src/pazpar2.c

   1 /* $Id: pazpar2.c,v 1.62 2007-04-04 21:05:37 marc Exp $ */
   2
   3 #include <stdlib.h>
   4 #include <stdio.h>
   5 #include <string.h>
   6 #include <sys/time.h>
   7 #include <unistd.h>
   8 #include <sys/socket.h>
   9 #include <netdb.h>
  10 #include <signal.h>
  11 #include <ctype.h>
  12 #include <assert.h>
  13
  14 #include <yaz/marcdisp.h>
  15 #include <yaz/comstack.h>
  16 #include <yaz/tcpip.h>
  17 #include <yaz/proto.h>
  18 #include <yaz/readconf.h>
  19 #include <yaz/pquery.h>
  20 #include <yaz/otherinfo.h>
  21 #include <yaz/yaz-util.h>
  22 #include <yaz/nmem.h>
  23
  24 #if HAVE_CONFIG_H
  25 #include "cconfig.h"
  26 #endif
  27
  28 #define USE_TIMING 0
  29 #if USE_TIMING
  30 #include <yaz/timing.h>
  31 #endif
  32
  33 #include <netinet/in.h>
  34
  35 #include "pazpar2.h"
  36 #include "eventl.h"
  37 #include "http.h"
  38 #include "termlists.h"
  39 #include "reclists.h"
  40 #include "relevance.h"
  41 #include "config.h"
  42 #include "database.h"
  43 #include "settings.h"
  44
  45 #define MAX_CHUNK 15
  46
  47 static void client_fatal(struct client *cl);
  48 static void connection_destroy(struct connection *co);
  49 static int client_prep_connection(struct client *cl);
  50 static void ingest_records(struct client *cl, Z_Records *r);
  51 //static struct conf_retrievalprofile *database_retrieval_profile(struct database *db);
  52 void session_alert_watch(struct session *s, int what);
  53 char *session_setting_oneval(struct session *s, struct database *db, const char *name);
  54
  55 IOCHAN channel_list = 0;  // Master list of connections we're handling events to
  56
  57 static struct connection *connection_freelist = 0;
  58 static struct client *client_freelist = 0;
  59
  60 static char *client_states[] = {
  61     "Client_Connecting",
  62     "Client_Connected",
  63     "Client_Idle",
  64     "Client_Initializing",
  65     "Client_Searching",
  66     "Client_Presenting",
  67     "Client_Error",
  68     "Client_Failed",
  69     "Client_Disconnected",
  70     "Client_Stopped"
  71 };
  72
  73 // Note: Some things in this structure will eventually move to configuration
  74 struct parameters global_parameters =
  75 {
  76     "",
  77     "",
  78     "",
  79     "",
  80     0,
  81     0,
  82     30,
  83     "81",
  84     "Index Data PazPar2 (MasterKey)",
  85     VERSION,
  86     600, // 10 minutes
  87     60,
  88     100,
  89     MAX_CHUNK,
  90     0,
  91     0,
  92     0
  93 };
  94
  95 static int send_apdu(struct client *c, Z_APDU *a)
  96 {
  97     struct connection *co = c->connection;
  98     char *buf;
  99     int len, r;
 100
 101     if (!z_APDU(global_parameters.odr_out, &a, 0, 0))
 102     {
 103         odr_perror(global_parameters.odr_out, "Encoding APDU");
 104         abort();
 105     }
 106     buf = odr_getbuf(global_parameters.odr_out, &len, 0);
 107     r = cs_put(co->link, buf, len);
 108     if (r < 0)
 109     {
 110         yaz_log(YLOG_WARN, "cs_put: %s", cs_errmsg(cs_errno(co->link)));
 111         return -1;
 112     }
 113     else if (r == 1)
 114     {
 115         fprintf(stderr, "cs_put incomplete (ParaZ does not handle that)\n");
 116         exit(1);
 117     }
 118     odr_reset(global_parameters.odr_out); /* release the APDU structure  */
 119     co->state = Conn_Waiting;
 120     return 0;
 121 }
 122
 123
 124 static void send_init(IOCHAN i)
 125 {
 126
 127     struct connection *co = iochan_getdata(i);
 128     struct client *cl = co->client;
 129     Z_APDU *a = zget_APDU(global_parameters.odr_out, Z_APDU_initRequest);
 130
 131     a->u.initRequest->implementationId = global_parameters.implementationId;
 132     a->u.initRequest->implementationName = global_parameters.implementationName;
 133     a->u.initRequest->implementationVersion =
 134         global_parameters.implementationVersion;
 135     ODR_MASK_SET(a->u.initRequest->options, Z_Options_search);
 136     ODR_MASK_SET(a->u.initRequest->options, Z_Options_present);
 137     ODR_MASK_SET(a->u.initRequest->options, Z_Options_namedResultSets);
 138
 139     ODR_MASK_SET(a->u.initRequest->protocolVersion, Z_ProtocolVersion_1);
 140     ODR_MASK_SET(a->u.initRequest->protocolVersion, Z_ProtocolVersion_2);
 141     ODR_MASK_SET(a->u.initRequest->protocolVersion, Z_ProtocolVersion_3);
 142
 143
 144     /* add virtual host if tunneling through Z39.50 proxy */
 145
 146     if (0 < strlen(global_parameters.zproxy_override)
 147         && 0 < strlen(cl->database->url))
 148         yaz_oi_set_string_oidval(&a->u.initRequest->otherInfo,
 149                                  global_parameters.odr_out, VAL_PROXY,
 150                                  1, cl->database->url);
 151
 152
 153
 154     if (send_apdu(cl, a) >= 0)
 155     {
 156         iochan_setflags(i, EVENT_INPUT);
 157         cl->state = Client_Initializing;
 158     }
 159     else
 160         cl->state = Client_Error;
 161     odr_reset(global_parameters.odr_out);
 162 }
 163
 164 static void pull_terms(NMEM nmem, struct ccl_rpn_node *n, char **termlist, int *num)
 165 {
 166     switch (n->kind)
 167     {
 168         case CCL_RPN_AND:
 169         case CCL_RPN_OR:
 170         case CCL_RPN_NOT:
 171         case CCL_RPN_PROX:
 172             pull_terms(nmem, n->u.p[0], termlist, num);
 173             pull_terms(nmem, n->u.p[1], termlist, num);
 174             break;
 175         case CCL_RPN_TERM:
 176             termlist[(*num)++] = nmem_strdup(nmem, n->u.t.term);
 177             break;
 178         default: // NOOP
 179             break;
 180     }
 181 }
 182
 183 // Extract terms from query into null-terminated termlist
 184 static void extract_terms(NMEM nmem, struct ccl_rpn_node *query, char **termlist)
 185 {
 186     int num = 0;
 187
 188     pull_terms(nmem, query, termlist, &num);
 189     termlist[num] = 0;
 190 }
 191
 192 static void send_search(IOCHAN i)
 193 {
 194     struct connection *co = iochan_getdata(i);
 195     struct client *cl = co->client;
 196     struct session *se = cl->session;
 197     struct database *db = cl->database;
 198     Z_APDU *a = zget_APDU(global_parameters.odr_out, Z_APDU_searchRequest);
 199     int ndb, cerror, cpos;
 200     char **databaselist;
 201     Z_Query *zquery;
 202     struct ccl_rpn_node *cn;
 203     int ssub = 0, lslb = 100000, mspn = 10;
 204     char *recsyn;
 205     char *piggyback;
 206
 207     yaz_log(YLOG_DEBUG, "Sending search");
 208
 209     cn = ccl_find_str(db->ccl_map, se->query, &cerror, &cpos);
 210     if (!cn)
 211         return;
 212
 213     if (!se->relevance)
 214     {
 215         // Initialize relevance structure with query terms
 216         char *p[512];
 217         extract_terms(se->nmem, cn, p);
 218         se->relevance = relevance_create(se->nmem, (const char **) p,
 219                 se->expected_maxrecs);
 220     }
 221
 222     a->u.searchRequest->query = zquery = odr_malloc(global_parameters.odr_out,
 223             sizeof(Z_Query));
 224     zquery->which = Z_Query_type_1;
 225     zquery->u.type_1 = ccl_rpn_query(global_parameters.odr_out, cn);
 226     ccl_rpn_delete(cn);
 227
 228     for (ndb = 0; db->databases[ndb]; ndb++)
 229         ;
 230     databaselist = odr_malloc(global_parameters.odr_out, sizeof(char*) * ndb);
 231     for (ndb = 0; db->databases[ndb]; ndb++)
 232         databaselist[ndb] = db->databases[ndb];
 233
 234     if (!(piggyback = session_setting_oneval(se, db, "pz:piggyback")) || *piggyback == '1')
 235     {
 236         if ((recsyn = session_setting_oneval(se, db, "pz:syntax")))
 237             a->u.searchRequest->preferredRecordSyntax =
 238                     yaz_str_to_z3950oid(global_parameters.odr_out,
 239                     CLASS_RECSYN, recsyn);
 240         a->u.searchRequest->smallSetUpperBound = &ssub;
 241         a->u.searchRequest->largeSetLowerBound = &lslb;
 242         a->u.searchRequest->mediumSetPresentNumber = &mspn;
 243     }
 244     a->u.searchRequest->resultSetName = "Default";
 245     a->u.searchRequest->databaseNames = databaselist;
 246     a->u.searchRequest->num_databaseNames = ndb;
 247
 248     if (send_apdu(cl, a) >= 0)
 249     {
 250         iochan_setflags(i, EVENT_INPUT);
 251         cl->state = Client_Searching;
 252         cl->requestid = se->requestid;
 253     }
 254     else
 255         cl->state = Client_Error;
 256
 257     odr_reset(global_parameters.odr_out);
 258 }
 259
 260 static void send_present(IOCHAN i)
 261 {
 262     struct connection *co = iochan_getdata(i);
 263     struct client *cl = co->client;
 264     struct session *se = cl->session;
 265     struct database *db = cl->database;
 266     Z_APDU *a = zget_APDU(global_parameters.odr_out, Z_APDU_presentRequest);
 267     int toget;
 268     int start = cl->records + 1;
 269     char *recsyn;
 270
 271     toget = global_parameters.chunk;
 272     if (toget > global_parameters.toget - cl->records)
 273         toget = global_parameters.toget - cl->records;
 274     if (toget > cl->hits - cl->records)
 275         toget = cl->hits - cl->records;
 276
 277     yaz_log(YLOG_DEBUG, "Trying to present %d records\n", toget);
 278
 279     a->u.presentRequest->resultSetStartPoint = &start;
 280     a->u.presentRequest->numberOfRecordsRequested = &toget;
 281
 282     a->u.presentRequest->resultSetId = "Default";
 283
 284     if ((recsyn = session_setting_oneval(se, db, "pz:syntax")))
 285         a->u.presentRequest->preferredRecordSyntax =
 286                 yaz_str_to_z3950oid(global_parameters.odr_out,
 287                 CLASS_RECSYN, recsyn);
 288
 289     if (send_apdu(cl, a) >= 0)
 290     {
 291         iochan_setflags(i, EVENT_INPUT);
 292         cl->state = Client_Presenting;
 293     }
 294     else
 295         cl->state = Client_Error;
 296     odr_reset(global_parameters.odr_out);
 297 }
 298
 299 static void do_initResponse(IOCHAN i, Z_APDU *a)
 300 {
 301     struct connection *co = iochan_getdata(i);
 302     struct client *cl = co->client;
 303     Z_InitResponse *r = a->u.initResponse;
 304
 305     yaz_log(YLOG_DEBUG, "Init response %s", cl->database->url);
 306
 307     if (*r->result)
 308     {
 309         cl->state = Client_Idle;
 310     }
 311     else
 312         cl->state = Client_Failed; // FIXME need to do something to the connection
 313 }
 314
 315 static void do_searchResponse(IOCHAN i, Z_APDU *a)
 316 {
 317     struct connection *co = iochan_getdata(i);
 318     struct client *cl = co->client;
 319     struct session *se = cl->session;
 320     Z_SearchResponse *r = a->u.searchResponse;
 321
 322     yaz_log(YLOG_DEBUG, "Search response %s (status=%d)",
 323             cl->database->url, *r->searchStatus);
 324
 325     if (*r->searchStatus)
 326     {
 327         cl->hits = *r->resultCount;
 328         se->total_hits += cl->hits;
 329         if (r->presentStatus && !*r->presentStatus && r->records)
 330         {
 331             yaz_log(YLOG_DEBUG, "Records in search response %s",
 332                     cl->database->url);
 333             ingest_records(cl, r->records);
 334         }
 335         cl->state = Client_Idle;
 336     }
 337     else
 338     {          /*"FAILED"*/
 339         cl->hits = 0;
 340         cl->state = Client_Error;
 341         if (r->records) {
 342             Z_Records *recs = r->records;
 343             if (recs->which == Z_Records_NSD)
 344             {
 345                 yaz_log(YLOG_WARN,
 346                         "Search response: Non-surrogate diagnostic %s",
 347                         cl->database->url);
 348                 cl->diagnostic = *recs->u.nonSurrogateDiagnostic->condition;
 349                 cl->state = Client_Error;
 350             }
 351         }
 352     }
 353 }
 354
 355 static void do_closeResponse(IOCHAN i, Z_APDU *a)
 356 {
 357     struct connection *co = iochan_getdata(i);
 358     struct client *cl = co->client;
 359     /* Z_Close *r = a->u.close; */
 360
 361     yaz_log(YLOG_WARN, "Close response %s", cl->database->url);
 362
 363     cl->state = Client_Failed;
 364     connection_destroy(co);
 365 }
 366
 367
 368 char *normalize_mergekey(char *buf, int skiparticle)
 369 {
 370     char *p = buf, *pout = buf;
 371
 372     if (skiparticle)
 373     {
 374         char firstword[64];
 375         char articles[] = "the den der die des an a "; // must end in space
 376
 377         while (*p && !isalnum(*p))
 378             p++;
 379         pout = firstword;
 380         while (*p && *p != ' ' && pout - firstword < 62)
 381             *(pout++) = tolower(*(p++));
 382         *(pout++) = ' ';
 383         *(pout++) = '\0';
 384         if (!strstr(articles, firstword))
 385             p = buf;
 386         pout = buf;
 387     }
 388
 389     while (*p)
 390     {
 391         while (*p && !isalnum(*p))
 392             p++;
 393         while (isalnum(*p))
 394             *(pout++) = tolower(*(p++));
 395         if (*p)
 396             *(pout++) = ' ';
 397         while (*p && !isalnum(*p))
 398             p++;
 399     }
 400     if (buf != pout)
 401         do {
 402             *(pout--) = '\0';
 403         }
 404         while (pout > buf && *pout == ' ');
 405
 406     return buf;
 407 }
 408
 409 static void add_facet(struct session *s, const char *type, const char *value)
 410 {
 411     int i;
 412
 413     if (!*value)
 414         return;
 415     for (i = 0; i < s->num_termlists; i++)
 416         if (!strcmp(s->termlists[i].name, type))
 417             break;
 418     if (i == s->num_termlists)
 419     {
 420         if (i == SESSION_MAX_TERMLISTS)
 421         {
 422             yaz_log(YLOG_FATAL, "Too many termlists");
 423             exit(1);
 424         }
 425         s->termlists[i].name = nmem_strdup(s->nmem, type);
 426         s->termlists[i].termlist = termlist_create(s->nmem, s->expected_maxrecs, 15);
 427         s->num_termlists = i + 1;
 428     }
 429     termlist_insert(s->termlists[i].termlist, value);
 430 }
 431
 432 static xmlDoc *normalize_record(struct client *cl, Z_External *rec)
 433 {
 434     struct conf_retrievalprofile *rprofile = cl->database->rprofile;
 435     struct conf_retrievalmap *m;
 436     xmlNode *res;
 437     xmlDoc *rdoc;
 438
 439     // First normalize to XML
 440     if (rprofile->native_syntax == Nativesyn_iso2709)
 441     {
 442         char *buf;
 443         int len;
 444         if (rec->which != Z_External_octet)
 445         {
 446             yaz_log(YLOG_WARN, "Unexpected external branch, probably BER");
 447             return 0;
 448         }
 449         buf = (char*) rec->u.octet_aligned->buf;
 450         len = rec->u.octet_aligned->len;
 451         if (yaz_marc_read_iso2709(rprofile->yaz_marc, buf, len) < 0)
 452         {
 453             yaz_log(YLOG_WARN, "Failed to decode MARC");
 454             return 0;
 455         }
 456         if (yaz_marc_write_xml(rprofile->yaz_marc, &res,
 457                     "http://www.loc.gov/MARC21/slim", 0, 0) < 0)
 458         {
 459             yaz_log(YLOG_WARN, "Failed to encode as XML");
 460             return 0;
 461         }
 462         rdoc = xmlNewDoc((xmlChar *) "1.0");
 463         xmlDocSetRootElement(rdoc, res);
 464     }
 465     else
 466     {
 467         yaz_log(YLOG_FATAL, "Unknown native_syntax in normalize_record");
 468         exit(1);
 469     }
 470
 471     if (global_parameters.dump_records)
 472     {
 473         fprintf(stderr, "Input Record (normalized):\n----------------\n");
 474 #if LIBXML_VERSION >= 20600
 475         xmlDocFormatDump(stderr, rdoc, 1);
 476 #else
 477         xmlDocDump(stderr, rdoc);
 478 #endif
 479     }
 480
 481     for (m = rprofile->maplist; m; m = m->next)
 482     {
 483         xmlDoc *new;
 484         if (m->type != Map_xslt)
 485         {
 486             yaz_log(YLOG_WARN, "Unknown map type");
 487             return 0;
 488         }
 489         if (!(new = xsltApplyStylesheet(m->stylesheet, rdoc, 0)))
 490         {
 491             yaz_log(YLOG_WARN, "XSLT transformation failed");
 492             return 0;
 493         }
 494         xmlFreeDoc(rdoc);
 495         rdoc = new;
 496     }
 497     if (global_parameters.dump_records)
 498     {
 499         fprintf(stderr, "Record:\n----------------\n");
 500 #if LIBXML_VERSION >= 20600
 501         xmlDocFormatDump(stderr, rdoc, 1);
 502 #else
 503         xmlDocDump(stderr, rdoc);
 504 #endif
 505     }
 506     return rdoc;
 507 }
 508
 509 // Extract what appears to be years from buf, storing highest and
 510 // lowest values.
 511 static int extract_years(const char *buf, int *first, int *last)
 512 {
 513     *first = -1;
 514     *last = -1;
 515     while (*buf)
 516     {
 517         const char *e;
 518         int len;
 519
 520         while (*buf && !isdigit(*buf))
 521             buf++;
 522         len = 0;
 523         for (e = buf; *e && isdigit(*e); e++)
 524             len++;
 525         if (len == 4)
 526         {
 527             int value = atoi(buf);
 528             if (*first < 0 || value < *first)
 529                 *first = value;
 530             if (*last < 0 || value > *last)
 531                 *last = value;
 532         }
 533         buf = e;
 534     }
 535     return *first;
 536 }
 537
 538 static struct record *ingest_record(struct client *cl, Z_External *rec)
 539 {
 540     xmlDoc *xdoc = normalize_record(cl, rec);
 541     xmlNode *root, *n;
 542     struct record *res;
 543     struct record_cluster *cluster;
 544     struct session *se = cl->session;
 545     xmlChar *mergekey, *mergekey_norm;
 546     xmlChar *type = 0;
 547     xmlChar *value = 0;
 548     struct conf_service *service = global_parameters.server->service;
 549
 550     if (!xdoc)
 551         return 0;
 552
 553     root = xmlDocGetRootElement(xdoc);
 554     if (!(mergekey = xmlGetProp(root, (xmlChar *) "mergekey")))
 555     {
 556         yaz_log(YLOG_WARN, "No mergekey found in record");
 557         xmlFreeDoc(xdoc);
 558         return 0;
 559     }
 560
 561     res = nmem_malloc(se->nmem, sizeof(struct record));
 562     res->next = 0;
 563     res->client = cl;
 564     res->metadata = nmem_malloc(se->nmem,
 565             sizeof(struct record_metadata*) * service->num_metadata);
 566     memset(res->metadata, 0, sizeof(struct record_metadata*) * service->num_metadata);
 567
 568     mergekey_norm = (xmlChar *) nmem_strdup(se->nmem, (char*) mergekey);
 569     xmlFree(mergekey);
 570     normalize_mergekey((char *) mergekey_norm, 0);
 571
 572     cluster = reclist_insert(se->reclist, res, (char *) mergekey_norm,
 573                              &se->total_merged);
 574     if (global_parameters.dump_records)
 575         yaz_log(YLOG_LOG, "Cluster id %d from %s (#%d)", cluster->recid,
 576                 cl->database->url, cl->records);
 577     if (!cluster)
 578     {
 579         /* no room for record */
 580         xmlFreeDoc(xdoc);
 581         return 0;
 582     }
 583     relevance_newrec(se->relevance, cluster);
 584
 585     for (n = root->children; n; n = n->next)
 586     {
 587         if (type)
 588             xmlFree(type);
 589         if (value)
 590             xmlFree(value);
 591         type = value = 0;
 592
 593         if (n->type != XML_ELEMENT_NODE)
 594             continue;
 595         if (!strcmp((const char *) n->name, "metadata"))
 596         {
 597             struct conf_metadata *md = 0;
 598             struct conf_sortkey *sk = 0;
 599             struct record_metadata **wheretoput, *newm;
 600             int imeta;
 601             int first, last;
 602
 603             type = xmlGetProp(n, (xmlChar *) "type");
 604             value = xmlNodeListGetString(xdoc, n->children, 0);
 605
 606             if (!type || !value)
 607                 continue;
 608
 609             // First, find out what field we're looking at
 610             for (imeta = 0; imeta < service->num_metadata; imeta++)
 611                 if (!strcmp((const char *) type, service->metadata[imeta].name))
 612                 {
 613                     md = &service->metadata[imeta];
 614                     if (md->sortkey_offset >= 0)
 615                         sk = &service->sortkeys[md->sortkey_offset];
 616                     break;
 617                 }
 618             if (!md)
 619             {
 620                 yaz_log(YLOG_WARN, "Ignoring unknown metadata element: %s", type);
 621                 continue;
 622             }
 623
 624             // Find out where we are putting it
 625             if (md->merge == Metadata_merge_no)
 626                 wheretoput = &res->metadata[imeta];
 627             else
 628                 wheretoput = &cluster->metadata[imeta];
 629
 630             // Put it there
 631             newm = nmem_malloc(se->nmem, sizeof(struct record_metadata));
 632             newm->next = 0;
 633             if (md->type == Metadata_type_generic)
 634             {
 635                 char *p, *pe;
 636                 for (p = (char *) value; *p && isspace(*p); p++)
 637                     ;
 638                 for (pe = p + strlen(p) - 1;
 639                         pe > p && strchr(" ,/.:([", *pe); pe--)
 640                     *pe = '\0';
 641                 newm->data.text = nmem_strdup(se->nmem, p);
 642
 643             }
 644             else if (md->type == Metadata_type_year)
 645             {
 646                 if (extract_years((char *) value, &first, &last) < 0)
 647                     continue;
 648             }
 649             else
 650             {
 651                 yaz_log(YLOG_WARN, "Unknown type in metadata element %s", type);
 652                 continue;
 653             }
 654             if (md->type == Metadata_type_year && md->merge != Metadata_merge_range)
 655             {
 656                 yaz_log(YLOG_WARN, "Only range merging supported for years");
 657                 continue;
 658             }
 659             if (md->merge == Metadata_merge_unique)
 660             {
 661                 struct record_metadata *mnode;
 662                 for (mnode = *wheretoput; mnode; mnode = mnode->next)
 663                     if (!strcmp((const char *) mnode->data.text, newm->data.text))
 664                         break;
 665                 if (!mnode)
 666                 {
 667                     newm->next = *wheretoput;
 668                     *wheretoput = newm;
 669                 }
 670             }
 671             else if (md->merge == Metadata_merge_longest)
 672             {
 673                 if (!*wheretoput ||
 674                         strlen(newm->data.text) > strlen((*wheretoput)->data.text))
 675                 {
 676                     *wheretoput = newm;
 677                     if (sk)
 678                     {
 679                         char *s = nmem_strdup(se->nmem, newm->data.text);
 680                         if (!cluster->sortkeys[md->sortkey_offset])
 681                             cluster->sortkeys[md->sortkey_offset] =
 682                                 nmem_malloc(se->nmem, sizeof(union data_types));
 683                         normalize_mergekey(s,
 684                                 (sk->type == Metadata_sortkey_skiparticle));
 685                         cluster->sortkeys[md->sortkey_offset]->text = s;
 686                     }
 687                 }
 688             }
 689             else if (md->merge == Metadata_merge_all || md->merge == Metadata_merge_no)
 690             {
 691                 newm->next = *wheretoput;
 692                 *wheretoput = newm;
 693             }
 694             else if (md->merge == Metadata_merge_range)
 695             {
 696                 assert(md->type == Metadata_type_year);
 697                 if (!*wheretoput)
 698                 {
 699                     *wheretoput = newm;
 700                     (*wheretoput)->data.number.min = first;
 701                     (*wheretoput)->data.number.max = last;
 702                     if (sk)
 703                         cluster->sortkeys[md->sortkey_offset] = &newm->data;
 704                 }
 705                 else
 706                 {
 707                     if (first < (*wheretoput)->data.number.min)
 708                         (*wheretoput)->data.number.min = first;
 709                     if (last > (*wheretoput)->data.number.max)
 710                         (*wheretoput)->data.number.max = last;
 711                 }
 712 #ifdef GAGA
 713                 if (sk)
 714                 {
 715                     union data_types *sdata = cluster->sortkeys[md->sortkey_offset];
 716                     yaz_log(YLOG_LOG, "SK range: %d-%d", sdata->number.min, sdata->number.max);
 717                 }
 718 #endif
 719             }
 720             else
 721                 yaz_log(YLOG_WARN, "Don't know how to merge on element name %s", md->name);
 722
 723             if (md->rank)
 724                 relevance_countwords(se->relevance, cluster,
 725                                      (char *) value, md->rank);
 726             if (md->termlist)
 727             {
 728                 if (md->type == Metadata_type_year)
 729                 {
 730                     char year[64];
 731                     sprintf(year, "%d", last);
 732                     add_facet(se, (char *) type, year);
 733                     if (first != last)
 734                     {
 735                         sprintf(year, "%d", first);
 736                         add_facet(se, (char *) type, year);
 737                     }
 738                 }
 739                 else
 740                     add_facet(se, (char *) type, (char *) value);
 741             }
 742             xmlFree(type);
 743             xmlFree(value);
 744             type = value = 0;
 745         }
 746         else
 747             yaz_log(YLOG_WARN, "Unexpected element %s in internal record", n->name);
 748     }
 749     if (type)
 750         xmlFree(type);
 751     if (value)
 752         xmlFree(value);
 753
 754     xmlFreeDoc(xdoc);
 755
 756     relevance_donerecord(se->relevance, cluster);
 757     se->total_records++;
 758
 759     return res;
 760 }
 761
 762 // Retrieve first defined value for 'name' for given database.
 763 // Will be extended to take into account user associated with session
 764 char *session_setting_oneval(struct session *s, struct database *db, const char *name)
 765 {
 766     int offset = settings_offset(name);
 767
 768     if (offset < 0)
 769         return 0;
 770     if (!db->settings[offset])
 771         return 0;
 772     return db->settings[offset]->value;
 773 }
 774
 775 static void ingest_records(struct client *cl, Z_Records *r)
 776 {
 777 #if USE_TIMING
 778     yaz_timing_t t = yaz_timing_create();
 779 #endif
 780     struct record *rec;
 781     struct session *s = cl->session;
 782     Z_NamePlusRecordList *rlist;
 783     int i;
 784
 785     if (r->which != Z_Records_DBOSD)
 786         return;
 787     rlist = r->u.databaseOrSurDiagnostics;
 788     for (i = 0; i < rlist->num_records; i++)
 789     {
 790         Z_NamePlusRecord *npr = rlist->records[i];
 791
 792         cl->records++;
 793         if (npr->which != Z_NamePlusRecord_databaseRecord)
 794         {
 795             yaz_log(YLOG_WARN,
 796                     "Unexpected record type, probably diagnostic %s",
 797                     cl->database->url);
 798             continue;
 799         }
 800
 801         rec = ingest_record(cl, npr->u.databaseRecord);
 802         if (!rec)
 803             continue;
 804     }
 805     if (s->watchlist[SESSION_WATCH_RECORDS].fun && rlist->num_records)
 806         session_alert_watch(s, SESSION_WATCH_RECORDS);
 807
 808 #if USE_TIMING
 809     yaz_timing_stop(t);
 810     yaz_log(YLOG_LOG, "ingest_records %6.5f %3.2f %3.2f",
 811             yaz_timing_get_real(t), yaz_timing_get_user(t),
 812             yaz_timing_get_sys(t));
 813     yaz_timing_destroy(&t);
 814 #endif
 815 }
 816
 817 static void do_presentResponse(IOCHAN i, Z_APDU *a)
 818 {
 819     struct connection *co = iochan_getdata(i);
 820     struct client *cl = co->client;
 821     Z_PresentResponse *r = a->u.presentResponse;
 822
 823     if (r->records) {
 824         Z_Records *recs = r->records;
 825         if (recs->which == Z_Records_NSD)
 826         {
 827             yaz_log(YLOG_WARN, "Non-surrogate diagnostic %s",
 828                     cl->database->url);
 829             cl->diagnostic = *recs->u.nonSurrogateDiagnostic->condition;
 830             cl->state = Client_Error;
 831         }
 832     }
 833
 834     if (!*r->presentStatus && cl->state != Client_Error)
 835     {
 836         yaz_log(YLOG_DEBUG, "Good Present response %s",
 837                 cl->database->url);
 838         ingest_records(cl, r->records);
 839         cl->state = Client_Idle;
 840     }
 841     else if (*r->presentStatus)
 842     {
 843         yaz_log(YLOG_WARN, "Bad Present response %s",
 844                 cl->database->url);
 845         cl->state = Client_Error;
 846     }
 847 }
 848
 849 static void handler(IOCHAN i, int event)
 850 {
 851     struct connection *co = iochan_getdata(i);
 852     struct client *cl = co->client;
 853     struct session *se = 0;
 854
 855     if (cl)
 856         se = cl->session;
 857     else
 858     {
 859         yaz_log(YLOG_WARN, "Destroying orphan connection");
 860         connection_destroy(co);
 861         return;
 862     }
 863
 864     if (co->state == Conn_Connecting && event & EVENT_OUTPUT)
 865     {
 866         int errcode;
 867         socklen_t errlen = sizeof(errcode);
 868
 869         if (getsockopt(cs_fileno(co->link), SOL_SOCKET, SO_ERROR, &errcode,
 870             &errlen) < 0 || errcode != 0)
 871         {
 872             client_fatal(cl);
 873             return;
 874         }
 875         else
 876         {
 877             yaz_log(YLOG_DEBUG, "Connect OK");
 878             co->state = Conn_Open;
 879             if (cl)
 880                 cl->state = Client_Connected;
 881         }
 882     }
 883
 884     else if (event & EVENT_INPUT)
 885     {
 886         int len = cs_get(co->link, &co->ibuf, &co->ibufsize);
 887
 888         if (len < 0)
 889         {
 890             yaz_log(YLOG_WARN|YLOG_ERRNO, "Error reading from %s",
 891                     cl->database->url);
 892             connection_destroy(co);
 893             return;
 894         }
 895         else if (len == 0)
 896         {
 897             yaz_log(YLOG_WARN, "EOF reading from %s", cl->database->url);
 898             connection_destroy(co);
 899             return;
 900         }
 901         else if (len > 1) // We discard input if we have no connection
 902         {
 903             co->state = Conn_Open;
 904
 905             if (cl && (cl->requestid == se->requestid || cl->state == Client_Initializing))
 906             {
 907                 Z_APDU *a;
 908
 909                 odr_reset(global_parameters.odr_in);
 910                 odr_setbuf(global_parameters.odr_in, co->ibuf, len, 0);
 911                 if (!z_APDU(global_parameters.odr_in, &a, 0, 0))
 912                 {
 913                     client_fatal(cl);
 914                     return;
 915                 }
 916                 switch (a->which)
 917                 {
 918                     case Z_APDU_initResponse:
 919                         do_initResponse(i, a);
 920                         break;
 921                     case Z_APDU_searchResponse:
 922                         do_searchResponse(i, a);
 923                         break;
 924                     case Z_APDU_presentResponse:
 925                         do_presentResponse(i, a);
 926                         break;
 927                     case Z_APDU_close:
 928                         do_closeResponse(i, a);
 929                         break;
 930                     default:
 931                         yaz_log(YLOG_WARN,
 932                                 "Unexpected Z39.50 response from %s",
 933                                 cl->database->url);
 934                         client_fatal(cl);
 935                         return;
 936                 }
 937                 // We aren't expecting staggered output from target
 938                 // if (cs_more(t->link))
 939                 //    iochan_setevent(i, EVENT_INPUT);
 940             }
 941             else  // we throw away response and go to idle mode
 942             {
 943                 yaz_log(YLOG_DEBUG, "Ignoring result of expired operation");
 944                 cl->state = Client_Idle;
 945             }
 946         }
 947         /* if len==1 we do nothing but wait for more input */
 948     }
 949
 950     if (cl->state == Client_Connected) {
 951         send_init(i);
 952     }
 953
 954     if (cl->state == Client_Idle)
 955     {
 956         if (cl->requestid != se->requestid && *se->query) {
 957             send_search(i);
 958         }
 959         else if (cl->hits > 0 && cl->records < global_parameters.toget &&
 960             cl->records < cl->hits) {
 961             send_present(i);
 962         }
 963     }
 964 }
 965
 966 // Disassociate connection from client
 967 static void connection_release(struct connection *co)
 968 {
 969     struct client *cl = co->client;
 970
 971     yaz_log(YLOG_DEBUG, "Connection release %s", co->host->hostport);
 972     if (!cl)
 973         return;
 974     cl->connection = 0;
 975     co->client = 0;
 976 }
 977
 978 // Close connection and recycle structure
 979 static void connection_destroy(struct connection *co)
 980 {
 981     struct host *h = co->host;
 982     cs_close(co->link);
 983     iochan_destroy(co->iochan);
 984
 985     yaz_log(YLOG_DEBUG, "Connection destroy %s", co->host->hostport);
 986     if (h->connections == co)
 987         h->connections = co->next;
 988     else
 989     {
 990         struct connection *pco;
 991         for (pco = h->connections; pco && pco->next != co; pco = pco->next)
 992             ;
 993         if (pco)
 994             pco->next = co->next;
 995         else
 996             abort();
 997     }
 998     if (co->client)
 999     {
1000         if (co->client->state != Client_Idle)
1001             co->client->state = Client_Disconnected;
1002         co->client->connection = 0;
1003     }
1004     co->next = connection_freelist;
1005     connection_freelist = co;
1006 }
1007
1008 // Creates a new connection for client, associated with the host of
1009 // client's database
1010 static struct connection *connection_create(struct client *cl)
1011 {
1012     struct connection *new;
1013     COMSTACK link;
1014     int res;
1015     void *addr;
1016
1017
1018     if (!(link = cs_create(tcpip_type, 0, PROTO_Z3950)))
1019         {
1020             yaz_log(YLOG_FATAL|YLOG_ERRNO, "Failed to create comstack");
1021             exit(1);
1022         }
1023
1024     if (0 == strlen(global_parameters.zproxy_override)){
1025         /* no Z39.50 proxy needed - direct connect */
1026         yaz_log(YLOG_DEBUG, "Connection create %s", cl->database->url);
1027
1028         if (!(addr = cs_straddr(link, cl->database->host->ipport)))
1029             {
1030                 yaz_log(YLOG_WARN|YLOG_ERRNO,
1031                         "Lookup of IP address %s failed",
1032                         cl->database->host->ipport);
1033                 return 0;
1034             }
1035
1036     } else {
1037         /* Z39.50 proxy connect */
1038         yaz_log(YLOG_DEBUG, "Connection create %s proxy %s",
1039                 cl->database->url, global_parameters.zproxy_override);
1040
1041         if (!(addr = cs_straddr(link, global_parameters.zproxy_override)))
1042             {
1043                 yaz_log(YLOG_WARN|YLOG_ERRNO,
1044                         "Lookup of IP address %s failed",
1045                         global_parameters.zproxy_override);
1046                 return 0;
1047             }
1048     }
1049
1050     res = cs_connect(link, addr);
1051     if (res < 0)
1052     {
1053         yaz_log(YLOG_WARN|YLOG_ERRNO, "cs_connect %s", cl->database->url);
1054         return 0;
1055     }
1056
1057     if ((new = connection_freelist))
1058         connection_freelist = new->next;
1059     else
1060     {
1061         new = xmalloc(sizeof (struct connection));
1062         new->ibuf = 0;
1063         new->ibufsize = 0;
1064     }
1065     new->state = Conn_Connecting;
1066     new->host = cl->database->host;
1067     new->next = new->host->connections;
1068     new->host->connections = new;
1069     new->client = cl;
1070     cl->connection = new;
1071     new->link = link;
1072
1073     new->iochan = iochan_create(cs_fileno(link), 0, handler, 0);
1074     iochan_setdata(new->iochan, new);
1075     new->iochan->next = channel_list;
1076     channel_list = new->iochan;
1077     return new;
1078 }
1079
1080 // Close connection and set state to error
1081 static void client_fatal(struct client *cl)
1082 {
1083     yaz_log(YLOG_WARN, "Fatal error from %s", cl->database->url);
1084     connection_destroy(cl->connection);
1085     cl->state = Client_Error;
1086 }
1087
1088 // Ensure that client has a connection associated
1089 static int client_prep_connection(struct client *cl)
1090 {
1091     struct connection *co;
1092     struct session *se = cl->session;
1093     struct host *host = cl->database->host;
1094
1095     co = cl->connection;
1096
1097     yaz_log(YLOG_DEBUG, "Client prep %s", cl->database->url);
1098
1099     if (!co)
1100     {
1101         // See if someone else has an idle connection
1102         // We should look at timestamps here to select the longest-idle connection
1103         for (co = host->connections; co; co = co->next)
1104             if (co->state == Conn_Open && (!co->client || co->client->session != se))
1105                 break;
1106         if (co)
1107         {
1108             connection_release(co);
1109             cl->connection = co;
1110             co->client = cl;
1111         }
1112         else
1113             co = connection_create(cl);
1114     }
1115     if (co)
1116     {
1117         if (co->state == Conn_Connecting)
1118         {
1119             cl->state = Client_Connecting;
1120             iochan_setflag(co->iochan, EVENT_OUTPUT);
1121         }
1122         else if (co->state == Conn_Open)
1123         {
1124             if (cl->state == Client_Error || cl->state == Client_Disconnected)
1125                 cl->state = Client_Idle;
1126             iochan_setflag(co->iochan, EVENT_OUTPUT);
1127         }
1128         return 1;
1129     }
1130     else
1131         return 0;
1132 }
1133
1134 #ifdef GAGA // Moved to database.c
1135
1136 // This function will most likely vanish when a proper target profile mechanism is
1137 // introduced.
1138 void load_simpletargets(const char *fn)
1139 {
1140     FILE *f = fopen(fn, "r");
1141     char line[256];
1142
1143     if (!f)
1144     {
1145         yaz_log(YLOG_WARN|YLOG_ERRNO, "open %s", fn);
1146         exit(1);
1147     }
1148
1149     while (fgets(line, 255, f))
1150     {
1151         char *url, *db;
1152         char *name;
1153         struct host *host;
1154         struct database *database;
1155
1156         if (strncmp(line, "target ", 7))
1157             continue;
1158         line[strlen(line) - 1] = '\0';
1159
1160         if ((name = strchr(line, ';')))
1161             *(name++) = '\0';
1162
1163         url = line + 7;
1164         if ((db = strchr(url, '/')))
1165             *(db++) = '\0';
1166         else
1167             db = "Default";
1168
1169         yaz_log(YLOG_LOG, "Target: %s, '%s'", url, db);
1170         for (host = hosts; host; host = host->next)
1171             if (!strcmp((const char *) url, host->hostport))
1172                 break;
1173         if (!host)
1174         {
1175             struct addrinfo *addrinfo, hints;
1176             char *port;
1177             char ipport[128];
1178             unsigned char addrbuf[4];
1179             int res;
1180
1181             host = xmalloc(sizeof(struct host));
1182             host->hostport = xstrdup(url);
1183             host->connections = 0;
1184
1185             if ((port = strchr(url, ':')))
1186                 *(port++) = '\0';
1187             else
1188                 port = "210";
1189
1190             hints.ai_flags = 0;
1191             hints.ai_family = PF_INET;
1192             hints.ai_socktype = SOCK_STREAM;
1193             hints.ai_protocol = IPPROTO_TCP;
1194             hints.ai_addrlen = 0;
1195             hints.ai_addr = 0;
1196             hints.ai_canonname = 0;
1197             hints.ai_next = 0;
1198             // This is not robust code. It assumes that getaddrinfo returns AF_INET
1199             // address.
1200             if ((res = getaddrinfo(url, port, &hints, &addrinfo)))
1201             {
1202                 yaz_log(YLOG_WARN, "Failed to resolve %s: %s", url, gai_strerror(res));
1203                 xfree(host->hostport);
1204                 xfree(host);
1205                 continue;
1206             }
1207             assert(addrinfo->ai_family == PF_INET);
1208             memcpy(addrbuf, &((struct sockaddr_in*)addrinfo->ai_addr)->sin_addr.s_addr, 4);
1209             sprintf(ipport, "%u.%u.%u.%u:%s",
1210                     addrbuf[0], addrbuf[1], addrbuf[2], addrbuf[3], port);
1211             host->ipport = xstrdup(ipport);
1212             freeaddrinfo(addrinfo);
1213             host->next = hosts;
1214             hosts = host;
1215         }
1216         database = xmalloc(sizeof(struct database));
1217         database->host = host;
1218         database->url = xmalloc(strlen(url) + strlen(db) + 2);
1219         strcpy(database->url, url);
1220         strcat(database->url, "/");
1221         strcat(database->url, db);
1222         if (name)
1223             database->name = xstrdup(name);
1224         else
1225             database->name = 0;
1226
1227         database->databases = xmalloc(2 * sizeof(char *));
1228         database->databases[0] = xstrdup(db);
1229         database->databases[1] = 0;
1230         database->errors = 0;
1231         database->qprofile = 0;
1232         database->rprofile = database_retrieval_profile(database);
1233         database->next = databases;
1234         databases = database;
1235
1236     }
1237     fclose(f);
1238 }
1239
1240 #endif
1241
1242 static struct client *client_create(void)
1243 {
1244     struct client *r;
1245     if (client_freelist)
1246     {
1247         r = client_freelist;
1248         client_freelist = client_freelist->next;
1249     }
1250     else
1251         r = xmalloc(sizeof(struct client));
1252     r->database = 0;
1253     r->connection = 0;
1254     r->session = 0;
1255     r->hits = 0;
1256     r->records = 0;
1257     r->setno = 0;
1258     r->requestid = -1;
1259     r->diagnostic = 0;
1260     r->state = Client_Disconnected;
1261     r->next = 0;
1262     return r;
1263 }
1264
1265 void client_destroy(struct client *c)
1266 {
1267     struct session *se = c->session;
1268     if (c == se->clients)
1269         se->clients = c->next;
1270     else
1271     {
1272         struct client *cc;
1273         for (cc = se->clients; cc && cc->next != c; cc = cc->next)
1274             ;
1275         if (cc)
1276             cc->next = c->next;
1277     }
1278     if (c->connection)
1279         connection_release(c->connection);
1280     c->next = client_freelist;
1281     client_freelist = c;
1282 }
1283
1284 void session_set_watch(struct session *s, int what, session_watchfun fun, void *data)
1285 {
1286     s->watchlist[what].fun = fun;
1287     s->watchlist[what].data = data;
1288 }
1289
1290 void session_alert_watch(struct session *s, int what)
1291 {
1292     if (!s->watchlist[what].fun)
1293         return;
1294     (*s->watchlist[what].fun)(s->watchlist[what].data);
1295     s->watchlist[what].fun = 0;
1296     s->watchlist[what].data = 0;
1297 }
1298
1299 //callback for grep_databases
1300 static void select_targets_callback(void *context, struct database *db)
1301 {
1302     struct session *se = (struct session*) context;
1303     struct client *cl = client_create();
1304     cl->database = db;
1305     cl->session = se;
1306     cl->next = se->clients;
1307     se->clients = cl;
1308 }
1309
1310 // This should be extended with parameters to control selection criteria
1311 // Associates a set of clients with a session;
1312 int select_targets(struct session *se, struct database_criterion *crit)
1313 {
1314     while (se->clients)
1315         client_destroy(se->clients);
1316
1317     return grep_databases(se, crit, select_targets_callback);
1318 }
1319
1320 int session_active_clients(struct session *s)
1321 {
1322     struct client *c;
1323     int res = 0;
1324
1325     for (c = s->clients; c; c = c->next)
1326         if (c->connection && (c->state == Client_Connecting ||
1327                     c->state == Client_Initializing ||
1328                     c->state == Client_Searching ||
1329                     c->state == Client_Presenting))
1330             res++;
1331
1332     return res;
1333 }
1334
1335 // parses crit1=val1,crit2=val2|val3,...
1336 static struct database_criterion *parse_filter(NMEM m, const char *buf)
1337 {
1338     struct database_criterion *res = 0;
1339     char **values;
1340     int num;
1341     int i;
1342
1343     if (!buf || !*buf)
1344         return 0;
1345     nmem_strsplit(m, ",", buf,  &values, &num);
1346     for (i = 0; i < num; i++)
1347     {
1348         char **subvalues;
1349         int subnum;
1350         int subi;
1351         struct database_criterion *new = nmem_malloc(m, sizeof(*new));
1352         char *eq = strchr(values[i], '=');
1353         if (!eq)
1354         {
1355             yaz_log(YLOG_WARN, "Missing equal-sign in filter");
1356             return 0;
1357         }
1358         *(eq++) = '\0';
1359         new->name = values[i];
1360         nmem_strsplit(m, "|", eq, &subvalues, &subnum);
1361         new->values = 0;
1362         for (subi = 0; subi < subnum; subi++)
1363         {
1364             struct database_criterion_value *newv = nmem_malloc(m, sizeof(*newv));
1365             newv->value = subvalues[subi];
1366             newv->next = new->values;
1367             new->values = newv;
1368         }
1369         new->next = res;
1370         res = new;
1371     }
1372     return res;
1373 }
1374
1375 char *search(struct session *se, char *query, char *filter)
1376 {
1377     int live_channels = 0;
1378     struct client *cl;
1379     struct database_criterion *criteria;
1380
1381     yaz_log(YLOG_DEBUG, "Search");
1382
1383     nmem_reset(se->nmem);
1384     criteria = parse_filter(se->nmem, filter);
1385     strcpy(se->query, query);
1386     se->requestid++;
1387     // Release any existing clients
1388     select_targets(se, criteria);
1389     for (cl = se->clients; cl; cl = cl->next)
1390     {
1391         if (client_prep_connection(cl))
1392             live_channels++;
1393     }
1394     if (live_channels)
1395     {
1396         int maxrecs = live_channels * global_parameters.toget;
1397         se->num_termlists = 0;
1398         se->reclist = reclist_create(se->nmem, maxrecs);
1399         // This will be initialized in send_search()
1400         se->relevance = 0;
1401         se->total_records = se->total_hits = se->total_merged = 0;
1402         se->expected_maxrecs = maxrecs;
1403     }
1404     else
1405         return "NOTARGETS";
1406
1407     return 0;
1408 }
1409
1410 void destroy_session(struct session *s)
1411 {
1412     yaz_log(YLOG_LOG, "Destroying session");
1413     while (s->clients)
1414         client_destroy(s->clients);
1415     nmem_destroy(s->nmem);
1416     wrbuf_destroy(s->wrbuf);
1417 }
1418
1419 struct session *new_session()
1420 {
1421     int i;
1422     struct session *session = xmalloc(sizeof(*session));
1423
1424     yaz_log(YLOG_DEBUG, "New pazpar2 session");
1425
1426     session->total_hits = 0;
1427     session->total_records = 0;
1428     session->num_termlists = 0;
1429     session->reclist = 0;
1430     session->requestid = -1;
1431     session->clients = 0;
1432     session->expected_maxrecs = 0;
1433     session->query[0] = '\0';
1434     session->nmem = nmem_create();
1435     session->wrbuf = wrbuf_alloc();
1436     for (i = 0; i <= SESSION_WATCH_MAX; i++)
1437     {
1438         session->watchlist[i].data = 0;
1439         session->watchlist[i].fun = 0;
1440     }
1441
1442     return session;
1443 }
1444
1445 struct hitsbytarget *hitsbytarget(struct session *se, int *count)
1446 {
1447     static struct hitsbytarget res[1000]; // FIXME MM
1448     struct client *cl;
1449
1450     *count = 0;
1451     for (cl = se->clients; cl; cl = cl->next)
1452     {
1453         res[*count].id = cl->database->url;
1454         res[*count].name = cl->database->name;
1455         res[*count].hits = cl->hits;
1456         res[*count].records = cl->records;
1457         res[*count].diagnostic = cl->diagnostic;
1458         res[*count].state = client_states[cl->state];
1459         res[*count].connected  = cl->connection ? 1 : 0;
1460         (*count)++;
1461     }
1462
1463     return res;
1464 }
1465
1466 struct termlist_score **termlist(struct session *s, const char *name, int *num)
1467 {
1468     int i;
1469
1470     for (i = 0; i < s->num_termlists; i++)
1471         if (!strcmp((const char *) s->termlists[i].name, name))
1472             return termlist_highscore(s->termlists[i].termlist, num);
1473     return 0;
1474 }
1475
1476 #ifdef MISSING_HEADERS
1477 void report_nmem_stats(void)
1478 {
1479     size_t in_use, is_free;
1480
1481     nmem_get_memory_in_use(&in_use);
1482     nmem_get_memory_free(&is_free);
1483
1484     yaz_log(YLOG_LOG, "nmem stat: use=%ld free=%ld",
1485             (long) in_use, (long) is_free);
1486 }
1487 #endif
1488
1489 struct record_cluster *show_single(struct session *s, int id)
1490 {
1491     struct record_cluster *r;
1492
1493     reclist_rewind(s->reclist);
1494     while ((r = reclist_read_record(s->reclist)))
1495         if (r->recid == id)
1496             return r;
1497     return 0;
1498 }
1499
1500 struct record_cluster **show(struct session *s, struct reclist_sortparms *sp, int start,
1501         int *num, int *total, int *sumhits, NMEM nmem_show)
1502 {
1503     struct record_cluster **recs = nmem_malloc(nmem_show, *num
1504                                        * sizeof(struct record_cluster *));
1505     struct reclist_sortparms *spp;
1506     int i;
1507 #if USE_TIMING
1508     yaz_timing_t t = yaz_timing_create();
1509 #endif
1510
1511     for (spp = sp; spp; spp = spp->next)
1512         if (spp->type == Metadata_sortkey_relevance)
1513         {
1514             relevance_prepare_read(s->relevance, s->reclist);
1515             break;
1516         }
1517     reclist_sort(s->reclist, sp);
1518
1519     *total = s->reclist->num_records;
1520     *sumhits = s->total_hits;
1521
1522     for (i = 0; i < start; i++)
1523         if (!reclist_read_record(s->reclist))
1524         {
1525             *num = 0;
1526             recs = 0;
1527             break;
1528         }
1529
1530     for (i = 0; i < *num; i++)
1531     {
1532         struct record_cluster *r = reclist_read_record(s->reclist);
1533         if (!r)
1534         {
1535             *num = i;
1536             break;
1537         }
1538         recs[i] = r;
1539     }
1540 #if USE_TIMING
1541     yaz_timing_stop(t);
1542     yaz_log(YLOG_LOG, "show %6.5f %3.2f %3.2f",
1543             yaz_timing_get_real(t), yaz_timing_get_user(t),
1544             yaz_timing_get_sys(t));
1545     yaz_timing_destroy(&t);
1546 #endif
1547     return recs;
1548 }
1549
1550 void statistics(struct session *se, struct statistics *stat)
1551 {
1552     struct client *cl;
1553     int count = 0;
1554
1555     memset(stat, 0, sizeof(*stat));
1556     for (cl = se->clients; cl; cl = cl->next)
1557     {
1558         if (!cl->connection)
1559             stat->num_no_connection++;
1560         switch (cl->state)
1561         {
1562             case Client_Connecting: stat->num_connecting++; break;
1563             case Client_Initializing: stat->num_initializing++; break;
1564             case Client_Searching: stat->num_searching++; break;
1565             case Client_Presenting: stat->num_presenting++; break;
1566             case Client_Idle: stat->num_idle++; break;
1567             case Client_Failed: stat->num_failed++; break;
1568             case Client_Error: stat->num_error++; break;
1569             default: break;
1570         }
1571         count++;
1572     }
1573     stat->num_hits = se->total_hits;
1574     stat->num_records = se->total_records;
1575
1576     stat->num_clients = count;
1577 }
1578
1579 static void start_http_listener(void)
1580 {
1581     char hp[128] = "";
1582     struct conf_server *ser = global_parameters.server;
1583
1584     if (*global_parameters.listener_override)
1585         strcpy(hp, global_parameters.listener_override);
1586     else
1587     {
1588         strcpy(hp, ser->host ? ser->host : "");
1589         if (ser->port)
1590         {
1591             if (*hp)
1592                 strcat(hp, ":");
1593             sprintf(hp + strlen(hp), "%d", ser->port);
1594         }
1595     }
1596     http_init(hp);
1597 }
1598
1599 // Initialize CCL map for a target
1600 // Note: This approach ignores user-specific CCL maps, for which I
1601 // don't presently see any application.
1602 static void prepare_cclmap(void *context, struct database *db)
1603 {
1604     struct setting *s;
1605
1606     if (!db->settings)
1607         return;
1608     db->ccl_map = ccl_qual_mk();
1609     for (s = db->settings[PZ_CCLMAP]; s; s = s->next)
1610         if (!*s->user)
1611         {
1612             char *p = strchr(s->name + 3, ':');
1613             if (!p)
1614             {
1615                 yaz_log(YLOG_FATAL, "Malformed cclmap name: %s", s->name);
1616                 exit(1);
1617             }
1618             p++;
1619             ccl_qual_fitem(db->ccl_map, s->value, p);
1620         }
1621 }
1622
1623 // Read settings for each database, and prepare a CCL map for that database
1624 static void prepare_cclmaps(void)
1625 {
1626     grep_databases(0, 0, prepare_cclmap);
1627 }
1628
1629 static void start_proxy(void)
1630 {
1631     char hp[128] = "";
1632     struct conf_server *ser = global_parameters.server;
1633
1634     if (*global_parameters.proxy_override)
1635         strcpy(hp, global_parameters.proxy_override);
1636     else if (ser->proxy_host || ser->proxy_port)
1637     {
1638         strcpy(hp, ser->proxy_host ? ser->proxy_host : "");
1639         if (ser->proxy_port)
1640         {
1641             if (*hp)
1642                 strcat(hp, ":");
1643             sprintf(hp + strlen(hp), "%d", ser->proxy_port);
1644         }
1645     }
1646     else
1647         return;
1648
1649     http_set_proxyaddr(hp, ser->myurl ? ser->myurl : "");
1650 }
1651
1652 static void start_zproxy(void)
1653 {
1654     struct conf_server *ser = global_parameters.server;
1655
1656     if (*global_parameters.zproxy_override){
1657         yaz_log(YLOG_LOG, "Z39.50 proxy  %s",
1658                 global_parameters.zproxy_override);
1659         return;
1660     }
1661
1662     else if (ser->zproxy_host || ser->zproxy_port)
1663     {
1664         char hp[128] = "";
1665
1666         strcpy(hp, ser->zproxy_host ? ser->zproxy_host : "");
1667         if (ser->zproxy_port)
1668         {
1669             if (*hp)
1670                 strcat(hp, ":");
1671             else
1672                 strcat(hp, "@:");
1673
1674             sprintf(hp + strlen(hp), "%d", ser->zproxy_port);
1675         }
1676         strcpy(global_parameters.zproxy_override, hp);
1677         yaz_log(YLOG_LOG, "Z39.50 proxy  %s",
1678                 global_parameters.zproxy_override);
1679
1680     }
1681     else
1682         return;
1683 }
1684
1685
1686
1687 int main(int argc, char **argv)
1688 {
1689     int ret;
1690     char *arg;
1691
1692     if (signal(SIGPIPE, SIG_IGN) == SIG_ERR)
1693         yaz_log(YLOG_WARN|YLOG_ERRNO, "signal");
1694
1695     yaz_log_init(YLOG_DEFAULT_LEVEL, "pazpar2", 0);
1696
1697     while ((ret = options("t:f:x:h:p:z:s:d", argv, argc, &arg)) != -2)
1698     {
1699         switch (ret) {
1700             case 'f':
1701                 if (!read_config(arg))
1702                     exit(1);
1703                 break;
1704             case 'h':
1705                 strcpy(global_parameters.listener_override, arg);
1706                 break;
1707             case 'p':
1708                 strcpy(global_parameters.proxy_override, arg);
1709                 break;
1710             case 'z':
1711                 strcpy(global_parameters.zproxy_override, arg);
1712                 break;
1713             case 't':
1714                 strcpy(global_parameters.settings_path_override, arg);
1715                 break;
1716             case 's':
1717                 load_simpletargets(arg);
1718                 break;
1719             case 'd':
1720                 global_parameters.dump_records = 1;
1721                 break;
1722             default:
1723                 fprintf(stderr, "Usage: pazpar2\n"
1724                         "    -f configfile\n"
1725                         "    -h [host:]port          (REST protocol listener)\n"
1726                         "    -C cclconfig\n"
1727                         "    -s simpletargetfile\n"
1728                         "    -p hostname[:portno]    (HTTP proxy)\n"
1729                         "    -z hostname[:portno]    (Z39.50 proxy)\n"
1730                         "    -d                      (show internal records)\n");
1731                 exit(1);
1732         }
1733     }
1734
1735     if (!config)
1736     {
1737         yaz_log(YLOG_FATAL, "Load config with -f");
1738         exit(1);
1739     }
1740     global_parameters.server = config->servers;
1741
1742     start_http_listener();
1743     start_proxy();
1744     start_zproxy();
1745
1746     if (*global_parameters.settings_path_override)
1747         settings_read(global_parameters.settings_path_override);
1748     else if (global_parameters.server->settings)
1749         settings_read(global_parameters.server->settings);
1750     else
1751         yaz_log(YLOG_WARN, "No settings-directory specified. Problems may ensue!");
1752     prepare_cclmaps();
1753     global_parameters.yaz_marc = yaz_marc_create();
1754     yaz_marc_subfield_str(global_parameters.yaz_marc, "\t");
1755     global_parameters.odr_in = odr_createmem(ODR_DECODE);
1756     global_parameters.odr_out = odr_createmem(ODR_ENCODE);
1757
1758     event_loop(&channel_list);
1759
1760     return 0;
1761 }
1762
1763 /*
1764  * Local variables:
1765  * c-basic-offset: 4
1766  * indent-tabs-mode: nil
1767  * End:
1768  * vim: shiftwidth=4 tabstop=8 expandtab
1769  */