src/logic.c

   1 /* $Id: logic.c,v 1.16 2007-04-21 12:00:54 adam Exp $
   2    Copyright (c) 2006-2007, Index Data.
   3
   4 This file is part of Pazpar2.
   5
   6 Pazpar2 is free software; you can redistribute it and/or modify it under
   7 the terms of the GNU General Public License as published by the Free
   8 Software Foundation; either version 2, or (at your option) any later
   9 version.
  10
  11 Pazpar2 is distributed in the hope that it will be useful, but WITHOUT ANY
  12 WARRANTY; without even the implied warranty of MERCHANTABILITY or
  13 FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License
  14 for more details.
  15
  16 You should have received a copy of the GNU General Public License
  17 along with Pazpar2; see the file LICENSE.  If not, write to the
  18 Free Software Foundation, 59 Temple Place - Suite 330, Boston, MA
  19 02111-1307, USA.
  20  */
  21
  22 // This file contains the primary business logic. Several parts of it should
  23 // Eventually be factored into separate modules.
  24
  25 #include <stdlib.h>
  26 #include <stdio.h>
  27 #include <string.h>
  28 #include <sys/time.h>
  29 #include <unistd.h>
  30 #include <sys/socket.h>
  31 #include <netdb.h>
  32 #include <signal.h>
  33 #include <ctype.h>
  34 #include <assert.h>
  35
  36 #include <yaz/marcdisp.h>
  37 #include <yaz/comstack.h>
  38 #include <yaz/tcpip.h>
  39 #include <yaz/proto.h>
  40 #include <yaz/readconf.h>
  41 #include <yaz/pquery.h>
  42 #include <yaz/otherinfo.h>
  43 #include <yaz/yaz-util.h>
  44 #include <yaz/nmem.h>
  45 #include <yaz/query-charset.h>
  46 #include <yaz/querytowrbuf.h>
  47 #if YAZ_VERSIONL >= 0x020163
  48 #include <yaz/oid_db.h>
  49 #endif
  50
  51 #if HAVE_CONFIG_H
  52 #include "cconfig.h"
  53 #endif
  54
  55 #define USE_TIMING 0
  56 #if USE_TIMING
  57 #include <yaz/timing.h>
  58 #endif
  59
  60 #include <netinet/in.h>
  61
  62 #include "pazpar2.h"
  63 #include "eventl.h"
  64 #include "http.h"
  65 #include "termlists.h"
  66 #include "reclists.h"
  67 #include "relevance.h"
  68 #include "config.h"
  69 #include "database.h"
  70 #include "settings.h"
  71
  72 #define MAX_CHUNK 15
  73
  74 static void client_fatal(struct client *cl);
  75 static void connection_destroy(struct connection *co);
  76 static int client_prep_connection(struct client *cl);
  77 static void ingest_records(struct client *cl, Z_Records *r);
  78 void session_alert_watch(struct session *s, int what);
  79
  80 static struct connection *connection_freelist = 0;
  81 static struct client *client_freelist = 0;
  82
  83 static char *client_states[] = {
  84     "Client_Connecting",
  85     "Client_Connected",
  86     "Client_Idle",
  87     "Client_Initializing",
  88     "Client_Searching",
  89     "Client_Presenting",
  90     "Client_Error",
  91     "Client_Failed",
  92     "Client_Disconnected",
  93     "Client_Stopped"
  94 };
  95
  96 // Note: Some things in this structure will eventually move to configuration
  97 struct parameters global_parameters =
  98 {
  99     "",
 100     "",
 101     "",
 102     "",
 103     0,
 104     0,
 105     30,
 106     "81",
 107     "Index Data PazPar2",
 108     VERSION,
 109     600, // 10 minutes
 110     60,
 111     100,
 112     MAX_CHUNK,
 113     0,
 114     0
 115 };
 116
 117 static int send_apdu(struct client *c, Z_APDU *a)
 118 {
 119     struct connection *co = c->connection;
 120     char *buf;
 121     int len, r;
 122
 123     if (!z_APDU(global_parameters.odr_out, &a, 0, 0))
 124     {
 125         odr_perror(global_parameters.odr_out, "Encoding APDU");
 126         abort();
 127     }
 128     buf = odr_getbuf(global_parameters.odr_out, &len, 0);
 129     r = cs_put(co->link, buf, len);
 130     if (r < 0)
 131     {
 132         yaz_log(YLOG_WARN, "cs_put: %s", cs_errmsg(cs_errno(co->link)));
 133         return -1;
 134     }
 135     else if (r == 1)
 136     {
 137         fprintf(stderr, "cs_put incomplete (ParaZ does not handle that)\n");
 138         exit(1);
 139     }
 140     odr_reset(global_parameters.odr_out); /* release the APDU structure  */
 141     co->state = Conn_Waiting;
 142     return 0;
 143 }
 144
 145 // Set authentication token in init if one is set for the client
 146 // TODO: Extend this to handle other schemes than open (should be simple)
 147 static void init_authentication(struct client *cl, Z_InitRequest *req)
 148 {
 149     struct session_database *sdb = cl->database;
 150     char *auth = session_setting_oneval(sdb, PZ_AUTHENTICATION);
 151
 152     if (auth)
 153     {
 154         Z_IdAuthentication *idAuth = odr_malloc(global_parameters.odr_out,
 155                 sizeof(*idAuth));
 156         idAuth->which = Z_IdAuthentication_open;
 157         idAuth->u.open = auth;
 158         req->idAuthentication = idAuth;
 159     }
 160 }
 161
 162 static void send_init(IOCHAN i)
 163 {
 164
 165     struct connection *co = iochan_getdata(i);
 166     struct client *cl = co->client;
 167     Z_APDU *a = zget_APDU(global_parameters.odr_out, Z_APDU_initRequest);
 168
 169     a->u.initRequest->implementationId = global_parameters.implementationId;
 170     a->u.initRequest->implementationName = global_parameters.implementationName;
 171     a->u.initRequest->implementationVersion =
 172         global_parameters.implementationVersion;
 173     ODR_MASK_SET(a->u.initRequest->options, Z_Options_search);
 174     ODR_MASK_SET(a->u.initRequest->options, Z_Options_present);
 175     ODR_MASK_SET(a->u.initRequest->options, Z_Options_namedResultSets);
 176
 177     ODR_MASK_SET(a->u.initRequest->protocolVersion, Z_ProtocolVersion_1);
 178     ODR_MASK_SET(a->u.initRequest->protocolVersion, Z_ProtocolVersion_2);
 179     ODR_MASK_SET(a->u.initRequest->protocolVersion, Z_ProtocolVersion_3);
 180
 181     init_authentication(cl, a->u.initRequest);
 182
 183     /* add virtual host if tunneling through Z39.50 proxy */
 184
 185     if (0 < strlen(global_parameters.zproxy_override)
 186         && 0 < strlen(cl->database->database->url))
 187     {
 188 #if YAZ_VERSIONL >= 0x020163
 189         yaz_oi_set_string_oid(&a->u.initRequest->otherInfo,
 190                               global_parameters.odr_out,
 191                               yaz_oid_userinfo_proxy,
 192                               1, cl->database->database->url);
 193 #else
 194         yaz_oi_set_string_oidval(&a->u.initRequest->otherInfo,
 195                                  global_parameters.odr_out, VAL_PROXY,
 196                                  1, cl->database->database->url);
 197 #endif
 198     }
 199
 200     if (send_apdu(cl, a) >= 0)
 201     {
 202         iochan_setflags(i, EVENT_INPUT);
 203         cl->state = Client_Initializing;
 204     }
 205     else
 206         cl->state = Client_Error;
 207     odr_reset(global_parameters.odr_out);
 208 }
 209
 210 // Recursively traverse query structure to extract terms.
 211 static void pull_terms(NMEM nmem, struct ccl_rpn_node *n, char **termlist, int *num)
 212 {
 213     char **words;
 214     int numwords;
 215     int i;
 216
 217     switch (n->kind)
 218     {
 219         case CCL_RPN_AND:
 220         case CCL_RPN_OR:
 221         case CCL_RPN_NOT:
 222         case CCL_RPN_PROX:
 223             pull_terms(nmem, n->u.p[0], termlist, num);
 224             pull_terms(nmem, n->u.p[1], termlist, num);
 225             break;
 226         case CCL_RPN_TERM:
 227             nmem_strsplit(nmem, " ", n->u.t.term, &words, &numwords);
 228             for (i = 0; i < numwords; i++)
 229                 termlist[(*num)++] = words[i];
 230             break;
 231         default: // NOOP
 232             break;
 233     }
 234 }
 235
 236 // Extract terms from query into null-terminated termlist
 237 static void extract_terms(NMEM nmem, struct ccl_rpn_node *query, char **termlist)
 238 {
 239     int num = 0;
 240
 241     pull_terms(nmem, query, termlist, &num);
 242     termlist[num] = 0;
 243 }
 244
 245 static void send_search(IOCHAN i)
 246 {
 247     struct connection *co = iochan_getdata(i);
 248     struct client *cl = co->client;
 249     struct session *se = cl->session;
 250     struct session_database *sdb = cl->database;
 251     Z_APDU *a = zget_APDU(global_parameters.odr_out, Z_APDU_searchRequest);
 252     int ndb;
 253     char **databaselist;
 254     Z_Query *zquery;
 255     int ssub = 0, lslb = 100000, mspn = 10;
 256     char *recsyn = 0;
 257     char *piggyback = 0;
 258     char *queryenc = 0;
 259     yaz_iconv_t iconv = 0;
 260
 261     yaz_log(YLOG_DEBUG, "Sending search to %s", cl->database->database->url);
 262
 263     // constructing RPN query
 264     a->u.searchRequest->query = zquery = odr_malloc(global_parameters.odr_out,
 265             sizeof(Z_Query));
 266     zquery->which = Z_Query_type_1;
 267     zquery->u.type_1 = p_query_rpn(global_parameters.odr_out, cl->pquery);
 268
 269     // converting to target encoding
 270     if ((queryenc = session_setting_oneval(sdb, PZ_QUERYENCODING))){
 271         iconv = yaz_iconv_open(queryenc, "UTF-8");
 272         if (iconv){
 273             yaz_query_charset_convert_rpnquery(zquery->u.type_1,
 274                                                global_parameters.odr_out,
 275                                                iconv);
 276             yaz_iconv_close(iconv);
 277         } else
 278             yaz_log(YLOG_WARN, "Query encoding failed %s %s",
 279                     cl->database->database->url, queryenc);
 280     }
 281
 282     for (ndb = 0; sdb->database->databases[ndb]; ndb++)
 283         ;
 284     databaselist = odr_malloc(global_parameters.odr_out, sizeof(char*) * ndb);
 285     for (ndb = 0; sdb->database->databases[ndb]; ndb++)
 286         databaselist[ndb] = sdb->database->databases[ndb];
 287
 288     if (!(piggyback = session_setting_oneval(sdb, PZ_PIGGYBACK)) || *piggyback == '1')
 289     {
 290         if ((recsyn = session_setting_oneval(sdb, PZ_REQUESTSYNTAX)))
 291         {
 292 #if YAZ_VERSIONL >= 0x020163
 293             a->u.searchRequest->preferredRecordSyntax =
 294                 yaz_string_to_oid_odr(yaz_oid_std(),
 295                                       CLASS_RECSYN, recsyn,
 296                                       global_parameters.odr_out);
 297 #else
 298             a->u.searchRequest->preferredRecordSyntax =
 299                 yaz_str_to_z3950oid(global_parameters.odr_out,
 300                                     CLASS_RECSYN, recsyn);
 301 #endif
 302         }
 303         a->u.searchRequest->smallSetUpperBound = &ssub;
 304         a->u.searchRequest->largeSetLowerBound = &lslb;
 305         a->u.searchRequest->mediumSetPresentNumber = &mspn;
 306     }
 307     a->u.searchRequest->resultSetName = "Default";
 308     a->u.searchRequest->databaseNames = databaselist;
 309     a->u.searchRequest->num_databaseNames = ndb;
 310
 311
 312     {  //scope for sending and logging queries
 313         WRBUF wbquery = wrbuf_alloc();
 314         yaz_query_to_wrbuf(wbquery, zquery);
 315
 316
 317         if (send_apdu(cl, a) >= 0)
 318             {
 319                 iochan_setflags(i, EVENT_INPUT);
 320                 cl->state = Client_Searching;
 321                 cl->requestid = se->requestid;
 322                 yaz_log(YLOG_LOG, "SearchRequest %s %s %s",
 323                          cl->database->database->url,
 324                         queryenc ? queryenc : "UTF-8",
 325                         wrbuf_cstr(wbquery));
 326             }
 327         else {
 328             cl->state = Client_Error;
 329                 yaz_log(YLOG_WARN, "Failed SearchRequest %s  %s %s",
 330                          cl->database->database->url,
 331                         queryenc ? queryenc : "UTF-8",
 332                         wrbuf_cstr(wbquery));
 333         }
 334
 335         wrbuf_destroy(wbquery);
 336     }
 337
 338     odr_reset(global_parameters.odr_out);
 339 }
 340
 341 static void send_present(IOCHAN i)
 342 {
 343     struct connection *co = iochan_getdata(i);
 344     struct client *cl = co->client;
 345     struct session_database *sdb = cl->database;
 346     Z_APDU *a = zget_APDU(global_parameters.odr_out, Z_APDU_presentRequest);
 347     int toget;
 348     int start = cl->records + 1;
 349     char *recsyn;
 350
 351     toget = global_parameters.chunk;
 352     if (toget > global_parameters.toget - cl->records)
 353         toget = global_parameters.toget - cl->records;
 354     if (toget > cl->hits - cl->records)
 355         toget = cl->hits - cl->records;
 356
 357     yaz_log(YLOG_DEBUG, "Trying to present %d records\n", toget);
 358
 359     a->u.presentRequest->resultSetStartPoint = &start;
 360     a->u.presentRequest->numberOfRecordsRequested = &toget;
 361
 362     a->u.presentRequest->resultSetId = "Default";
 363
 364     if ((recsyn = session_setting_oneval(sdb, PZ_REQUESTSYNTAX)))
 365     {
 366 #if YAZ_VERSIONL >= 0x020163
 367         a->u.presentRequest->preferredRecordSyntax =
 368             yaz_string_to_oid_odr(yaz_oid_std(),
 369                                   CLASS_RECSYN, recsyn,
 370                                   global_parameters.odr_out);
 371 #else
 372         a->u.presentRequest->preferredRecordSyntax =
 373             yaz_str_to_z3950oid(global_parameters.odr_out,
 374                                 CLASS_RECSYN, recsyn);
 375 #endif
 376     }
 377
 378     if (send_apdu(cl, a) >= 0)
 379     {
 380         iochan_setflags(i, EVENT_INPUT);
 381         cl->state = Client_Presenting;
 382     }
 383     else
 384         cl->state = Client_Error;
 385     odr_reset(global_parameters.odr_out);
 386 }
 387
 388 static void do_initResponse(IOCHAN i, Z_APDU *a)
 389 {
 390     struct connection *co = iochan_getdata(i);
 391     struct client *cl = co->client;
 392     Z_InitResponse *r = a->u.initResponse;
 393
 394     yaz_log(YLOG_DEBUG, "Init response %s", cl->database->database->url);
 395
 396     if (*r->result)
 397     {
 398         cl->state = Client_Idle;
 399     }
 400     else
 401         cl->state = Client_Failed; // FIXME need to do something to the connection
 402 }
 403
 404 static void do_searchResponse(IOCHAN i, Z_APDU *a)
 405 {
 406     struct connection *co = iochan_getdata(i);
 407     struct client *cl = co->client;
 408     struct session *se = cl->session;
 409     Z_SearchResponse *r = a->u.searchResponse;
 410
 411     yaz_log(YLOG_DEBUG, "Search response %s (status=%d)",
 412             cl->database->database->url, *r->searchStatus);
 413
 414     if (*r->searchStatus)
 415     {
 416         cl->hits = *r->resultCount;
 417         se->total_hits += cl->hits;
 418         if (r->presentStatus && !*r->presentStatus && r->records)
 419         {
 420             yaz_log(YLOG_DEBUG, "Records in search response %s",
 421                     cl->database->database->url);
 422             ingest_records(cl, r->records);
 423         }
 424         cl->state = Client_Idle;
 425     }
 426     else
 427     {          /*"FAILED"*/
 428         cl->hits = 0;
 429         cl->state = Client_Error;
 430         if (r->records) {
 431             Z_Records *recs = r->records;
 432             if (recs->which == Z_Records_NSD)
 433             {
 434                 yaz_log(YLOG_WARN,
 435                         "Search response: Non-surrogate diagnostic %s",
 436                         cl->database->database->url);
 437                 cl->diagnostic = *recs->u.nonSurrogateDiagnostic->condition;
 438                 cl->state = Client_Error;
 439             }
 440         }
 441     }
 442 }
 443
 444 static void do_closeResponse(IOCHAN i, Z_APDU *a)
 445 {
 446     struct connection *co = iochan_getdata(i);
 447     struct client *cl = co->client;
 448     /* Z_Close *r = a->u.close; */
 449
 450     yaz_log(YLOG_WARN, "Close response %s", cl->database->database->url);
 451
 452     cl->state = Client_Failed;
 453     connection_destroy(co);
 454 }
 455
 456
 457 char *normalize_mergekey(char *buf, int skiparticle)
 458 {
 459     char *p = buf, *pout = buf;
 460
 461     if (skiparticle)
 462     {
 463         char firstword[64];
 464         char articles[] = "the den der die des an a "; // must end in space
 465
 466         while (*p && !isalnum(*p))
 467             p++;
 468         pout = firstword;
 469         while (*p && *p != ' ' && pout - firstword < 62)
 470             *(pout++) = tolower(*(p++));
 471         *(pout++) = ' ';
 472         *(pout++) = '\0';
 473         if (!strstr(articles, firstword))
 474             p = buf;
 475         pout = buf;
 476     }
 477
 478     while (*p)
 479     {
 480         while (*p && !isalnum(*p))
 481             p++;
 482         while (isalnum(*p))
 483             *(pout++) = tolower(*(p++));
 484         if (*p)
 485             *(pout++) = ' ';
 486         while (*p && !isalnum(*p))
 487             p++;
 488     }
 489     if (buf != pout)
 490         do {
 491             *(pout--) = '\0';
 492         }
 493         while (pout > buf && *pout == ' ');
 494
 495     return buf;
 496 }
 497
 498 static void add_facet(struct session *s, const char *type, const char *value)
 499 {
 500     int i;
 501
 502     if (!*value)
 503         return;
 504     for (i = 0; i < s->num_termlists; i++)
 505         if (!strcmp(s->termlists[i].name, type))
 506             break;
 507     if (i == s->num_termlists)
 508     {
 509         if (i == SESSION_MAX_TERMLISTS)
 510         {
 511             yaz_log(YLOG_FATAL, "Too many termlists");
 512             exit(1);
 513         }
 514         s->termlists[i].name = nmem_strdup(s->nmem, type);
 515         s->termlists[i].termlist = termlist_create(s->nmem, s->expected_maxrecs, 15);
 516         s->num_termlists = i + 1;
 517     }
 518     termlist_insert(s->termlists[i].termlist, value);
 519 }
 520
 521 static xmlDoc *normalize_record(struct client *cl, Z_External *rec)
 522 {
 523     struct database_retrievalmap *m;
 524     struct session_database *sdb = cl->database;
 525     struct database *db = sdb->database;
 526     xmlNode *res;
 527     xmlDoc *rdoc;
 528
 529     // First normalize to XML
 530     if (sdb->yaz_marc)
 531     {
 532         char *buf;
 533         int len;
 534         if (rec->which != Z_External_octet)
 535         {
 536             yaz_log(YLOG_WARN, "Unexpected external branch, probably BER %s",
 537                     db->url);
 538             return 0;
 539         }
 540         buf = (char*) rec->u.octet_aligned->buf;
 541         len = rec->u.octet_aligned->len;
 542         if (yaz_marc_read_iso2709(sdb->yaz_marc, buf, len) < 0)
 543         {
 544             yaz_log(YLOG_WARN, "Failed to decode MARC %s", db->url);
 545             return 0;
 546         }
 547
 548         yaz_marc_write_using_libxml2(sdb->yaz_marc, 1);
 549         if (yaz_marc_write_xml(sdb->yaz_marc, &res,
 550                     "http://www.loc.gov/MARC21/slim", 0, 0) < 0)
 551         {
 552             yaz_log(YLOG_WARN, "Failed to encode as XML %s",
 553                     db->url);
 554             return 0;
 555         }
 556         rdoc = xmlNewDoc((xmlChar *) "1.0");
 557         xmlDocSetRootElement(rdoc, res);
 558
 559     }
 560     else
 561     {
 562         yaz_log(YLOG_FATAL,
 563                 "Unknown native_syntax in normalize_record from %s",
 564                 db->url);
 565         exit(1);
 566     }
 567
 568     if (global_parameters.dump_records){
 569         fprintf(stderr,
 570                 "Input Record (normalized) from %s\n----------------\n",
 571                 db->url);
 572 #if LIBXML_VERSION >= 20600
 573         xmlDocFormatDump(stderr, rdoc, 1);
 574 #else
 575         xmlDocDump(stderr, rdoc);
 576 #endif
 577     }
 578
 579     for (m = sdb->map; m; m = m->next){
 580         xmlDoc *new = 0;
 581
 582 #if 1
 583         {
 584             xmlNodePtr root = 0;
 585             new = xsltApplyStylesheet(m->stylesheet, rdoc, 0);
 586             root= xmlDocGetRootElement(new);
 587         if (!new || !root || !(root->children))
 588         {
 589             yaz_log(YLOG_WARN, "XSLT transformation failed from %s",
 590                     cl->database->database->url);
 591             xmlFreeDoc(new);
 592             xmlFreeDoc(rdoc);
 593             return 0;
 594         }
 595         }
 596 #endif
 597
 598 #if 0
 599         // do it another way to detect transformation errors right now
 600         // but does not seem to work either!
 601         {
 602             xsltTransformContextPtr ctxt;
 603             ctxt = xsltNewTransformContext(m->stylesheet, rdoc);
 604             new = xsltApplyStylesheetUser(m->stylesheet, rdoc, 0, 0, 0, ctxt);
 605             if ((ctxt->state == XSLT_STATE_ERROR) ||
 606                 (ctxt->state == XSLT_STATE_STOPPED)){
 607                 yaz_log(YLOG_WARN, "XSLT transformation failed from %s",
 608                         cl->database->database->url);
 609                 xmlFreeDoc(new);
 610                 xmlFreeDoc(rdoc);
 611                 return 0;
 612             }
 613         }
 614 #endif
 615
 616         xmlFreeDoc(rdoc);
 617         rdoc = new;
 618     }
 619     if (global_parameters.dump_records)
 620     {
 621         fprintf(stderr, "Record from %s\n----------------\n",
 622                 cl->database->database->url);
 623 #if LIBXML_VERSION >= 20600
 624         xmlDocFormatDump(stderr, rdoc, 1);
 625 #else
 626         xmlDocDump(stderr, rdoc);
 627 #endif
 628     }
 629     return rdoc;
 630 }
 631
 632 // Extract what appears to be years from buf, storing highest and
 633 // lowest values.
 634 static int extract_years(const char *buf, int *first, int *last)
 635 {
 636     *first = -1;
 637     *last = -1;
 638     while (*buf)
 639     {
 640         const char *e;
 641         int len;
 642
 643         while (*buf && !isdigit(*buf))
 644             buf++;
 645         len = 0;
 646         for (e = buf; *e && isdigit(*e); e++)
 647             len++;
 648         if (len == 4)
 649         {
 650             int value = atoi(buf);
 651             if (*first < 0 || value < *first)
 652                 *first = value;
 653             if (*last < 0 || value > *last)
 654                 *last = value;
 655         }
 656         buf = e;
 657     }
 658     return *first;
 659 }
 660
 661 static struct record *ingest_record(struct client *cl, Z_External *rec)
 662 {
 663     xmlDoc *xdoc = normalize_record(cl, rec);
 664     xmlNode *root, *n;
 665     struct record *res;
 666     struct record_cluster *cluster;
 667     struct session *se = cl->session;
 668     xmlChar *mergekey, *mergekey_norm;
 669     xmlChar *type = 0;
 670     xmlChar *value = 0;
 671     struct conf_service *service = global_parameters.server->service;
 672
 673     if (!xdoc)
 674         return 0;
 675
 676     root = xmlDocGetRootElement(xdoc);
 677     if (!(mergekey = xmlGetProp(root, (xmlChar *) "mergekey")))
 678     {
 679         yaz_log(YLOG_WARN, "No mergekey found in record");
 680         xmlFreeDoc(xdoc);
 681         return 0;
 682     }
 683
 684     res = nmem_malloc(se->nmem, sizeof(struct record));
 685     res->next = 0;
 686     res->client = cl;
 687     res->metadata = nmem_malloc(se->nmem,
 688             sizeof(struct record_metadata*) * service->num_metadata);
 689     memset(res->metadata, 0, sizeof(struct record_metadata*) * service->num_metadata);
 690
 691     mergekey_norm = (xmlChar *) nmem_strdup(se->nmem, (char*) mergekey);
 692     xmlFree(mergekey);
 693     normalize_mergekey((char *) mergekey_norm, 0);
 694
 695     cluster = reclist_insert(se->reclist,
 696                              global_parameters.server->service,
 697                              res, (char *) mergekey_norm,
 698                              &se->total_merged);
 699     if (global_parameters.dump_records)
 700         yaz_log(YLOG_LOG, "Cluster id %d from %s (#%d)", cluster->recid,
 701                 cl->database->database->url, cl->records);
 702     if (!cluster)
 703     {
 704         /* no room for record */
 705         xmlFreeDoc(xdoc);
 706         return 0;
 707     }
 708     relevance_newrec(se->relevance, cluster);
 709
 710     for (n = root->children; n; n = n->next)
 711     {
 712         if (type)
 713             xmlFree(type);
 714         if (value)
 715             xmlFree(value);
 716         type = value = 0;
 717
 718         if (n->type != XML_ELEMENT_NODE)
 719             continue;
 720         if (!strcmp((const char *) n->name, "metadata"))
 721         {
 722             struct conf_metadata *md = 0;
 723             struct conf_sortkey *sk = 0;
 724             struct record_metadata **wheretoput, *newm;
 725             int imeta;
 726             int first, last;
 727
 728             type = xmlGetProp(n, (xmlChar *) "type");
 729             value = xmlNodeListGetString(xdoc, n->children, 0);
 730
 731             if (!type || !value)
 732                 continue;
 733
 734             // First, find out what field we're looking at
 735             for (imeta = 0; imeta < service->num_metadata; imeta++)
 736                 if (!strcmp((const char *) type, service->metadata[imeta].name))
 737                 {
 738                     md = &service->metadata[imeta];
 739                     if (md->sortkey_offset >= 0)
 740                         sk = &service->sortkeys[md->sortkey_offset];
 741                     break;
 742                 }
 743             if (!md)
 744             {
 745                 yaz_log(YLOG_WARN, "Ignoring unknown metadata element: %s", type);
 746                 continue;
 747             }
 748
 749             // Find out where we are putting it
 750             if (md->merge == Metadata_merge_no)
 751                 wheretoput = &res->metadata[imeta];
 752             else
 753                 wheretoput = &cluster->metadata[imeta];
 754
 755             // Put it there
 756             newm = nmem_malloc(se->nmem, sizeof(struct record_metadata));
 757             newm->next = 0;
 758             if (md->type == Metadata_type_generic)
 759             {
 760                 char *p, *pe;
 761                 for (p = (char *) value; *p && isspace(*p); p++)
 762                     ;
 763                 for (pe = p + strlen(p) - 1;
 764                         pe > p && strchr(" ,/.:([", *pe); pe--)
 765                     *pe = '\0';
 766                 newm->data.text = nmem_strdup(se->nmem, p);
 767
 768             }
 769             else if (md->type == Metadata_type_year)
 770             {
 771                 if (extract_years((char *) value, &first, &last) < 0)
 772                     continue;
 773             }
 774             else
 775             {
 776                 yaz_log(YLOG_WARN, "Unknown type in metadata element %s", type);
 777                 continue;
 778             }
 779             if (md->type == Metadata_type_year && md->merge != Metadata_merge_range)
 780             {
 781                 yaz_log(YLOG_WARN, "Only range merging supported for years");
 782                 continue;
 783             }
 784             if (md->merge == Metadata_merge_unique)
 785             {
 786                 struct record_metadata *mnode;
 787                 for (mnode = *wheretoput; mnode; mnode = mnode->next)
 788                     if (!strcmp((const char *) mnode->data.text, newm->data.text))
 789                         break;
 790                 if (!mnode)
 791                 {
 792                     newm->next = *wheretoput;
 793                     *wheretoput = newm;
 794                 }
 795             }
 796             else if (md->merge == Metadata_merge_longest)
 797             {
 798                 if (!*wheretoput ||
 799                         strlen(newm->data.text) > strlen((*wheretoput)->data.text))
 800                 {
 801                     *wheretoput = newm;
 802                     if (sk)
 803                     {
 804                         char *s = nmem_strdup(se->nmem, newm->data.text);
 805                         if (!cluster->sortkeys[md->sortkey_offset])
 806                             cluster->sortkeys[md->sortkey_offset] =
 807                                 nmem_malloc(se->nmem, sizeof(union data_types));
 808                         normalize_mergekey(s,
 809                                 (sk->type == Metadata_sortkey_skiparticle));
 810                         cluster->sortkeys[md->sortkey_offset]->text = s;
 811                     }
 812                 }
 813             }
 814             else if (md->merge == Metadata_merge_all || md->merge == Metadata_merge_no)
 815             {
 816                 newm->next = *wheretoput;
 817                 *wheretoput = newm;
 818             }
 819             else if (md->merge == Metadata_merge_range)
 820             {
 821                 assert(md->type == Metadata_type_year);
 822                 if (!*wheretoput)
 823                 {
 824                     *wheretoput = newm;
 825                     (*wheretoput)->data.number.min = first;
 826                     (*wheretoput)->data.number.max = last;
 827                     if (sk)
 828                         cluster->sortkeys[md->sortkey_offset] = &newm->data;
 829                 }
 830                 else
 831                 {
 832                     if (first < (*wheretoput)->data.number.min)
 833                         (*wheretoput)->data.number.min = first;
 834                     if (last > (*wheretoput)->data.number.max)
 835                         (*wheretoput)->data.number.max = last;
 836                 }
 837 #ifdef GAGA
 838                 if (sk)
 839                 {
 840                     union data_types *sdata = cluster->sortkeys[md->sortkey_offset];
 841                     yaz_log(YLOG_LOG, "SK range: %d-%d", sdata->number.min, sdata->number.max);
 842                 }
 843 #endif
 844             }
 845             else
 846                 yaz_log(YLOG_WARN, "Don't know how to merge on element name %s", md->name);
 847
 848             if (md->rank)
 849                 relevance_countwords(se->relevance, cluster,
 850                                      (char *) value, md->rank);
 851             if (md->termlist)
 852             {
 853                 if (md->type == Metadata_type_year)
 854                 {
 855                     char year[64];
 856                     sprintf(year, "%d", last);
 857                     add_facet(se, (char *) type, year);
 858                     if (first != last)
 859                     {
 860                         sprintf(year, "%d", first);
 861                         add_facet(se, (char *) type, year);
 862                     }
 863                 }
 864                 else
 865                     add_facet(se, (char *) type, (char *) value);
 866             }
 867             xmlFree(type);
 868             xmlFree(value);
 869             type = value = 0;
 870         }
 871         else
 872             yaz_log(YLOG_WARN, "Unexpected element %s in internal record", n->name);
 873     }
 874     if (type)
 875         xmlFree(type);
 876     if (value)
 877         xmlFree(value);
 878
 879     xmlFreeDoc(xdoc);
 880
 881     relevance_donerecord(se->relevance, cluster);
 882     se->total_records++;
 883
 884     return res;
 885 }
 886
 887 // Retrieve first defined value for 'name' for given database.
 888 // Will be extended to take into account user associated with session
 889 char *session_setting_oneval(struct session_database *db, int offset)
 890 {
 891     if (!db->settings[offset])
 892         return "";
 893     return db->settings[offset]->value;
 894 }
 895
 896 static void ingest_records(struct client *cl, Z_Records *r)
 897 {
 898 #if USE_TIMING
 899     yaz_timing_t t = yaz_timing_create();
 900 #endif
 901     struct record *rec;
 902     struct session *s = cl->session;
 903     Z_NamePlusRecordList *rlist;
 904     int i;
 905
 906     if (r->which != Z_Records_DBOSD)
 907         return;
 908     rlist = r->u.databaseOrSurDiagnostics;
 909     for (i = 0; i < rlist->num_records; i++)
 910     {
 911         Z_NamePlusRecord *npr = rlist->records[i];
 912
 913         cl->records++;
 914         if (npr->which != Z_NamePlusRecord_databaseRecord)
 915         {
 916             yaz_log(YLOG_WARN,
 917                     "Unexpected record type, probably diagnostic %s",
 918                     cl->database->database->url);
 919             continue;
 920         }
 921
 922         rec = ingest_record(cl, npr->u.databaseRecord);
 923         if (!rec)
 924             continue;
 925     }
 926     if (s->watchlist[SESSION_WATCH_RECORDS].fun && rlist->num_records)
 927         session_alert_watch(s, SESSION_WATCH_RECORDS);
 928
 929 #if USE_TIMING
 930     yaz_timing_stop(t);
 931     yaz_log(YLOG_LOG, "ingest_records %6.5f %3.2f %3.2f",
 932             yaz_timing_get_real(t), yaz_timing_get_user(t),
 933             yaz_timing_get_sys(t));
 934     yaz_timing_destroy(&t);
 935 #endif
 936 }
 937
 938 static void do_presentResponse(IOCHAN i, Z_APDU *a)
 939 {
 940     struct connection *co = iochan_getdata(i);
 941     struct client *cl = co->client;
 942     Z_PresentResponse *r = a->u.presentResponse;
 943
 944     if (r->records) {
 945         Z_Records *recs = r->records;
 946         if (recs->which == Z_Records_NSD)
 947         {
 948             yaz_log(YLOG_WARN, "Non-surrogate diagnostic %s",
 949                     cl->database->database->url);
 950             cl->diagnostic = *recs->u.nonSurrogateDiagnostic->condition;
 951             cl->state = Client_Error;
 952         }
 953     }
 954
 955     if (!*r->presentStatus && cl->state != Client_Error)
 956     {
 957         yaz_log(YLOG_DEBUG, "Good Present response %s",
 958                 cl->database->database->url);
 959         ingest_records(cl, r->records);
 960         cl->state = Client_Idle;
 961     }
 962     else if (*r->presentStatus)
 963     {
 964         yaz_log(YLOG_WARN, "Bad Present response %s",
 965                 cl->database->database->url);
 966         cl->state = Client_Error;
 967     }
 968 }
 969
 970 void connection_handler(IOCHAN i, int event)
 971 {
 972     struct connection *co = iochan_getdata(i);
 973     struct client *cl = co->client;
 974     struct session *se = 0;
 975
 976     if (cl)
 977         se = cl->session;
 978     else
 979     {
 980         yaz_log(YLOG_WARN, "Destroying orphan connection");
 981         connection_destroy(co);
 982         return;
 983     }
 984
 985     if (co->state == Conn_Connecting && event & EVENT_OUTPUT)
 986     {
 987         int errcode;
 988         socklen_t errlen = sizeof(errcode);
 989
 990         if (getsockopt(cs_fileno(co->link), SOL_SOCKET, SO_ERROR, &errcode,
 991             &errlen) < 0 || errcode != 0)
 992         {
 993             client_fatal(cl);
 994             return;
 995         }
 996         else
 997         {
 998             yaz_log(YLOG_DEBUG, "Connect OK");
 999             co->state = Conn_Open;
1000             if (cl)
1001                 cl->state = Client_Connected;
1002         }
1003     }
1004
1005     else if (event & EVENT_INPUT)
1006     {
1007         int len = cs_get(co->link, &co->ibuf, &co->ibufsize);
1008
1009         if (len < 0)
1010         {
1011             yaz_log(YLOG_WARN|YLOG_ERRNO, "Error reading from %s",
1012                     cl->database->database->url);
1013             connection_destroy(co);
1014             return;
1015         }
1016         else if (len == 0)
1017         {
1018             yaz_log(YLOG_WARN, "EOF reading from %s", cl->database->database->url);
1019             connection_destroy(co);
1020             return;
1021         }
1022         else if (len > 1) // We discard input if we have no connection
1023         {
1024             co->state = Conn_Open;
1025
1026             if (cl && (cl->requestid == se->requestid || cl->state == Client_Initializing))
1027             {
1028                 Z_APDU *a;
1029
1030                 odr_reset(global_parameters.odr_in);
1031                 odr_setbuf(global_parameters.odr_in, co->ibuf, len, 0);
1032                 if (!z_APDU(global_parameters.odr_in, &a, 0, 0))
1033                 {
1034                     client_fatal(cl);
1035                     return;
1036                 }
1037                 switch (a->which)
1038                 {
1039                     case Z_APDU_initResponse:
1040                         do_initResponse(i, a);
1041                         break;
1042                     case Z_APDU_searchResponse:
1043                         do_searchResponse(i, a);
1044                         break;
1045                     case Z_APDU_presentResponse:
1046                         do_presentResponse(i, a);
1047                         break;
1048                     case Z_APDU_close:
1049                         do_closeResponse(i, a);
1050                         break;
1051                     default:
1052                         yaz_log(YLOG_WARN,
1053                                 "Unexpected Z39.50 response from %s",
1054                                 cl->database->database->url);
1055                         client_fatal(cl);
1056                         return;
1057                 }
1058                 // We aren't expecting staggered output from target
1059                 // if (cs_more(t->link))
1060                 //    iochan_setevent(i, EVENT_INPUT);
1061             }
1062             else  // we throw away response and go to idle mode
1063             {
1064                 yaz_log(YLOG_DEBUG, "Ignoring result of expired operation");
1065                 cl->state = Client_Idle;
1066             }
1067         }
1068         /* if len==1 we do nothing but wait for more input */
1069     }
1070
1071     if (cl->state == Client_Connected) {
1072         send_init(i);
1073     }
1074
1075     if (cl->state == Client_Idle)
1076     {
1077         if (cl->requestid != se->requestid && cl->pquery) {
1078             send_search(i);
1079         }
1080         else if (cl->hits > 0 && cl->records < global_parameters.toget &&
1081             cl->records < cl->hits) {
1082             send_present(i);
1083         }
1084     }
1085 }
1086
1087 // Disassociate connection from client
1088 static void connection_release(struct connection *co)
1089 {
1090     struct client *cl = co->client;
1091
1092     yaz_log(YLOG_DEBUG, "Connection release %s", co->host->hostport);
1093     if (!cl)
1094         return;
1095     cl->connection = 0;
1096     co->client = 0;
1097 }
1098
1099 // Close connection and recycle structure
1100 static void connection_destroy(struct connection *co)
1101 {
1102     struct host *h = co->host;
1103     cs_close(co->link);
1104     iochan_destroy(co->iochan);
1105
1106     yaz_log(YLOG_DEBUG, "Connection destroy %s", co->host->hostport);
1107     if (h->connections == co)
1108         h->connections = co->next;
1109     else
1110     {
1111         struct connection *pco;
1112         for (pco = h->connections; pco && pco->next != co; pco = pco->next)
1113             ;
1114         if (pco)
1115             pco->next = co->next;
1116         else
1117             abort();
1118     }
1119     if (co->client)
1120     {
1121         if (co->client->state != Client_Idle)
1122             co->client->state = Client_Disconnected;
1123         co->client->connection = 0;
1124     }
1125     co->next = connection_freelist;
1126     connection_freelist = co;
1127 }
1128
1129
1130 // Creates a new connection for client, associated with the host of
1131 // client's database
1132 static struct connection *connection_create(struct client *cl)
1133 {
1134     struct connection *new;
1135     COMSTACK link;
1136     int res;
1137     void *addr;
1138
1139
1140     if (!cl->database->database->host->ipport)
1141     {
1142         yaz_log(YLOG_WARN, "Not yet resolved: %s",
1143                 cl->database->database->url);
1144         return 0;
1145     }
1146
1147     if (!(link = cs_create(tcpip_type, 0, PROTO_Z3950)))
1148         {
1149             yaz_log(YLOG_FATAL|YLOG_ERRNO, "Failed to create comstack");
1150             exit(1);
1151         }
1152
1153     if (0 == strlen(global_parameters.zproxy_override)){
1154         /* no Z39.50 proxy needed - direct connect */
1155         yaz_log(YLOG_DEBUG, "Connection create %s", cl->database->database->url);
1156
1157         if (!(addr = cs_straddr(link, cl->database->database->host->ipport)))
1158             {
1159                 yaz_log(YLOG_WARN|YLOG_ERRNO,
1160                         "Lookup of IP address %s failed",
1161                         cl->database->database->host->ipport);
1162                 return 0;
1163             }
1164
1165     } else {
1166         /* Z39.50 proxy connect */
1167         yaz_log(YLOG_DEBUG, "Connection create %s proxy %s",
1168                 cl->database->database->url, global_parameters.zproxy_override);
1169
1170         if (!(addr = cs_straddr(link, global_parameters.zproxy_override)))
1171             {
1172                 yaz_log(YLOG_WARN|YLOG_ERRNO,
1173                         "Lookup of IP address %s failed",
1174                         global_parameters.zproxy_override);
1175                 return 0;
1176             }
1177     }
1178
1179     res = cs_connect(link, addr);
1180     if (res < 0)
1181     {
1182         yaz_log(YLOG_WARN|YLOG_ERRNO, "cs_connect %s", cl->database->database->url);
1183         return 0;
1184     }
1185
1186     if ((new = connection_freelist))
1187         connection_freelist = new->next;
1188     else
1189     {
1190         new = xmalloc(sizeof (struct connection));
1191         new->ibuf = 0;
1192         new->ibufsize = 0;
1193     }
1194     new->state = Conn_Connecting;
1195     new->host = cl->database->database->host;
1196     new->next = new->host->connections;
1197     new->host->connections = new;
1198     new->client = cl;
1199     cl->connection = new;
1200     new->link = link;
1201
1202     new->iochan = iochan_create(cs_fileno(link), connection_handler, 0);
1203     iochan_setdata(new->iochan, new);
1204     pazpar2_add_channel(new->iochan);
1205     return new;
1206 }
1207
1208 // Close connection and set state to error
1209 static void client_fatal(struct client *cl)
1210 {
1211     yaz_log(YLOG_WARN, "Fatal error from %s", cl->database->database->url);
1212     connection_destroy(cl->connection);
1213     cl->state = Client_Error;
1214 }
1215
1216 // Ensure that client has a connection associated
1217 static int client_prep_connection(struct client *cl)
1218 {
1219     struct connection *co;
1220     struct session *se = cl->session;
1221     struct host *host = cl->database->database->host;
1222
1223     co = cl->connection;
1224
1225     yaz_log(YLOG_DEBUG, "Client prep %s", cl->database->database->url);
1226
1227     if (!co)
1228     {
1229         // See if someone else has an idle connection
1230         // We should look at timestamps here to select the longest-idle connection
1231         for (co = host->connections; co; co = co->next)
1232             if (co->state == Conn_Open && (!co->client || co->client->session != se))
1233                 break;
1234         if (co)
1235         {
1236             connection_release(co);
1237             cl->connection = co;
1238             co->client = cl;
1239         }
1240         else
1241             co = connection_create(cl);
1242     }
1243     if (co)
1244     {
1245         if (co->state == Conn_Connecting)
1246         {
1247             cl->state = Client_Connecting;
1248             iochan_setflag(co->iochan, EVENT_OUTPUT);
1249         }
1250         else if (co->state == Conn_Open)
1251         {
1252             if (cl->state == Client_Error || cl->state == Client_Disconnected)
1253                 cl->state = Client_Idle;
1254             iochan_setflag(co->iochan, EVENT_OUTPUT);
1255         }
1256         return 1;
1257     }
1258     else
1259         return 0;
1260 }
1261
1262 // Initialize YAZ Map structures for MARC-based targets
1263 static int prepare_yazmarc(struct session_database *sdb)
1264 {
1265     char *s;
1266
1267     if (!sdb->settings)
1268     {
1269         yaz_log(YLOG_WARN, "No settings for %s", sdb->database->url);
1270         return -1;
1271     }
1272     if ((s = session_setting_oneval(sdb, PZ_NATIVESYNTAX)) && !strncmp(s, "iso2709", 7))
1273     {
1274         char *encoding = "marc-8s", *e;
1275         yaz_iconv_t cm;
1276
1277         // See if a native encoding is specified
1278         if ((e = strchr(s, ';')))
1279             encoding = e + 1;
1280
1281         sdb->yaz_marc = yaz_marc_create();
1282         yaz_marc_subfield_str(sdb->yaz_marc, "\t");
1283
1284         cm = yaz_iconv_open("utf-8", encoding);
1285         if (!cm)
1286         {
1287             yaz_log(YLOG_FATAL,
1288                     "Unable to map from %s to UTF-8 for target %s",
1289                     encoding, sdb->database->url);
1290             return -1;
1291         }
1292         yaz_marc_iconv(sdb->yaz_marc, cm);
1293     }
1294     return 0;
1295 }
1296
1297 // Prepare XSLT stylesheets for record normalization
1298 // Structures are allocated on the session_wide nmem to avoid having
1299 // to recompute this for every search. This would lead
1300 // to leaking if a single session was to repeatedly change the PZ_XSLT
1301 // setting. However, this is not a realistic use scenario.
1302 static int prepare_map(struct session *se, struct session_database *sdb)
1303 {
1304    char *s;
1305
1306     if (!sdb->settings)
1307     {
1308         yaz_log(YLOG_WARN, "No settings on %s", sdb->database->url);
1309         return -1;
1310     }
1311     if ((s = session_setting_oneval(sdb, PZ_XSLT)))
1312     {
1313         char **stylesheets;
1314         struct database_retrievalmap **m = &sdb->map;
1315         int num, i;
1316
1317         nmem_strsplit(se->session_nmem, ",", s, &stylesheets, &num);
1318         for (i = 0; i < num; i++)
1319         {
1320             (*m) = nmem_malloc(se->session_nmem, sizeof(**m));
1321             (*m)->next = 0;
1322             if (!((*m)->stylesheet = conf_load_stylesheet(stylesheets[i])))
1323             {
1324                 yaz_log(YLOG_FATAL, "Unable to load stylesheet: %s",
1325                         stylesheets[i]);
1326                 return -1;
1327             }
1328             m = &(*m)->next;
1329         }
1330     }
1331     if (!sdb->map)
1332         yaz_log(YLOG_WARN, "No Normalization stylesheet for target %s",
1333                 sdb->database->url);
1334     return 0;
1335 }
1336
1337 // This analyzes settings and recomputes any supporting data structures
1338 // if necessary.
1339 static int prepare_session_database(struct session *se, struct session_database *sdb)
1340 {
1341     if (!sdb->settings)
1342     {
1343         yaz_log(YLOG_WARN, "No settings associated with %s", sdb->database->url);
1344         return -1;
1345     }
1346     if (sdb->settings[PZ_NATIVESYNTAX] && !sdb->yaz_marc)
1347     {
1348         if (prepare_yazmarc(sdb) < 0)
1349             return -1;
1350     }
1351     if (sdb->settings[PZ_XSLT] && !sdb->map)
1352     {
1353         if (prepare_map(se, sdb) < 0)
1354             return -1;
1355     }
1356     return 0;
1357 }
1358
1359 // Initialize CCL map for a target
1360 static CCL_bibset prepare_cclmap(struct client *cl)
1361 {
1362     struct session_database *sdb = cl->database;
1363     struct setting *s;
1364     CCL_bibset res;
1365
1366     if (!sdb->settings)
1367         return 0;
1368     res = ccl_qual_mk();
1369     for (s = sdb->settings[PZ_CCLMAP]; s; s = s->next)
1370     {
1371         char *p = strchr(s->name + 3, ':');
1372         if (!p)
1373         {
1374             yaz_log(YLOG_WARN, "Malformed cclmap name: %s", s->name);
1375             ccl_qual_rm(&res);
1376             return 0;
1377         }
1378         p++;
1379         ccl_qual_fitem(res, s->value, p);
1380     }
1381     return res;
1382 }
1383
1384 // Parse the query given the settings specific to this client
1385 static int client_parse_query(struct client *cl, const char *query)
1386 {
1387     struct session *se = cl->session;
1388     struct ccl_rpn_node *cn;
1389     int cerror, cpos;
1390     CCL_bibset ccl_map = prepare_cclmap(cl);
1391
1392     if (!ccl_map)
1393         return -1;
1394     cn = ccl_find_str(ccl_map, query, &cerror, &cpos);
1395     ccl_qual_rm(&ccl_map);
1396     if (!cn)
1397     {
1398         cl->state = Client_Error;
1399         yaz_log(YLOG_WARN, "Failed to parse query for %s",
1400                          cl->database->database->url);
1401         return -1;
1402     }
1403     wrbuf_rewind(se->wrbuf);
1404     ccl_pquery(se->wrbuf, cn);
1405     wrbuf_putc(se->wrbuf, '\0');
1406     if (cl->pquery)
1407         xfree(cl->pquery);
1408     cl->pquery = xstrdup(wrbuf_buf(se->wrbuf));
1409
1410     if (!se->relevance)
1411     {
1412         // Initialize relevance structure with query terms
1413         char *p[512];
1414         extract_terms(se->nmem, cn, p);
1415         se->relevance = relevance_create(se->nmem, (const char **) p,
1416                 se->expected_maxrecs);
1417     }
1418
1419     ccl_rpn_delete(cn);
1420     return 0;
1421 }
1422
1423 static struct client *client_create(void)
1424 {
1425     struct client *r;
1426     if (client_freelist)
1427     {
1428         r = client_freelist;
1429         client_freelist = client_freelist->next;
1430     }
1431     else
1432         r = xmalloc(sizeof(struct client));
1433     r->pquery = 0;
1434     r->database = 0;
1435     r->connection = 0;
1436     r->session = 0;
1437     r->hits = 0;
1438     r->records = 0;
1439     r->setno = 0;
1440     r->requestid = -1;
1441     r->diagnostic = 0;
1442     r->state = Client_Disconnected;
1443     r->next = 0;
1444     return r;
1445 }
1446
1447 void client_destroy(struct client *c)
1448 {
1449     struct session *se = c->session;
1450     if (c == se->clients)
1451         se->clients = c->next;
1452     else
1453     {
1454         struct client *cc;
1455         for (cc = se->clients; cc && cc->next != c; cc = cc->next)
1456             ;
1457         if (cc)
1458             cc->next = c->next;
1459     }
1460     if (c->connection)
1461         connection_release(c->connection);
1462     c->next = client_freelist;
1463     client_freelist = c;
1464 }
1465
1466 void session_set_watch(struct session *s, int what, session_watchfun fun, void *data)
1467 {
1468     s->watchlist[what].fun = fun;
1469     s->watchlist[what].data = data;
1470 }
1471
1472 void session_alert_watch(struct session *s, int what)
1473 {
1474     if (!s->watchlist[what].fun)
1475         return;
1476     (*s->watchlist[what].fun)(s->watchlist[what].data);
1477     s->watchlist[what].fun = 0;
1478     s->watchlist[what].data = 0;
1479 }
1480
1481 //callback for grep_databases
1482 static void select_targets_callback(void *context, struct session_database *db)
1483 {
1484     struct session *se = (struct session*) context;
1485     struct client *cl = client_create();
1486     cl->database = db;
1487     cl->session = se;
1488     cl->next = se->clients;
1489     se->clients = cl;
1490 }
1491
1492 // Associates a set of clients with a session;
1493 // Note: Session-databases represent databases with per-session setting overrides
1494 int select_targets(struct session *se, struct database_criterion *crit)
1495 {
1496     while (se->clients)
1497         client_destroy(se->clients);
1498
1499     return session_grep_databases(se, crit, select_targets_callback);
1500 }
1501
1502 int session_active_clients(struct session *s)
1503 {
1504     struct client *c;
1505     int res = 0;
1506
1507     for (c = s->clients; c; c = c->next)
1508         if (c->connection && (c->state == Client_Connecting ||
1509                     c->state == Client_Initializing ||
1510                     c->state == Client_Searching ||
1511                     c->state == Client_Presenting))
1512             res++;
1513
1514     return res;
1515 }
1516
1517 // parses crit1=val1,crit2=val2|val3,...
1518 static struct database_criterion *parse_filter(NMEM m, const char *buf)
1519 {
1520     struct database_criterion *res = 0;
1521     char **values;
1522     int num;
1523     int i;
1524
1525     if (!buf || !*buf)
1526         return 0;
1527     nmem_strsplit(m, ",", buf,  &values, &num);
1528     for (i = 0; i < num; i++)
1529     {
1530         char **subvalues;
1531         int subnum;
1532         int subi;
1533         struct database_criterion *new = nmem_malloc(m, sizeof(*new));
1534         char *eq = strchr(values[i], '=');
1535         if (!eq)
1536         {
1537             yaz_log(YLOG_WARN, "Missing equal-sign in filter");
1538             return 0;
1539         }
1540         *(eq++) = '\0';
1541         new->name = values[i];
1542         nmem_strsplit(m, "|", eq, &subvalues, &subnum);
1543         new->values = 0;
1544         for (subi = 0; subi < subnum; subi++)
1545         {
1546             struct database_criterion_value *newv = nmem_malloc(m, sizeof(*newv));
1547             newv->value = subvalues[subi];
1548             newv->next = new->values;
1549             new->values = newv;
1550         }
1551         new->next = res;
1552         res = new;
1553     }
1554     return res;
1555 }
1556
1557 char *search(struct session *se, char *query, char *filter)
1558 {
1559     int live_channels = 0;
1560     struct client *cl;
1561     struct database_criterion *criteria;
1562
1563     yaz_log(YLOG_DEBUG, "Search");
1564
1565     nmem_reset(se->nmem);
1566     criteria = parse_filter(se->nmem, filter);
1567     se->requestid++;
1568     live_channels = select_targets(se, criteria);
1569     if (live_channels)
1570     {
1571         int maxrecs = live_channels * global_parameters.toget;
1572         se->num_termlists = 0;
1573         se->reclist = reclist_create(se->nmem, maxrecs);
1574         // This will be initialized in send_search()
1575         se->total_records = se->total_hits = se->total_merged = 0;
1576         se->expected_maxrecs = maxrecs;
1577     }
1578     else
1579         return "NOTARGETS";
1580
1581     se->relevance = 0;
1582
1583     for (cl = se->clients; cl; cl = cl->next)
1584     {
1585         if (prepare_session_database(se, cl->database) < 0)
1586             return "CONFIG_ERROR";
1587         if (client_parse_query(cl, query) < 0)  // Query must parse for all targets
1588             return "QUERY";
1589     }
1590
1591     for (cl = se->clients; cl; cl = cl->next)
1592     {
1593         client_prep_connection(cl);
1594     }
1595
1596     return 0;
1597 }
1598
1599 // Apply a session override to a database
1600 void session_apply_setting(struct session *se, char *dbname, char *setting, char *value)
1601 {
1602     struct session_database *sdb;
1603
1604     for (sdb = se->databases; sdb; sdb = sdb->next)
1605         if (!strcmp(dbname, sdb->database->url))
1606         {
1607             struct setting *new = nmem_malloc(se->session_nmem, sizeof(*new));
1608             int offset = settings_offset(setting);
1609
1610             if (offset < 0)
1611             {
1612                 yaz_log(YLOG_WARN, "Unknown setting %s", setting);
1613                 return;
1614             }
1615             new->precedence = 0;
1616             new->target = dbname;
1617             new->name = setting;
1618             new->value = value;
1619             new->next = sdb->settings[offset];
1620             sdb->settings[offset] = new;
1621
1622             // Force later recompute of settings-driven data structures
1623             // (happens when a search starts and client connections are prepared)
1624             switch (offset)
1625             {
1626                 case PZ_NATIVESYNTAX:
1627                     if (sdb->yaz_marc)
1628                     {
1629                         yaz_marc_destroy(sdb->yaz_marc);
1630                         sdb->yaz_marc = 0;
1631                     }
1632                     break;
1633                 case PZ_XSLT:
1634                     if (sdb->map)
1635                     {
1636                         struct database_retrievalmap *m;
1637                         // We don't worry about the map structure -- it's in nmem
1638                         for (m = sdb->map; m; m = m->next)
1639                             xsltFreeStylesheet(m->stylesheet);
1640                         sdb->map = 0;
1641                     }
1642                     break;
1643             }
1644             break;
1645         }
1646     if (!sdb)
1647         yaz_log(YLOG_WARN, "Unknown database in setting override: %s", dbname);
1648 }
1649
1650 void session_init_databases_fun(void *context, struct database *db)
1651 {
1652     struct session *se = (struct session *) context;
1653     struct session_database *new = nmem_malloc(se->session_nmem, sizeof(*new));
1654     int num = settings_num();
1655     int i;
1656
1657     new->database = db;
1658     new->yaz_marc = 0;
1659     new->map = 0;
1660     new->settings = nmem_malloc(se->session_nmem, sizeof(struct settings *) * num);
1661     for (i = 0; i < num; i++)
1662         new->settings[i] = db->settings[i];
1663     new->next = se->databases;
1664     se->databases = new;
1665 }
1666
1667 // Initialize session_database list -- this represents this session's view
1668 // of the database list -- subject to modification by the settings ws command
1669 void session_init_databases(struct session *se)
1670 {
1671     se->databases = 0;
1672     grep_databases(se, 0, session_init_databases_fun);
1673 }
1674
1675 void destroy_session(struct session *s)
1676 {
1677     yaz_log(YLOG_LOG, "Destroying session");
1678     while (s->clients)
1679         client_destroy(s->clients);
1680     nmem_destroy(s->nmem);
1681     wrbuf_destroy(s->wrbuf);
1682 }
1683
1684 struct session *new_session(NMEM nmem)
1685 {
1686     int i;
1687     struct session *session = nmem_malloc(nmem, sizeof(*session));
1688
1689     yaz_log(YLOG_DEBUG, "New Pazpar2 session");
1690
1691     session->total_hits = 0;
1692     session->total_records = 0;
1693     session->num_termlists = 0;
1694     session->reclist = 0;
1695     session->requestid = -1;
1696     session->clients = 0;
1697     session->expected_maxrecs = 0;
1698     session->session_nmem = nmem;
1699     session->nmem = nmem_create();
1700     session->wrbuf = wrbuf_alloc();
1701     session_init_databases(session);
1702     for (i = 0; i <= SESSION_WATCH_MAX; i++)
1703     {
1704         session->watchlist[i].data = 0;
1705         session->watchlist[i].fun = 0;
1706     }
1707
1708     return session;
1709 }
1710
1711 struct hitsbytarget *hitsbytarget(struct session *se, int *count)
1712 {
1713     static struct hitsbytarget res[1000]; // FIXME MM
1714     struct client *cl;
1715
1716     *count = 0;
1717     for (cl = se->clients; cl; cl = cl->next)
1718     {
1719         char *name = session_setting_oneval(cl->database, PZ_NAME);
1720
1721         res[*count].id = cl->database->database->url;
1722         res[*count].name = *name ? name : "Unknown";
1723         res[*count].hits = cl->hits;
1724         res[*count].records = cl->records;
1725         res[*count].diagnostic = cl->diagnostic;
1726         res[*count].state = client_states[cl->state];
1727         res[*count].connected  = cl->connection ? 1 : 0;
1728         (*count)++;
1729     }
1730
1731     return res;
1732 }
1733
1734 struct termlist_score **termlist(struct session *s, const char *name, int *num)
1735 {
1736     int i;
1737
1738     for (i = 0; i < s->num_termlists; i++)
1739         if (!strcmp((const char *) s->termlists[i].name, name))
1740             return termlist_highscore(s->termlists[i].termlist, num);
1741     return 0;
1742 }
1743
1744 #ifdef MISSING_HEADERS
1745 void report_nmem_stats(void)
1746 {
1747     size_t in_use, is_free;
1748
1749     nmem_get_memory_in_use(&in_use);
1750     nmem_get_memory_free(&is_free);
1751
1752     yaz_log(YLOG_LOG, "nmem stat: use=%ld free=%ld",
1753             (long) in_use, (long) is_free);
1754 }
1755 #endif
1756
1757 struct record_cluster *show_single(struct session *s, int id)
1758 {
1759     struct record_cluster *r;
1760
1761     reclist_rewind(s->reclist);
1762     while ((r = reclist_read_record(s->reclist)))
1763         if (r->recid == id)
1764             return r;
1765     return 0;
1766 }
1767
1768 struct record_cluster **show(struct session *s, struct reclist_sortparms *sp, int start,
1769         int *num, int *total, int *sumhits, NMEM nmem_show)
1770 {
1771     struct record_cluster **recs = nmem_malloc(nmem_show, *num
1772                                        * sizeof(struct record_cluster *));
1773     struct reclist_sortparms *spp;
1774     int i;
1775 #if USE_TIMING
1776     yaz_timing_t t = yaz_timing_create();
1777 #endif
1778
1779     for (spp = sp; spp; spp = spp->next)
1780         if (spp->type == Metadata_sortkey_relevance)
1781         {
1782             relevance_prepare_read(s->relevance, s->reclist);
1783             break;
1784         }
1785     reclist_sort(s->reclist, sp);
1786
1787     *total = s->reclist->num_records;
1788     *sumhits = s->total_hits;
1789
1790     for (i = 0; i < start; i++)
1791         if (!reclist_read_record(s->reclist))
1792         {
1793             *num = 0;
1794             recs = 0;
1795             break;
1796         }
1797
1798     for (i = 0; i < *num; i++)
1799     {
1800         struct record_cluster *r = reclist_read_record(s->reclist);
1801         if (!r)
1802         {
1803             *num = i;
1804             break;
1805         }
1806         recs[i] = r;
1807     }
1808 #if USE_TIMING
1809     yaz_timing_stop(t);
1810     yaz_log(YLOG_LOG, "show %6.5f %3.2f %3.2f",
1811             yaz_timing_get_real(t), yaz_timing_get_user(t),
1812             yaz_timing_get_sys(t));
1813     yaz_timing_destroy(&t);
1814 #endif
1815     return recs;
1816 }
1817
1818 void statistics(struct session *se, struct statistics *stat)
1819 {
1820     struct client *cl;
1821     int count = 0;
1822
1823     memset(stat, 0, sizeof(*stat));
1824     for (cl = se->clients; cl; cl = cl->next)
1825     {
1826         if (!cl->connection)
1827             stat->num_no_connection++;
1828         switch (cl->state)
1829         {
1830             case Client_Connecting: stat->num_connecting++; break;
1831             case Client_Initializing: stat->num_initializing++; break;
1832             case Client_Searching: stat->num_searching++; break;
1833             case Client_Presenting: stat->num_presenting++; break;
1834             case Client_Idle: stat->num_idle++; break;
1835             case Client_Failed: stat->num_failed++; break;
1836             case Client_Error: stat->num_error++; break;
1837             default: break;
1838         }
1839         count++;
1840     }
1841     stat->num_hits = se->total_hits;
1842     stat->num_records = se->total_records;
1843
1844     stat->num_clients = count;
1845 }
1846
1847 void start_http_listener(void)
1848 {
1849     char hp[128] = "";
1850     struct conf_server *ser = global_parameters.server;
1851
1852     if (*global_parameters.listener_override)
1853         strcpy(hp, global_parameters.listener_override);
1854     else
1855     {
1856         strcpy(hp, ser->host ? ser->host : "");
1857         if (ser->port)
1858         {
1859             if (*hp)
1860                 strcat(hp, ":");
1861             sprintf(hp + strlen(hp), "%d", ser->port);
1862         }
1863     }
1864     http_init(hp);
1865 }
1866
1867 void start_proxy(void)
1868 {
1869     char hp[128] = "";
1870     struct conf_server *ser = global_parameters.server;
1871
1872     if (*global_parameters.proxy_override)
1873         strcpy(hp, global_parameters.proxy_override);
1874     else if (ser->proxy_host || ser->proxy_port)
1875     {
1876         strcpy(hp, ser->proxy_host ? ser->proxy_host : "");
1877         if (ser->proxy_port)
1878         {
1879             if (*hp)
1880                 strcat(hp, ":");
1881             sprintf(hp + strlen(hp), "%d", ser->proxy_port);
1882         }
1883     }
1884     else
1885         return;
1886
1887     http_set_proxyaddr(hp, ser->myurl ? ser->myurl : "");
1888 }
1889
1890 void start_zproxy(void)
1891 {
1892     struct conf_server *ser = global_parameters.server;
1893
1894     if (*global_parameters.zproxy_override){
1895         yaz_log(YLOG_LOG, "Z39.50 proxy  %s",
1896                 global_parameters.zproxy_override);
1897         return;
1898     }
1899
1900     else if (ser->zproxy_host || ser->zproxy_port)
1901     {
1902         char hp[128] = "";
1903
1904         strcpy(hp, ser->zproxy_host ? ser->zproxy_host : "");
1905         if (ser->zproxy_port)
1906         {
1907             if (*hp)
1908                 strcat(hp, ":");
1909             else
1910                 strcat(hp, "@:");
1911
1912             sprintf(hp + strlen(hp), "%d", ser->zproxy_port);
1913         }
1914         strcpy(global_parameters.zproxy_override, hp);
1915         yaz_log(YLOG_LOG, "Z39.50 proxy  %s",
1916                 global_parameters.zproxy_override);
1917
1918     }
1919     else
1920         return;
1921 }
1922
1923 // Master list of connections we're handling events to
1924 static IOCHAN channel_list = 0;
1925 void pazpar2_add_channel(IOCHAN chan)
1926 {
1927     chan->next = channel_list;
1928     channel_list = chan;
1929 }
1930
1931 void pazpar2_event_loop()
1932 {
1933     event_loop(&channel_list);
1934 }
1935
1936 /*
1937  * Local variables:
1938  * c-basic-offset: 4
1939  * indent-tabs-mode: nil
1940  * End:
1941  * vim: shiftwidth=4 tabstop=8 expandtab
1942  */