
1301         if (!reclist_read_record(se->reclist))
1302         {
1303             *num = 0;
1304             break;
1305         }
1306
1307     if (*num > 0)
1308         recs =
1309             nmem_malloc(se->nmem, *num * sizeof(struct record_cluster *));
1310     for (i = 0; i < *num; i++)
1311     {
1312         struct record_cluster *r = reclist_read_record(se->reclist);
1313         if (!r)
1314         {
1315             *num = i;
1316             break;
1317         }
1318         recs[i] = r;
1319     }
1320     reclist_leave(se->reclist);
1321 #if USE_TIMING
1322     yaz_timing_stop(t);
1323     yaz_log(YLOG_LOG, "show %6.5f %3.2f %3.2f",
1324             yaz_timing_get_real(t), yaz_timing_get_user(t),
1325             yaz_timing_get_sys(t));
1326     yaz_timing_destroy(&t);
1327 #endif
1328     return recs;
1329 }
1330
1331 void show_range_stop(struct session *se, struct record_cluster **recs)
1332 {
1333 }
1334
1335 void statistics(struct session *se, struct statistics *stat)
1336 {
1337     struct client_list *l;
1338     int count = 0;
1339
1340     memset(stat, 0, sizeof(*stat));
1341     stat->num_hits = 0;
1342     for (l = se->clients_active; l; l = l->next)
1343     {
1344         struct client *cl = l->client;
1345         if (!client_get_connection(cl))
1346             stat->num_no_connection++;
1347         stat->num_hits += client_get_hits(cl);
1348         switch (client_get_state(cl))
1349         {
1350         case Client_Connecting: stat->num_connecting++; break;
1351         case Client_Working: stat->num_working++; break;
1352         case Client_Idle: stat->num_idle++; break;
1353         case Client_Failed: stat->num_failed++; break;
1354         case Client_Error: stat->num_error++; break;
1355         default: break;
1356         }
1357         count++;
1358     }
1359     stat->num_records = se->total_records;
1360
1361     stat->num_clients = count;
1362 }
1363
1364 static struct record_metadata *record_metadata_init(
1365     NMEM nmem, const char *value, enum conf_metadata_type type,
1366     struct _xmlAttr *attr)
1367 {
1368     struct record_metadata *rec_md = record_metadata_create(nmem);
1369     struct record_metadata_attr **attrp = &rec_md->attributes;
1370
1371     for (; attr; attr = attr->next)
1372     {
1373         if (attr->children && attr->children->content)
1374         {
1375             if (strcmp((const char *) attr->name, "type"))
1376             {  /* skip the "type" attribute.. Its value is already part of
1377                   the element in output (md-%s) and so repeating it here
1378                   is redundant */
1379                 *attrp = nmem_malloc(nmem, sizeof(**attrp));
1380                 (*attrp)->name =
1381                     nmem_strdup(nmem, (const char *) attr->name);
1382                 (*attrp)->value =
1383                     nmem_strdup(nmem, (const char *) attr->children->content);
1384                 attrp = &(*attrp)->next;
1385             }
1386         }
1387     }
1388     *attrp = 0;
1389
1390     if (type == Metadata_type_generic)
1391     {
1392         char *p = nmem_strdup(nmem, value);
1393
1394         p = normalize7bit_generic(p, " ,/.:([");
1395
1396         rec_md->data.text.disp = p;
1397         rec_md->data.text.sort = 0;
1398     }
1399     else if (type == Metadata_type_year || type == Metadata_type_date)
1400     {
1401         int first, last;
1402         int longdate = 0;
1403
1404         if (type == Metadata_type_date)
1405             longdate = 1;
1406         if (extract7bit_dates((char *) value, &first, &last, longdate) < 0)
1407             return 0;
1408
1409         rec_md->data.number.min = first;
1410         rec_md->data.number.max = last;
1411     }
1412     else
1413         return 0;
1414     return rec_md;
1415 }
1416
1417 static int get_mergekey_from_doc(xmlDoc *doc, xmlNode *root, const char *name,
1418                                  struct conf_service *service, WRBUF norm_wr)
1419 {
1420     xmlNode *n;
1421     int no_found = 0;
1422     for (n = root->children; n; n = n->next)
1423     {
1424         if (n->type != XML_ELEMENT_NODE)
1425             continue;
1426         if (!strcmp((const char *) n->name, "metadata"))
1427         {
1428             xmlChar *type = xmlGetProp(n, (xmlChar *) "type");
1429             if (type == NULL) {
1430                 yaz_log(YLOG_FATAL, "Missing type attribute on metadata element. Skipping!");
1431             }
1432             else if (!strcmp(name, (const char *) type))
1433             {
1434                 xmlChar *value = xmlNodeListGetString(doc, n->children, 1);
1435                 if (value)
1436                 {
1437                     const char *norm_str;
1438                     pp2_charset_token_t prt =
1439                         pp2_charset_token_create(service->charsets, "mergekey");
1440
1441                     pp2_charset_token_first(prt, (const char *) value, 0);
1442                     if (wrbuf_len(norm_wr) > 0)
1443                         wrbuf_puts(norm_wr, " ");
1444                     wrbuf_puts(norm_wr, name);
1445                     while ((norm_str =
1446                             pp2_charset_token_next(prt)))
1447                     {
1448                         if (*norm_str)
1449                         {
1450                             wrbuf_puts(norm_wr, " ");
1451                             wrbuf_puts(norm_wr, norm_str);
1452                         }
1453                     }
1454                     xmlFree(value);
1455                     pp2_charset_token_destroy(prt);
1456                     no_found++;
1457                 }
1458             }
1459             xmlFree(type);
1460         }
1461     }
1462     return no_found;
1463 }
1464
1465 static const char *get_mergekey(xmlDoc *doc, struct client *cl, int record_no,
1466                                 struct conf_service *service, NMEM nmem)
1467 {
1468     char *mergekey_norm = 0;
1469     xmlNode *root = xmlDocGetRootElement(doc);
1470     WRBUF norm_wr = wrbuf_alloc();
1471
1472     /* consider mergekey from XSL first */
1473     xmlChar *mergekey = xmlGetProp(root, (xmlChar *) "mergekey");
1474     if (mergekey)
1475     {
1476         const char *norm_str;
1477         pp2_charset_token_t prt =
1478             pp2_charset_token_create(service->charsets, "mergekey");
1479
1480         pp2_charset_token_first(prt, (const char *) mergekey, 0);
1481         while ((norm_str = pp2_charset_token_next(prt)))
1482         {
1483             if (*norm_str)
1484             {
1485                 if (wrbuf_len(norm_wr))
1486                     wrbuf_puts(norm_wr, " ");
1487                 wrbuf_puts(norm_wr, norm_str);
1488             }
1489         }
1490         pp2_charset_token_destroy(prt);
1491         xmlFree(mergekey);
1492     }
1493     else
1494     {
1495         /* no mergekey defined in XSL. Look for mergekey metadata instead */
1496         int field_id;
1497         for (field_id = 0; field_id < service->num_metadata; field_id++)
1498         {
1499             struct conf_metadata *ser_md = &service->metadata[field_id];
1500             if (ser_md->mergekey != Metadata_mergekey_no)
1501             {
1502                 int r = get_mergekey_from_doc(doc, root, ser_md->name,
1503                                               service, norm_wr);
1504                 if (r == 0 && ser_md->mergekey == Metadata_mergekey_required)
1505                 {
1506                     /* no mergekey on this one and it is required..
1507                        Generate unique key instead */
1508                     wrbuf_rewind(norm_wr);
1509                     break;
1510                 }
1511             }
1512         }
1513     }
1514
1515     /* generate unique key if none is not generated already or is empty */
1516     if (wrbuf_len(norm_wr) == 0)
1517     {
1518         wrbuf_printf(norm_wr, "position: %s-%d",
1519                      client_get_id(cl), record_no);
1520     }
1521     else
1522     {
1523         const char *lead = "content: ";
1524         wrbuf_insert(norm_wr, 0, lead, strlen(lead));
1525     }
1526     if (wrbuf_len(norm_wr) > 0)
1527         mergekey_norm = nmem_strdup(nmem, wrbuf_cstr(norm_wr));
1528     wrbuf_destroy(norm_wr);
1529     return mergekey_norm;
1530 }
1531
1532 /** \brief see if metadata for pz:recordfilter exists
1533     \param root xml root element of normalized record
1534     \param sdb session database for client
1535     \retval 0 if there is no metadata for pz:recordfilter
1536     \retval 1 if there is metadata for pz:recordfilter
1537
1538     If there is no pz:recordfilter defined, this function returns 1
1539     as well.
1540 */
1541
1542 static int check_record_filter(xmlNode *root, struct session_database *sdb)
1543 {
1544     int match = 0;
1545     xmlNode *n;
1546     const char *s;
1547     s = session_setting_oneval(sdb, PZ_RECORDFILTER);
1548
1549     if (!s || !*s)
1550         return 1;
1551
1552     for (n = root->children; n; n = n->next)
1553     {
1554         if (n->type != XML_ELEMENT_NODE)
1555             continue;
1556         if (!strcmp((const char *) n->name, "metadata"))
1557         {
1558             xmlChar *type = xmlGetProp(n, (xmlChar *) "type");
1559             if (type)
1560             {
1561                 size_t len;
1562                 int substring;
1563                 const char *eq;
1564
1565                 if ((eq = strchr(s, '=')))
1566                     substring = 0;
1567                 else if ((eq = strchr(s, '~')))
1568                     substring = 1;
1569                 if (eq)
1570                     len = eq - s;
1571                 else
1572                     len = strlen(s);
1573                 if (len == strlen((const char *)type) &&
1574                     !memcmp((const char *) type, s, len))
1575                 {
1576                     xmlChar *value = xmlNodeGetContent(n);
1577                     if (value && *value)
1578                     {
1579                         if (!eq ||
1580                             (substring && strstr((const char *) value, eq+1)) ||
1581                             (!substring && !strcmp((const char *) value, eq + 1)))
1582                             match = 1;
1583                     }
1584                     xmlFree(value);
1585                 }
1586                 xmlFree(type);
1587             }
1588         }
1589     }
1590     return match;
1591 }
1592
1593
1594 static int ingest_to_cluster(struct client *cl,
1595                              xmlDoc *xdoc,
1596                              xmlNode *root,
1597                              int record_no,
1598                              const char *mergekey_norm);
1599
1600 /** \brief ingest XML record
1601     \param cl client holds the result set for record
1602     \param rec record buffer (0 terminated)
1603     \param record_no record position (1, 2, ..)
1604     \param nmem working NMEM
1605     \retval 0 OK
1606     \retval -1 failure
1607     \retval -2 Filtered
1608 */
1609 int ingest_record(struct client *cl, const char *rec,
1610                   int record_no, NMEM nmem)
1611 {
1612     struct session *se = client_get_session(cl);
1613     int ret = 0;
1614     struct session_database *sdb = client_get_database(cl);
1615     struct conf_service *service = se->service;
1616     xmlDoc *xdoc = normalize_record(se, sdb, service, rec, nmem);
1617     xmlNode *root;
1618     const char *mergekey_norm;
1619
1620     if (!xdoc)
1621         return -1;
1622
1623     root = xmlDocGetRootElement(xdoc);
1624
1625     if (!check_record_filter(root, sdb))
1626     {
1627         session_log(se, YLOG_LOG, "Filtered out record no %d from %s", record_no, sdb->database->id);
1628         xmlFreeDoc(xdoc);
1629         return -2;
1630     }
1631
1632     mergekey_norm = get_mergekey(xdoc, cl, record_no, service, nmem);
1633     if (!mergekey_norm)
1634     {
1635         session_log(se, YLOG_WARN, "Got no mergekey");
1636         xmlFreeDoc(xdoc);
1637         return -1;
1638     }
1639     assert(client_get_session(cl) == se);
1640
1641     ret = ingest_to_cluster(cl, xdoc, root, record_no, mergekey_norm);
1642
1643     xmlFreeDoc(xdoc);
1644     return ret;
1645 }
1646
1647 //    struct conf_metadata *ser_md = &service->metadata[md_field_id];
1648 //    struct record_metadata *rec_md = record->metadata[md_field_id];
1649 static int match_metadata_local(struct conf_metadata *ser_md,
1650                                 struct record_metadata *rec_md0,
1651                                 char **values, int num_v)
1652 {
1653     int i;
1654     struct record_metadata *rec_md = rec_md0;
1655     for (i = 0; i < num_v; )
1656     {
1657         if (rec_md)
1658         {
1659             if (ser_md->type == Metadata_type_year
1660                 || ser_md->type == Metadata_type_date)
1661             {
1662                 int y = atoi(values[i]);
1663                 if (y >= rec_md->data.number.min
1664                     && y <= rec_md->data.number.max)
1665                     break;
1666             }
1667             else
1668             {
1669                 yaz_log(YLOG_DEBUG, "cmp: '%s' '%s'", rec_md->data.text.disp, values[i]);
1670                 if (!strcmp(rec_md->data.text.disp, values[i]))
1671                 {
1672                     // Value equals, should not be filtered.
1673                     break;
1674                 }
1675             }
1676             rec_md = rec_md->next;
1677         }
1678         else
1679         {
1680             rec_md = rec_md0;
1681             i++;
1682         }
1683     }
1684     return i < num_v ? 1 : 0;
1685 }
1686
1687 int session_check_cluster_limit(struct session *se, struct record_cluster *rec)
1688 {
1689     int i;
1690     struct conf_service *service = se->service;
1691     int ret = 1;
1692     const char *name;
1693     const char *value;
1694     NMEM nmem_tmp = nmem_create();
1695
1696     for (i = 0; (name = facet_limits_get(se->facet_limits, i, &value)); i++)
1697     {
1698         int j;
1699         for (j = 0; j < service->num_metadata; j++)
1700         {
1701             struct conf_metadata *md = service->metadata + j;
1702             if (!strcmp(md->name, name) && md->limitcluster)
1703             {
1704                 char **values = 0;
1705                 int num = 0;
1706                 int md_field_id =
1707                     conf_service_metadata_field_id(service,
1708                                                    md->limitcluster);
1709
1710                 if (md_field_id < 0)
1711                 {
1712                     ret = 0;
1713                     break;
1714                 }
1715
1716                 nmem_strsplit_escape2(nmem_tmp, "|", value, &values,
1717                                       &num, 1, '\\', 1);
1718
1719                 if (!match_metadata_local(&service->metadata[md_field_id],
1720                                           rec->metadata[md_field_id],
1721                                           values, num))
1722                 {
1723                     ret = 0;
1724                     break;
1725                 }
1726             }
1727         }
1728     }
1729     nmem_destroy(nmem_tmp);
1730     return ret;
1731 }
1732
1733 // Skip record on non-zero
1734 static int check_limit_local(struct client *cl,
1735                              struct record *record,
1736                              int record_no)
1737 {
1738     int skip_record = 0;
1739     struct session *se = client_get_session(cl);
1740     struct conf_service *service = se->service;
1741     NMEM nmem_tmp = nmem_create();
1742     struct session_database *sdb = client_get_database(cl);
1743     int l = 0;
1744     while (!skip_record)
1745     {
1746         int md_field_id;
1747         char **values = 0;
1748         int num_v = 0;
1749         const char *name =
1750             client_get_facet_limit_local(cl, sdb, &l, nmem_tmp,
1751                                          &num_v, &values);
1752         if (!name)
1753             break;
1754
1755         if (!strcmp(name, "*"))
1756         {
1757             for (md_field_id = 0; md_field_id < service->num_metadata;
1758                  md_field_id++)
1759             {
1760                 if (match_metadata_local(
1761                         &service->metadata[md_field_id],
1762                         record->metadata[md_field_id],
1763                         values, num_v))
1764                     break;
1765             }
1766             if (md_field_id == service->num_metadata)
1767                 skip_record = 1;
1768         }
1769         else
1770         {
1771             md_field_id = conf_service_metadata_field_id(service, name);
1772             if (md_field_id < 0)
1773             {
1774                 skip_record = 1;
1775                 break;
1776             }
1777             if (!match_metadata_local(
1778                     &service->metadata[md_field_id],
1779                     record->metadata[md_field_id],
1780                     values, num_v))
1781             {
1782                 skip_record = 1;
1783             }
1784         }
1785     }
1786     nmem_destroy(nmem_tmp);
1787     return skip_record;
1788 }
1789
1790 static int ingest_to_cluster(struct client *cl,
1791                              xmlDoc *xdoc,
1792                              xmlNode *root,
1793                              int record_no,
1794                              const char *mergekey_norm)
1795 {
1796     xmlNode *n;
1797     xmlChar *type = 0;
1798     xmlChar *value = 0;
1799     struct session *se = client_get_session(cl);
1800     struct conf_service *service = se->service;
1801     int term_factor = 1;
1802     struct record_cluster *cluster;
1803     struct record_metadata **metadata0;
1804     struct session_database *sdb = client_get_database(cl);
1805     struct record *record = record_create(se->nmem,
1806                                           service->num_metadata,
1807                                           service->num_sortkeys, cl,
1808                                           record_no);
1809
1810     for (n = root->children; n; n = n->next)
1811     {
1812         if (type)
1813             xmlFree(type);
1814         if (value)
1815             xmlFree(value);
1816         type = value = 0;
1817
1818         if (n->type != XML_ELEMENT_NODE)
1819             continue;
1820         if (!strcmp((const char *) n->name, "metadata"))
1821         {
1822             struct conf_metadata *ser_md = 0;
1823             struct record_metadata **wheretoput = 0;
1824             struct record_metadata *rec_md = 0;
1825             int md_field_id = -1;
1826
1827             type = xmlGetProp(n, (xmlChar *) "type");
1828             value = xmlNodeListGetString(xdoc, n->children, 1);
1829
1830             if (!type || !value || !*value)
1831                 continue;
1832
1833             md_field_id
1834                 = conf_service_metadata_field_id(service, (const char *) type);
1835             if (md_field_id < 0)
1836             {
1837                 if (se->number_of_warnings_unknown_metadata == 0)
1838                 {
1839                     session_log(se, YLOG_WARN,
1840                             "Ignoring unknown metadata element: %s", type);
1841                 }
1842                 se->number_of_warnings_unknown_metadata++;
1843                 continue;
1844             }
1845
1846             ser_md = &service->metadata[md_field_id];
1847
1848             // non-merged metadata
1849             rec_md = record_metadata_init(se->nmem, (const char *) value,
1850                                           ser_md->type, n->properties);
1851             if (!rec_md)
1852             {
1853                 session_log(se, YLOG_WARN, "bad metadata data '%s' "
1854                             "for element '%s'", value, type);
1855                 continue;
1856             }
1857             wheretoput = &record->metadata[md_field_id];
1858             while (*wheretoput)
1859                 wheretoput = &(*wheretoput)->next;
1860             *wheretoput = rec_md;
1861         }
1862     }
1863
1864     if (check_limit_local(cl, record, record_no))
1865     {
1866         session_log(se, YLOG_LOG, "Facet filtered out record no %d from %s",
1867                     record_no, sdb->database->id);
1868         if (type)
1869             xmlFree(type);
1870         if (value)
1871             xmlFree(value);
1872         return -2;
1873     }
1874     cluster = reclist_insert(se->reclist, service, record,
1875                              mergekey_norm, &se->total_merged);
1876     if (!cluster)
1877         return 0; // complete match with existing record
1878
1879     {
1880         const char *use_term_factor_str =
1881             session_setting_oneval(sdb, PZ_TERMLIST_TERM_FACTOR);
1882         if (use_term_factor_str && use_term_factor_str[0] == '1')
1883         {
1884             int maxrecs = client_get_maxrecs(cl);
1885             int hits = (int) client_get_hits(cl);
1886             term_factor = MAX(hits, maxrecs) /  MAX(1, maxrecs);
1887             assert(term_factor >= 1);
1888             yaz_log(YLOG_DEBUG, "Using term factor: %d (%d / %d)", term_factor, MAX(hits, maxrecs), MAX(1, maxrecs));
1889         }
1890     }
1891
1892     if (global_parameters.dump_records)
1893         session_log(se, YLOG_LOG, "Cluster id %s from %s (#%d)", cluster->recid,
1894                     sdb->database->id, record_no);
1895
1896
1897     relevance_newrec(se->relevance, cluster);
1898
1899     // original metadata, to check if first existence of a field
1900     metadata0 = xmalloc(sizeof(*metadata0) * service->num_metadata);
1901     memcpy(metadata0, cluster->metadata,
1902            sizeof(*metadata0) * service->num_metadata);
1903
1904     // now parsing XML record and adding data to cluster or record metadata
1905     for (n = root->children; n; n = n->next)
1906     {
1907         pp2_charset_token_t prt;
1908         if (type)
1909             xmlFree(type);
1910         if (value)
1911             xmlFree(value);
1912         type = value = 0;
1913
1914         if (n->type != XML_ELEMENT_NODE)
1915             continue;
1916         if (!strcmp((const char *) n->name, "metadata"))
1917         {
1918             struct conf_metadata *ser_md = 0;
1919             struct conf_sortkey *ser_sk = 0;
1920             struct record_metadata **wheretoput = 0;
1921             struct record_metadata *rec_md = 0;
1922             int md_field_id = -1;
1923             int sk_field_id = -1;
1924             const char *rank;
1925             xmlChar *xml_rank;
1926
1927             type = xmlGetProp(n, (xmlChar *) "type");
1928             value = xmlNodeListGetString(xdoc, n->children, 1);
1929
1930             if (!type || !value || !*value)
1931                 continue;
1932
1933             md_field_id
1934                 = conf_service_metadata_field_id(service, (const char *) type);
1935             if (md_field_id < 0)
1936                 continue;
1937
1938             ser_md = &service->metadata[md_field_id];
1939
1940             if (ser_md->sortkey_offset >= 0)
1941             {
1942                 sk_field_id = ser_md->sortkey_offset;
1943                 ser_sk = &service->sortkeys[sk_field_id];
1944             }
1945
1946             // merged metadata
1947             rec_md = record_metadata_init(se->nmem, (const char *) value,
1948                                           ser_md->type, 0);
1949
1950             // see if the field was not in cluster already (from beginning)
1951
1952             if (!rec_md)
1953                 continue;
1954
1955             xml_rank = xmlGetProp(n, (xmlChar *) "rank");
1956             rank = xml_rank ? (const char *) xml_rank : ser_md->rank;
1957
1958             wheretoput = &cluster->metadata[md_field_id];
1959
1960             if (ser_md->merge == Metadata_merge_first)
1961             {
1962                 if (!metadata0[md_field_id])
1963                 {
1964                     while (*wheretoput)
1965                         wheretoput = &(*wheretoput)->next;
1966                     *wheretoput = rec_md;
1967                 }
1968             }
1969             else if (ser_md->merge == Metadata_merge_unique)
1970             {
1971                 while (*wheretoput)
1972                 {
1973                     if (!strcmp((const char *) (*wheretoput)->data.text.disp,
1974                                 rec_md->data.text.disp))
1975                         break;
1976                     wheretoput = &(*wheretoput)->next;
1977                 }
1978                 if (!*wheretoput)
1979                     *wheretoput = rec_md;
1980             }
1981             else if (ser_md->merge == Metadata_merge_longest)
1982             {
1983                 if (!*wheretoput
1984                     || strlen(rec_md->data.text.disp)
1985                     > strlen((*wheretoput)->data.text.disp))
1986                 {
1987                     *wheretoput = rec_md;
1988                     if (ser_sk)
1989                     {
1990                         const char *sort_str = 0;
1991                         int skip_article =
1992                             ser_sk->type == Metadata_sortkey_skiparticle;
1993
1994                         if (!cluster->sortkeys[sk_field_id])
1995                             cluster->sortkeys[sk_field_id] =
1996                                 nmem_malloc(se->nmem,
1997                                             sizeof(union data_types));
1998
1999                         prt =
2000                             pp2_charset_token_create(service->charsets, "sort");
2001
2002                         pp2_charset_token_first(prt, rec_md->data.text.disp,
2003                                                 skip_article);
2004
2005                         pp2_charset_token_next(prt);
2006
2007                         sort_str = pp2_get_sort(prt);
2008
2009                         cluster->sortkeys[sk_field_id]->text.disp =
2010                             rec_md->data.text.disp;
2011                         if (!sort_str)
2012                         {
2013                             sort_str = rec_md->data.text.disp;
2014                             session_log(se, YLOG_WARN,
2015                                     "Could not make sortkey. Bug #1858");
2016                         }
2017                         cluster->sortkeys[sk_field_id]->text.sort =
2018                             nmem_strdup(se->nmem, sort_str);
2019                         pp2_charset_token_destroy(prt);
2020                     }
2021                 }
2022             }
2023             else if (ser_md->merge == Metadata_merge_all)
2024             {
2025                 while (*wheretoput)
2026                     wheretoput = &(*wheretoput)->next;
2027                 *wheretoput = rec_md;
2028             }
2029             else if (ser_md->merge == Metadata_merge_range)
2030             {
2031                 if (!*wheretoput)
2032                 {
2033                     *wheretoput = rec_md;
2034                     if (ser_sk)
2035                         cluster->sortkeys[sk_field_id]
2036                             = &rec_md->data;
2037                 }
2038                 else
2039                 {
2040                     int this_min = rec_md->data.number.min;
2041                     int this_max = rec_md->data.number.max;
2042                     if (this_min < (*wheretoput)->data.number.min)
2043                         (*wheretoput)->data.number.min = this_min;
2044                     if (this_max > (*wheretoput)->data.number.max)
2045                         (*wheretoput)->data.number.max = this_max;
2046                 }
2047             }
2048
2049             // ranking of _all_ fields enabled ...
2050             if (rank)
2051             {
2052                 relevance_countwords(se->relevance, cluster,
2053                                      (char *) value, rank, ser_md->name);
2054             }
2055
2056             // construct facets ... unless the client already has reported them
2057             if (ser_md->termlist && !client_has_facet(cl, (char *) type))
2058             {
2059                 if (ser_md->type == Metadata_type_year)
2060                 {
2061                     char year[64];
2062                     sprintf(year, "%d", rec_md->data.number.max);
2063
2064                     add_facet(se, (char *) type, year, term_factor);
2065                     if (rec_md->data.number.max != rec_md->data.number.min)
2066                     {
2067                         sprintf(year, "%d", rec_md->data.number.min);
2068                         add_facet(se, (char *) type, year, term_factor);
2069                     }
2070                 }
2071                 else
2072                     add_facet(se, (char *) type, (char *) value, term_factor);
2073             }
2074
2075             // cleaning up
2076             if (xml_rank)
2077                 xmlFree(xml_rank);
2078             xmlFree(type);
2079             xmlFree(value);
2080             type = value = 0;
2081         }
2082         else
2083         {
2084             if (se->number_of_warnings_unknown_elements == 0)
2085                 session_log(se, YLOG_WARN,
2086                         "Unexpected element in internal record: %s", n->name);
2087             se->number_of_warnings_unknown_elements++;
2088         }
2089     }
2090     if (type)
2091         xmlFree(type);
2092     if (value)
2093         xmlFree(value);
2094
2095     xfree(metadata0);
2096     relevance_donerecord(se->relevance, cluster);
2097     se->total_records++;
2098
2099     return 0;
2100 }
2101
2102 void session_log(struct session *s, int level, const char *fmt, ...)
2103 {
2104     char buf[1024];
2105     va_list ap;
2106     va_start(ap, fmt);
2107
2108     yaz_vsnprintf(buf, sizeof(buf)-30, fmt, ap);
2109     yaz_log(level, "Session %u: %s", s->session_id, buf);
2110
2111     va_end(ap);
2112 }
2113
2114 /*
2115  * Local variables:
2116  * c-basic-offset: 4
2117  * c-file-style: "Stroustrup"
2118  * indent-tabs-mode: nil
2119  * End:
2120  * vim: shiftwidth=4 tabstop=8 expandtab
2121  */
2122