src/pazpar2.c

   1 /* $Id: pazpar2.c,v 1.7 2007-01-03 06:23:44 quinn 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/comstack.h>
  15 #include <yaz/tcpip.h>
  16 #include <yaz/proto.h>
  17 #include <yaz/readconf.h>
  18 #include <yaz/pquery.h>
  19 #include <yaz/yaz-util.h>
  20 #include <yaz/nmem.h>
  21
  22 #include "pazpar2.h"
  23 #include "eventl.h"
  24 #include "command.h"
  25 #include "http.h"
  26 #include "termlists.h"
  27 #include "reclists.h"
  28 #include "relevance.h"
  29 #include "config.h"
  30
  31 #define PAZPAR2_VERSION "0.1"
  32 #define MAX_CHUNK 15
  33
  34 static void client_fatal(struct client *cl);
  35 static void connection_destroy(struct connection *co);
  36 static int client_prep_connection(struct client *cl);
  37 static void ingest_records(struct client *cl, Z_Records *r);
  38 static struct conf_retrievalprofile *database_retrieval_profile(struct database *db);
  39 void session_alert_watch(struct session *s, int what);
  40
  41 IOCHAN channel_list = 0;  // Master list of connections we're handling events to
  42
  43 static struct connection *connection_freelist = 0;
  44 static struct client *client_freelist = 0;
  45
  46 static struct host *hosts = 0;  // The hosts we know about
  47 static struct database *databases = 0; // The databases we know about
  48
  49 static char *client_states[] = {
  50     "Client_Connecting",
  51     "Client_Connected",
  52     "Client_Idle",
  53     "Client_Initializing",
  54     "Client_Searching",
  55     "Client_Presenting",
  56     "Client_Error",
  57     "Client_Failed",
  58     "Client_Disconnected",
  59     "Client_Stopped"
  60 };
  61
  62 struct parameters global_parameters =
  63 {
  64     0,
  65     30,
  66     "81",
  67     "Index Data PazPar2 (MasterKey)",
  68     PAZPAR2_VERSION,
  69     600, // 10 minutes
  70     60,
  71     100,
  72     MAX_CHUNK,
  73     0,
  74     0,
  75     0,
  76     0
  77 };
  78
  79
  80 static int send_apdu(struct client *c, Z_APDU *a)
  81 {
  82     struct connection *co = c->connection;
  83     char *buf;
  84     int len, r;
  85
  86     if (!z_APDU(global_parameters.odr_out, &a, 0, 0))
  87     {
  88         odr_perror(global_parameters.odr_out, "Encoding APDU");
  89         abort();
  90     }
  91     buf = odr_getbuf(global_parameters.odr_out, &len, 0);
  92     r = cs_put(co->link, buf, len);
  93     if (r < 0)
  94     {
  95         yaz_log(YLOG_WARN, "cs_put: %s", cs_errmsg(cs_errno(co->link)));
  96         return -1;
  97     }
  98     else if (r == 1)
  99     {
 100         fprintf(stderr, "cs_put incomplete (ParaZ does not handle that)\n");
 101         exit(1);
 102     }
 103     odr_reset(global_parameters.odr_out); /* release the APDU structure  */
 104     co->state = Conn_Waiting;
 105     return 0;
 106 }
 107
 108
 109 static void send_init(IOCHAN i)
 110 {
 111     struct connection *co = iochan_getdata(i);
 112     struct client *cl = co->client;
 113     Z_APDU *a = zget_APDU(global_parameters.odr_out, Z_APDU_initRequest);
 114
 115     a->u.initRequest->implementationId = global_parameters.implementationId;
 116     a->u.initRequest->implementationName = global_parameters.implementationName;
 117     a->u.initRequest->implementationVersion =
 118         global_parameters.implementationVersion;
 119     ODR_MASK_SET(a->u.initRequest->options, Z_Options_search);
 120     ODR_MASK_SET(a->u.initRequest->options, Z_Options_present);
 121     ODR_MASK_SET(a->u.initRequest->options, Z_Options_namedResultSets);
 122
 123     ODR_MASK_SET(a->u.initRequest->protocolVersion, Z_ProtocolVersion_1);
 124     ODR_MASK_SET(a->u.initRequest->protocolVersion, Z_ProtocolVersion_2);
 125     ODR_MASK_SET(a->u.initRequest->protocolVersion, Z_ProtocolVersion_3);
 126     if (send_apdu(cl, a) >= 0)
 127     {
 128         iochan_setflags(i, EVENT_INPUT);
 129         cl->state = Client_Initializing;
 130     }
 131     else
 132         cl->state = Client_Error;
 133     odr_reset(global_parameters.odr_out);
 134 }
 135
 136 static void send_search(IOCHAN i)
 137 {
 138     struct connection *co = iochan_getdata(i);
 139     struct client *cl = co->client;
 140     struct session *se = cl->session;
 141     struct database *db = cl->database;
 142     Z_APDU *a = zget_APDU(global_parameters.odr_out, Z_APDU_searchRequest);
 143     int ndb, cerror, cpos;
 144     char **databaselist;
 145     Z_Query *zquery;
 146     struct ccl_rpn_node *cn;
 147     int ssub = 0, lslb = 100000, mspn = 10;
 148
 149     yaz_log(YLOG_DEBUG, "Sending search");
 150
 151     cn = ccl_find_str(global_parameters.ccl_filter, se->query, &cerror, &cpos);
 152     if (!cn)
 153         return;
 154     a->u.searchRequest->query = zquery = odr_malloc(global_parameters.odr_out,
 155             sizeof(Z_Query));
 156     zquery->which = Z_Query_type_1;
 157     zquery->u.type_1 = ccl_rpn_query(global_parameters.odr_out, cn);
 158     ccl_rpn_delete(cn);
 159
 160     for (ndb = 0; db->databases[ndb]; ndb++)
 161         ;
 162     databaselist = odr_malloc(global_parameters.odr_out, sizeof(char*) * ndb);
 163     for (ndb = 0; db->databases[ndb]; ndb++)
 164         databaselist[ndb] = db->databases[ndb];
 165
 166     a->u.presentRequest->preferredRecordSyntax =
 167             yaz_oidval_to_z3950oid(global_parameters.odr_out,
 168             CLASS_RECSYN, VAL_USMARC);
 169     a->u.searchRequest->smallSetUpperBound = &ssub;
 170     a->u.searchRequest->largeSetLowerBound = &lslb;
 171     a->u.searchRequest->mediumSetPresentNumber = &mspn;
 172     a->u.searchRequest->resultSetName = "Default";
 173     a->u.searchRequest->databaseNames = databaselist;
 174     a->u.searchRequest->num_databaseNames = ndb;
 175
 176     if (send_apdu(cl, a) >= 0)
 177     {
 178         iochan_setflags(i, EVENT_INPUT);
 179         cl->state = Client_Searching;
 180         cl->requestid = se->requestid;
 181     }
 182     else
 183         cl->state = Client_Error;
 184
 185     odr_reset(global_parameters.odr_out);
 186 }
 187
 188 static void send_present(IOCHAN i)
 189 {
 190     struct connection *co = iochan_getdata(i);
 191     struct client *cl = co->client;
 192     Z_APDU *a = zget_APDU(global_parameters.odr_out, Z_APDU_presentRequest);
 193     int toget;
 194     int start = cl->records + 1;
 195
 196     toget = global_parameters.chunk;
 197     if (toget > cl->hits - cl->records)
 198         toget = cl->hits - cl->records;
 199
 200     yaz_log(YLOG_DEBUG, "Trying to present %d records\n", toget);
 201
 202     a->u.presentRequest->resultSetStartPoint = &start;
 203     a->u.presentRequest->numberOfRecordsRequested = &toget;
 204
 205     a->u.presentRequest->resultSetId = "Default";
 206
 207     a->u.presentRequest->preferredRecordSyntax =
 208             yaz_oidval_to_z3950oid(global_parameters.odr_out,
 209             CLASS_RECSYN, VAL_USMARC);
 210
 211     if (send_apdu(cl, a) >= 0)
 212     {
 213         iochan_setflags(i, EVENT_INPUT);
 214         cl->state = Client_Presenting;
 215     }
 216     else
 217         cl->state = Client_Error;
 218     odr_reset(global_parameters.odr_out);
 219 }
 220
 221 static void do_initResponse(IOCHAN i, Z_APDU *a)
 222 {
 223     struct connection *co = iochan_getdata(i);
 224     struct client *cl = co->client;
 225     Z_InitResponse *r = a->u.initResponse;
 226
 227     yaz_log(YLOG_DEBUG, "Received init response");
 228
 229     if (*r->result)
 230     {
 231         cl->state = Client_Idle;
 232     }
 233     else
 234         cl->state = Client_Failed; // FIXME need to do something to the connection
 235 }
 236
 237 static void do_searchResponse(IOCHAN i, Z_APDU *a)
 238 {
 239     struct connection *co = iochan_getdata(i);
 240     struct client *cl = co->client;
 241     struct session *se = cl->session;
 242     Z_SearchResponse *r = a->u.searchResponse;
 243
 244     yaz_log(YLOG_DEBUG, "Searchresponse (status=%d)", *r->searchStatus);
 245
 246     if (*r->searchStatus)
 247     {
 248         cl->hits = *r->resultCount;
 249         se->total_hits += cl->hits;
 250         if (r->presentStatus && !*r->presentStatus && r->records)
 251         {
 252             yaz_log(YLOG_DEBUG, "Records in search response");
 253             cl->records += *r->numberOfRecordsReturned;
 254             ingest_records(cl, r->records);
 255         }
 256         cl->state = Client_Idle;
 257     }
 258     else
 259     {          /*"FAILED"*/
 260         cl->hits = 0;
 261         cl->state = Client_Error;
 262         if (r->records) {
 263             Z_Records *recs = r->records;
 264             if (recs->which == Z_Records_NSD)
 265             {
 266                 yaz_log(YLOG_WARN, "Non-surrogate diagnostic");
 267                 cl->diagnostic = *recs->u.nonSurrogateDiagnostic->condition;
 268                 cl->state = Client_Error;
 269             }
 270         }
 271     }
 272 }
 273
 274 char *normalize_mergekey(char *buf)
 275 {
 276     char *p = buf, *pout = buf;
 277
 278     while (*p)
 279     {
 280         while (*p && !isalnum(*p))
 281             p++;
 282         while (isalnum(*p))
 283             *(pout++) = tolower(*(p++));
 284         if (*p)
 285             *(pout++) = ' ';
 286         while (*p && !isalnum(*p))
 287             p++;
 288     }
 289     if (buf != pout)
 290         *pout = '\0';
 291
 292     return buf;
 293 }
 294
 295
 296 #ifdef GAGA
 297 // FIXME needs to be generalized. Should flexibly generate X lists per search
 298 static void extract_subject(struct session *s, const char *rec)
 299 {
 300     const char *field, *subfield;
 301
 302     while ((field = find_field(rec, "650")))
 303     {
 304         rec = field;
 305         if ((subfield = find_subfield(field, 'a')))
 306         {
 307             char *e, *ef;
 308             char buf[1024];
 309             int len;
 310
 311             ef = index(subfield, '\n');
 312             if (!ef)
 313                 return;
 314             if ((e = index(subfield, '\t')) && e < ef)
 315                 ef = e;
 316             while (ef > subfield && !isalpha(*(ef - 1)) && *(ef - 1) != ')')
 317                 ef--;
 318             len = ef - subfield;
 319             assert(len < 1023);
 320             memcpy(buf, subfield, len);
 321             buf[len] = '\0';
 322 #ifdef FIXME
 323             if (*buf)
 324                 termlist_insert(s->termlist, buf);
 325 #endif
 326         }
 327     }
 328 }
 329 #endif
 330
 331 static void add_facet(struct session *s, const char *type, const char *value)
 332 {
 333     int i;
 334
 335     for (i = 0; i < s->num_termlists; i++)
 336         if (!strcmp(s->termlists[i].name, type))
 337             break;
 338     if (i == s->num_termlists)
 339     {
 340         if (i == SESSION_MAX_TERMLISTS)
 341         {
 342             yaz_log(YLOG_FATAL, "Too many termlists");
 343             exit(1);
 344         }
 345         s->termlists[i].name = nmem_strdup(s->nmem, type);
 346         s->termlists[i].termlist = termlist_create(s->nmem, s->expected_maxrecs, 15);
 347         s->num_termlists = i + 1;
 348     }
 349     termlist_insert(s->termlists[i].termlist, value);
 350 }
 351
 352 static xmlDoc *normalize_record(struct client *cl, Z_External *rec)
 353 {
 354     struct conf_retrievalprofile *rprofile = cl->database->rprofile;
 355     struct conf_retrievalmap *m;
 356     xmlNode *res;
 357     xmlDoc *rdoc;
 358
 359     // First normalize to XML
 360     if (rprofile->native_syntax == Nativesyn_iso2709)
 361     {
 362         char *buf;
 363         int len;
 364         if (rec->which != Z_External_octet)
 365         {
 366             yaz_log(YLOG_WARN, "Unexpected external branch, probably BER");
 367             return 0;
 368         }
 369         buf = (char*) rec->u.octet_aligned->buf;
 370         len = rec->u.octet_aligned->len;
 371         if (yaz_marc_read_iso2709(rprofile->yaz_marc, buf, len) < 0)
 372         {
 373             yaz_log(YLOG_WARN, "Failed to decode MARC");
 374             return 0;
 375         }
 376         if (yaz_marc_write_xml(rprofile->yaz_marc, &res,
 377                     "http://www.loc.gov/MARC21/slim", 0, 0) < 0)
 378         {
 379             yaz_log(YLOG_WARN, "Failed to encode as XML");
 380             return 0;
 381         }
 382         rdoc = xmlNewDoc("1.0");
 383         xmlDocSetRootElement(rdoc, res);
 384     }
 385     else
 386     {
 387         yaz_log(YLOG_FATAL, "Unknown native_syntax in normalize_record");
 388         exit(1);
 389     }
 390     for (m = rprofile->maplist; m; m = m->next)
 391     {
 392         xmlDoc *new;
 393         if (m->type != Map_xslt)
 394         {
 395             yaz_log(YLOG_WARN, "Unknown map type");
 396             return 0;
 397         }
 398         if (!(new = xsltApplyStylesheet(m->stylesheet, rdoc, 0)))
 399         {
 400             yaz_log(YLOG_WARN, "XSLT transformation failed");
 401             return 0;
 402         }
 403         xmlFreeDoc(rdoc);
 404         rdoc = new;
 405     }
 406     if (global_parameters.dump_records)
 407     {
 408         fprintf(stderr, "Record:\n----------------\n");
 409         xmlDocFormatDump(stderr, rdoc, 1);
 410     }
 411     return rdoc;
 412 }
 413
 414 static struct record *ingest_record(struct client *cl, Z_External *rec)
 415 {
 416     xmlDoc *xdoc = normalize_record(cl, rec);
 417     xmlNode *root, *n;
 418     struct record *res, *head;
 419     struct session *se = cl->session;
 420     xmlChar *mergekey, *mergekey_norm;
 421
 422     if (!xdoc)
 423         return 0;
 424
 425     root = xmlDocGetRootElement(xdoc);
 426     if (!(mergekey = xmlGetProp(root, "mergekey")))
 427     {
 428         yaz_log(YLOG_WARN, "No mergekey found in record");
 429         return 0;
 430     }
 431
 432     res = nmem_malloc(se->nmem, sizeof(struct record));
 433     res->next_cluster = 0;
 434     res->target_offset = -1;
 435     res->term_frequency_vec = 0;
 436     res->title = "Unknown";
 437     res->relevance = 0;
 438
 439     mergekey_norm = nmem_strdup(se->nmem, (char*) mergekey);
 440     xmlFree(mergekey);
 441     res->merge_key = normalize_mergekey(mergekey_norm);
 442
 443     head = reclist_insert(se->reclist, res);
 444     relevance_newrec(se->relevance, head);
 445
 446     for (n = root->children; n; n = n->next)
 447     {
 448         if (n->type != XML_ELEMENT_NODE)
 449             continue;
 450         if (!strcmp(n->name, "facet"))
 451         {
 452             xmlChar *type = xmlGetProp(n, "type");
 453             xmlChar *value = xmlNodeListGetString(xdoc, n->children, 0);
 454             add_facet(se, type, value);
 455             relevance_countwords(se->relevance, head, value, 1);
 456             xmlFree(type);
 457             xmlFree(value);
 458         }
 459         else if (!strcmp(n->name, "metadata"))
 460         {
 461             xmlChar *type = xmlGetProp(n, "type"), *value;
 462             if (!strcmp(type, "title"))
 463                 res->title = nmem_strdup(se->nmem,
 464                         value = xmlNodeListGetString(xdoc, n->children, 0));
 465
 466             relevance_countwords(se->relevance, head, value, 4);
 467             xmlFree(type);
 468             xmlFree(value);
 469         }
 470         else
 471             yaz_log(YLOG_WARN, "Unexpected element %s in internal record", n->name);
 472     }
 473
 474     relevance_donerecord(se->relevance, head);
 475     se->total_records++;
 476
 477     return res;
 478 }
 479
 480 static void ingest_records(struct client *cl, Z_Records *r)
 481 {
 482     struct record *rec;
 483     struct session *s = cl->session;
 484     Z_NamePlusRecordList *rlist;
 485     int i;
 486
 487     if (r->which != Z_Records_DBOSD)
 488         return;
 489     rlist = r->u.databaseOrSurDiagnostics;
 490     for (i = 0; i < rlist->num_records; i++)
 491     {
 492         Z_NamePlusRecord *npr = rlist->records[i];
 493
 494         if (npr->which != Z_NamePlusRecord_databaseRecord)
 495         {
 496             yaz_log(YLOG_WARN, "Unexpected record type, probably diagnostic");
 497             continue;
 498         }
 499
 500         rec = ingest_record(cl, npr->u.databaseRecord);
 501         if (!rec)
 502             continue;
 503     }
 504     if (s->watchlist[SESSION_WATCH_RECORDS].fun && rlist->num_records)
 505         session_alert_watch(s, SESSION_WATCH_RECORDS);
 506 }
 507
 508 static void do_presentResponse(IOCHAN i, Z_APDU *a)
 509 {
 510     struct connection *co = iochan_getdata(i);
 511     struct client *cl = co->client;
 512     Z_PresentResponse *r = a->u.presentResponse;
 513
 514     if (r->records) {
 515         Z_Records *recs = r->records;
 516         if (recs->which == Z_Records_NSD)
 517         {
 518             yaz_log(YLOG_WARN, "Non-surrogate diagnostic");
 519             cl->diagnostic = *recs->u.nonSurrogateDiagnostic->condition;
 520             cl->state = Client_Error;
 521         }
 522     }
 523
 524     if (!*r->presentStatus && cl->state != Client_Error)
 525     {
 526         yaz_log(YLOG_DEBUG, "Good Present response");
 527         cl->records += *r->numberOfRecordsReturned;
 528         ingest_records(cl, r->records);
 529         cl->state = Client_Idle;
 530     }
 531     else if (*r->presentStatus)
 532     {
 533         yaz_log(YLOG_WARN, "Bad Present response");
 534         cl->state = Client_Error;
 535     }
 536 }
 537
 538 static void handler(IOCHAN i, int event)
 539 {
 540     struct connection *co = iochan_getdata(i);
 541     struct client *cl = co->client;
 542     struct session *se = 0;
 543
 544     if (cl)
 545         se = cl->session;
 546     else
 547     {
 548         yaz_log(YLOG_WARN, "Destroying orphan connection");
 549         connection_destroy(co);
 550         return;
 551     }
 552
 553     if (co->state == Conn_Connecting && event & EVENT_OUTPUT)
 554     {
 555         int errcode;
 556         socklen_t errlen = sizeof(errcode);
 557
 558         if (getsockopt(cs_fileno(co->link), SOL_SOCKET, SO_ERROR, &errcode,
 559             &errlen) < 0 || errcode != 0)
 560         {
 561             client_fatal(cl);
 562             return;
 563         }
 564         else
 565         {
 566             yaz_log(YLOG_DEBUG, "Connect OK");
 567             co->state = Conn_Open;
 568             if (cl)
 569                 cl->state = Client_Connected;
 570         }
 571     }
 572
 573     else if (event & EVENT_INPUT)
 574     {
 575         int len = cs_get(co->link, &co->ibuf, &co->ibufsize);
 576
 577         if (len < 0)
 578         {
 579             yaz_log(YLOG_WARN|YLOG_ERRNO, "Error reading from Z server");
 580             connection_destroy(co);
 581             return;
 582         }
 583         else if (len == 0)
 584         {
 585             yaz_log(YLOG_WARN, "EOF reading from Z server");
 586             connection_destroy(co);
 587             return;
 588         }
 589         else if (len > 1) // We discard input if we have no connection
 590         {
 591             co->state = Conn_Open;
 592
 593             if (cl && (cl->requestid == se->requestid || cl->state == Client_Initializing))
 594             {
 595                 Z_APDU *a;
 596
 597                 odr_reset(global_parameters.odr_in);
 598                 odr_setbuf(global_parameters.odr_in, co->ibuf, len, 0);
 599                 if (!z_APDU(global_parameters.odr_in, &a, 0, 0))
 600                 {
 601                     client_fatal(cl);
 602                     return;
 603                 }
 604                 switch (a->which)
 605                 {
 606                     case Z_APDU_initResponse:
 607                         do_initResponse(i, a);
 608                         break;
 609                     case Z_APDU_searchResponse:
 610                         do_searchResponse(i, a);
 611                         break;
 612                     case Z_APDU_presentResponse:
 613                         do_presentResponse(i, a);
 614                         break;
 615                     default:
 616                         yaz_log(YLOG_WARN, "Unexpected result from server");
 617                         client_fatal(cl);
 618                         return;
 619                 }
 620                 // We aren't expecting staggered output from target
 621                 // if (cs_more(t->link))
 622                 //    iochan_setevent(i, EVENT_INPUT);
 623             }
 624             else  // we throw away response and go to idle mode
 625             {
 626                 yaz_log(YLOG_DEBUG, "Ignoring result of expired operation");
 627                 cl->state = Client_Idle;
 628             }
 629         }
 630         /* if len==1 we do nothing but wait for more input */
 631     }
 632
 633     if (cl->state == Client_Connected) {
 634         send_init(i);
 635     }
 636
 637     if (cl->state == Client_Idle)
 638     {
 639         if (cl->requestid != se->requestid && *se->query) {
 640             send_search(i);
 641         }
 642         else if (cl->hits > 0 && cl->records < global_parameters.toget &&
 643             cl->records < cl->hits) {
 644             send_present(i);
 645         }
 646     }
 647 }
 648
 649 // Disassociate connection from client
 650 static void connection_release(struct connection *co)
 651 {
 652     struct client *cl = co->client;
 653
 654     yaz_log(YLOG_DEBUG, "Connection release %s", co->host->hostport);
 655     if (!cl)
 656         return;
 657     cl->connection = 0;
 658     co->client = 0;
 659 }
 660
 661 // Close connection and recycle structure
 662 static void connection_destroy(struct connection *co)
 663 {
 664     struct host *h = co->host;
 665     cs_close(co->link);
 666     iochan_destroy(co->iochan);
 667
 668     yaz_log(YLOG_DEBUG, "Connection destroy %s", co->host->hostport);
 669     if (h->connections == co)
 670         h->connections = co->next;
 671     else
 672     {
 673         struct connection *pco;
 674         for (pco = h->connections; pco && pco->next != co; pco = pco->next)
 675             ;
 676         if (pco)
 677             pco->next = co->next;
 678         else
 679             abort();
 680     }
 681     if (co->client)
 682     {
 683         if (co->client->state != Client_Idle)
 684             co->client->state = Client_Disconnected;
 685         co->client->connection = 0;
 686     }
 687     co->next = connection_freelist;
 688     connection_freelist = co;
 689 }
 690
 691 // Creates a new connection for client, associated with the host of
 692 // client's database
 693 static struct connection *connection_create(struct client *cl)
 694 {
 695     struct connection *new;
 696     COMSTACK link;
 697     int res;
 698     void *addr;
 699
 700     yaz_log(YLOG_DEBUG, "Connection create %s", cl->database->url);
 701     if (!(link = cs_create(tcpip_type, 0, PROTO_Z3950)))
 702     {
 703         yaz_log(YLOG_FATAL|YLOG_ERRNO, "Failed to create comstack");
 704         exit(1);
 705     }
 706
 707     if (!(addr = cs_straddr(link, cl->database->host->ipport)))
 708     {
 709         yaz_log(YLOG_WARN|YLOG_ERRNO, "Lookup of IP address failed?");
 710         return 0;
 711     }
 712
 713     res = cs_connect(link, addr);
 714     if (res < 0)
 715     {
 716         yaz_log(YLOG_WARN|YLOG_ERRNO, "cs_connect %s", cl->database->url);
 717         return 0;
 718     }
 719
 720     if ((new = connection_freelist))
 721         connection_freelist = new->next;
 722     else
 723     {
 724         new = xmalloc(sizeof (struct connection));
 725         new->ibuf = 0;
 726         new->ibufsize = 0;
 727     }
 728     new->state = Conn_Connecting;
 729     new->host = cl->database->host;
 730     new->next = new->host->connections;
 731     new->host->connections = new;
 732     new->client = cl;
 733     cl->connection = new;
 734     new->link = link;
 735
 736     new->iochan = iochan_create(cs_fileno(link), handler, 0);
 737     iochan_setdata(new->iochan, new);
 738     new->iochan->next = channel_list;
 739     channel_list = new->iochan;
 740     return new;
 741 }
 742
 743 // Close connection and set state to error
 744 static void client_fatal(struct client *cl)
 745 {
 746     yaz_log(YLOG_WARN, "Fatal error from %s", cl->database->url);
 747     connection_destroy(cl->connection);
 748     cl->state = Client_Error;
 749 }
 750
 751 // Ensure that client has a connection associated
 752 static int client_prep_connection(struct client *cl)
 753 {
 754     struct connection *co;
 755     struct session *se = cl->session;
 756     struct host *host = cl->database->host;
 757
 758     co = cl->connection;
 759
 760     yaz_log(YLOG_DEBUG, "Client prep %s", cl->database->url);
 761
 762     if (!co)
 763     {
 764         // See if someone else has an idle connection
 765         // We should look at timestamps here to select the longest-idle connection
 766         for (co = host->connections; co; co = co->next)
 767             if (co->state == Conn_Open && (!co->client || co->client->session != se))
 768                 break;
 769         if (co)
 770         {
 771             connection_release(co);
 772             cl->connection = co;
 773             co->client = cl;
 774         }
 775         else
 776             co = connection_create(cl);
 777     }
 778     if (co)
 779     {
 780         if (co->state == Conn_Connecting)
 781             cl->state = Client_Connecting;
 782         else if (co->state == Conn_Open)
 783         {
 784             if (cl->state == Client_Error || cl->state == Client_Disconnected)
 785                 cl->state = Client_Idle;
 786         }
 787         iochan_setflag(co->iochan, EVENT_OUTPUT);
 788         return 1;
 789     }
 790     else
 791         return 0;
 792 }
 793
 794 // This function will most likely vanish when a proper target profile mechanism is
 795 // introduced.
 796 void load_simpletargets(const char *fn)
 797 {
 798     FILE *f = fopen(fn, "r");
 799     char line[256];
 800
 801     if (!f)
 802     {
 803         yaz_log(YLOG_WARN|YLOG_ERRNO, "open %s", fn);
 804         exit(1);
 805     }
 806
 807     while (fgets(line, 255, f))
 808     {
 809         char *url, *db;
 810         struct host *host;
 811         struct database *database;
 812
 813         if (strncmp(line, "target ", 7))
 814             continue;
 815         url = line + 7;
 816         url[strlen(url) - 1] = '\0';
 817         yaz_log(YLOG_DEBUG, "Target: %s", url);
 818         if ((db = strchr(url, '/')))
 819             *(db++) = '\0';
 820         else
 821             db = "Default";
 822
 823         for (host = hosts; host; host = host->next)
 824             if (!strcmp(url, host->hostport))
 825                 break;
 826         if (!host)
 827         {
 828             struct addrinfo *addrinfo, hints;
 829             char *port;
 830             char ipport[128];
 831             unsigned char addrbuf[4];
 832             int res;
 833
 834             host = xmalloc(sizeof(struct host));
 835             host->hostport = xstrdup(url);
 836             host->connections = 0;
 837
 838             if ((port = strchr(url, ':')))
 839                 *(port++) = '\0';
 840             else
 841                 port = "210";
 842
 843             hints.ai_flags = 0;
 844             hints.ai_family = PF_INET;
 845             hints.ai_socktype = SOCK_STREAM;
 846             hints.ai_protocol = IPPROTO_TCP;
 847             hints.ai_addrlen = 0;
 848             hints.ai_addr = 0;
 849             hints.ai_canonname = 0;
 850             hints.ai_next = 0;
 851             // This is not robust code. It assumes that getaddrinfo returns AF_INET
 852             // address.
 853             if ((res = getaddrinfo(url, port, &hints, &addrinfo)))
 854             {
 855                 yaz_log(YLOG_WARN, "Failed to resolve %s: %s", url, gai_strerror(res));
 856                 xfree(host->hostport);
 857                 xfree(host);
 858                 continue;
 859             }
 860             assert(addrinfo->ai_family == PF_INET);
 861             memcpy(addrbuf, &((struct sockaddr_in*)addrinfo->ai_addr)->sin_addr.s_addr, 4);
 862             sprintf(ipport, "%hhd.%hhd.%hhd.%hhd:%s",
 863                     addrbuf[0], addrbuf[1], addrbuf[2], addrbuf[3], port);
 864             host->ipport = xstrdup(ipport);
 865             freeaddrinfo(addrinfo);
 866             host->next = hosts;
 867             hosts = host;
 868         }
 869         database = xmalloc(sizeof(struct database));
 870         database->host = host;
 871         database->url = xmalloc(strlen(url) + strlen(db) + 2);
 872         strcpy(database->url, url);
 873         strcat(database->url, "/");
 874         strcat(database->url, db);
 875
 876         database->databases = xmalloc(2 * sizeof(char *));
 877         database->databases[0] = xstrdup(db);
 878         database->databases[1] = 0;
 879         database->errors = 0;
 880         database->qprofile = 0;
 881         database->rprofile = database_retrieval_profile(database);
 882         database->next = databases;
 883         databases = database;
 884
 885     }
 886     fclose(f);
 887 }
 888
 889 static void pull_terms(NMEM nmem, struct ccl_rpn_node *n, char **termlist, int *num)
 890 {
 891     switch (n->kind)
 892     {
 893         case CCL_RPN_AND:
 894         case CCL_RPN_OR:
 895         case CCL_RPN_NOT:
 896         case CCL_RPN_PROX:
 897             pull_terms(nmem, n->u.p[0], termlist, num);
 898             pull_terms(nmem, n->u.p[1], termlist, num);
 899             break;
 900         case CCL_RPN_TERM:
 901             termlist[(*num)++] = nmem_strdup(nmem, n->u.t.term);
 902             break;
 903         default: // NOOP
 904             break;
 905     }
 906 }
 907
 908 // Extract terms from query into null-terminated termlist
 909 static int extract_terms(NMEM nmem, char *query, char **termlist)
 910 {
 911     int error, pos;
 912     struct ccl_rpn_node *n;
 913     int num = 0;
 914
 915     n = ccl_find_str(global_parameters.ccl_filter, query, &error, &pos);
 916     if (!n)
 917         return -1;
 918     pull_terms(nmem, n, termlist, &num);
 919     termlist[num] = 0;
 920     ccl_rpn_delete(n);
 921     return 0;
 922 }
 923
 924 static struct client *client_create(void)
 925 {
 926     struct client *r;
 927     if (client_freelist)
 928     {
 929         r = client_freelist;
 930         client_freelist = client_freelist->next;
 931     }
 932     else
 933         r = xmalloc(sizeof(struct client));
 934     r->database = 0;
 935     r->connection = 0;
 936     r->session = 0;
 937     r->hits = 0;
 938     r->records = 0;
 939     r->setno = 0;
 940     r->requestid = -1;
 941     r->diagnostic = 0;
 942     r->state = Client_Disconnected;
 943     r->next = 0;
 944     return r;
 945 }
 946
 947 void client_destroy(struct client *c)
 948 {
 949     struct session *se = c->session;
 950     if (c == se->clients)
 951         se->clients = c->next;
 952     else
 953     {
 954         struct client *cc;
 955         for (cc = se->clients; cc && cc->next != c; cc = cc->next)
 956             ;
 957         if (cc)
 958             cc->next = c->next;
 959     }
 960     if (c->connection)
 961         connection_release(c->connection);
 962     c->next = client_freelist;
 963     client_freelist = c;
 964 }
 965
 966 void session_set_watch(struct session *s, int what, session_watchfun fun, void *data)
 967 {
 968     s->watchlist[what].fun = fun;
 969     s->watchlist[what].data = data;
 970 }
 971
 972 void session_alert_watch(struct session *s, int what)
 973 {
 974     if (!s->watchlist[what].fun)
 975         return;
 976     (*s->watchlist[what].fun)(s->watchlist[what].data);
 977     s->watchlist[what].fun = 0;
 978     s->watchlist[what].data = 0;
 979 }
 980
 981 // This needs to be extended with selection criteria
 982 static struct conf_retrievalprofile *database_retrieval_profile(struct database *db)
 983 {
 984     if (!config)
 985     {
 986         yaz_log(YLOG_FATAL, "Must load configuration (-f)");
 987         exit(1);
 988     }
 989     if (!config->retrievalprofiles)
 990     {
 991         yaz_log(YLOG_FATAL, "No retrieval profiles defined");
 992     }
 993     return config->retrievalprofiles;
 994 }
 995
 996 // This should be extended with parameters to control selection criteria
 997 // Associates a set of clients with a session;
 998 int select_targets(struct session *se)
 999 {
1000     struct database *db;
1001     int c = 0;
1002
1003     while (se->clients)
1004         client_destroy(se->clients);
1005     for (db = databases; db; db = db->next)
1006     {
1007         struct client *cl = client_create();
1008         cl->database = db;
1009         cl->session = se;
1010         cl->next = se->clients;
1011         se->clients = cl;
1012         c++;
1013     }
1014     return c;
1015 }
1016
1017 char *search(struct session *se, char *query)
1018 {
1019     int live_channels = 0;
1020     struct client *cl;
1021
1022     yaz_log(YLOG_DEBUG, "Search");
1023
1024     strcpy(se->query, query);
1025     se->requestid++;
1026     nmem_reset(se->nmem);
1027     for (cl = se->clients; cl; cl = cl->next)
1028     {
1029         cl->hits = -1;
1030         cl->records = 0;
1031         cl->diagnostic = 0;
1032
1033         if (client_prep_connection(cl))
1034             live_channels++;
1035     }
1036     if (live_channels)
1037     {
1038         char *p[512];
1039         int maxrecs = live_channels * global_parameters.toget;
1040         se->num_termlists = 0;
1041         se->reclist = reclist_create(se->nmem, maxrecs);
1042         extract_terms(se->nmem, query, p);
1043         se->relevance = relevance_create(se->nmem, (const char **) p, maxrecs);
1044         se->total_records = se->total_hits = 0;
1045         se->expected_maxrecs = maxrecs;
1046     }
1047     else
1048         return "NOTARGETS";
1049
1050     return 0;
1051 }
1052
1053 void destroy_session(struct session *s)
1054 {
1055     yaz_log(YLOG_LOG, "Destroying session");
1056     while (s->clients)
1057         client_destroy(s->clients);
1058     nmem_destroy(s->nmem);
1059     wrbuf_free(s->wrbuf, 1);
1060 }
1061
1062 struct session *new_session()
1063 {
1064     int i;
1065     struct session *session = xmalloc(sizeof(*session));
1066
1067     yaz_log(YLOG_DEBUG, "New pazpar2 session");
1068
1069     session->total_hits = 0;
1070     session->total_records = 0;
1071     session->num_termlists = 0;
1072     session->reclist = 0;
1073     session->requestid = -1;
1074     session->clients = 0;
1075     session->expected_maxrecs = 0;
1076     session->query[0] = '\0';
1077     session->nmem = nmem_create();
1078     session->wrbuf = wrbuf_alloc();
1079     for (i = 0; i <= SESSION_WATCH_MAX; i++)
1080     {
1081         session->watchlist[i].data = 0;
1082         session->watchlist[i].fun = 0;
1083     }
1084
1085     select_targets(session);
1086
1087     return session;
1088 }
1089
1090 struct hitsbytarget *hitsbytarget(struct session *se, int *count)
1091 {
1092     static struct hitsbytarget res[1000]; // FIXME MM
1093     struct client *cl;
1094
1095     *count = 0;
1096     for (cl = se->clients; cl; cl = cl->next)
1097     {
1098         strcpy(res[*count].id, cl->database->host->hostport);
1099         res[*count].hits = cl->hits;
1100         res[*count].records = cl->records;
1101         res[*count].diagnostic = cl->diagnostic;
1102         res[*count].state = client_states[cl->state];
1103         res[*count].connected  = cl->connection ? 1 : 0;
1104         (*count)++;
1105     }
1106
1107     return res;
1108 }
1109
1110 struct termlist_score **termlist(struct session *s, const char *name, int *num)
1111 {
1112     int i;
1113
1114     for (i = 0; i < s->num_termlists; i++)
1115         if (!strcmp(s->termlists[i].name, name))
1116             return termlist_highscore(s->termlists[i].termlist, num);
1117     return 0;
1118 }
1119
1120 #ifdef REPORT_NMEM
1121 // conditional compilation by SH: This lead to a warning with currently installed
1122 // YAZ header files on us1
1123 void report_nmem_stats(void)
1124 {
1125     size_t in_use, is_free;
1126
1127     nmem_get_memory_in_use(&in_use);
1128     nmem_get_memory_free(&is_free);
1129
1130     yaz_log(YLOG_LOG, "nmem stat: use=%ld free=%ld",
1131             (long) in_use, (long) is_free);
1132 }
1133 #endif
1134
1135 struct record **show(struct session *s, int start, int *num, int *total,
1136                      int *sumhits, NMEM nmem_show)
1137 {
1138     struct record **recs = nmem_malloc(nmem_show, *num
1139                                        * sizeof(struct record *));
1140     int i;
1141
1142     relevance_prepare_read(s->relevance, s->reclist);
1143
1144     *total = s->reclist->num_records;
1145     *sumhits = s->total_hits;
1146
1147     for (i = 0; i < start; i++)
1148         if (!reclist_read_record(s->reclist))
1149         {
1150             *num = 0;
1151             return 0;
1152         }
1153
1154     for (i = 0; i < *num; i++)
1155     {
1156         struct record *r = reclist_read_record(s->reclist);
1157         if (!r)
1158         {
1159             *num = i;
1160             break;
1161         }
1162         recs[i] = r;
1163     }
1164     return recs;
1165 }
1166
1167 void statistics(struct session *se, struct statistics *stat)
1168 {
1169     struct client *cl;
1170     int count = 0;
1171
1172     bzero(stat, sizeof(*stat));
1173     for (cl = se->clients; cl; cl = cl->next)
1174     {
1175         if (!cl->connection)
1176             stat->num_no_connection++;
1177         switch (cl->state)
1178         {
1179             case Client_Connecting: stat->num_connecting++; break;
1180             case Client_Initializing: stat->num_initializing++; break;
1181             case Client_Searching: stat->num_searching++; break;
1182             case Client_Presenting: stat->num_presenting++; break;
1183             case Client_Idle: stat->num_idle++; break;
1184             case Client_Failed: stat->num_failed++; break;
1185             case Client_Error: stat->num_error++; break;
1186             default: break;
1187         }
1188         count++;
1189     }
1190     stat->num_hits = se->total_hits;
1191     stat->num_records = se->total_records;
1192
1193     stat->num_clients = count;
1194 }
1195
1196 static CCL_bibset load_cclfile(const char *fn)
1197 {
1198     CCL_bibset res = ccl_qual_mk();
1199     if (ccl_qual_fname(res, fn) < 0)
1200     {
1201         yaz_log(YLOG_FATAL|YLOG_ERRNO, "%s", fn);
1202         exit(1);
1203     }
1204     return res;
1205 }
1206
1207 int main(int argc, char **argv)
1208 {
1209     int ret;
1210     char *arg;
1211     int setport = 0;
1212
1213     if (signal(SIGPIPE, SIG_IGN) < 0)
1214         yaz_log(YLOG_WARN|YLOG_ERRNO, "signal");
1215
1216     yaz_log_init(YLOG_DEFAULT_LEVEL, "pazpar2", 0);
1217
1218     while ((ret = options("f:x:c:h:p:C:s:d", argv, argc, &arg)) != -2)
1219     {
1220         switch (ret) {
1221             case 'f':
1222                 if (!read_config(arg))
1223                     exit(1);
1224                 break;
1225             case 'c':
1226                 command_init(atoi(arg));
1227                 setport++;
1228                 break;
1229             case 'h':
1230                 http_init(arg);
1231                 setport++;
1232                 break;
1233             case 'C':
1234                 global_parameters.ccl_filter = load_cclfile(arg);
1235                 break;
1236             case 'p':
1237                 http_set_proxyaddr(arg);
1238                 break;
1239             case 's':
1240                 load_simpletargets(arg);
1241                 break;
1242             case 'd':
1243                 global_parameters.dump_records = 1;
1244                 break;
1245             default:
1246                 fprintf(stderr, "Usage: pazpar2\n"
1247                         "    -f configfile\n"
1248                         "    -h [host:]port          (REST protocol listener)\n"
1249                         "    -c cmdport              (telnet-style)\n"
1250                         "    -C cclconfig\n"
1251                         "    -s simpletargetfile\n"
1252                         "    -p hostname[:portno]    (HTTP proxy)\n");
1253                 exit(1);
1254         }
1255     }
1256
1257     if (!setport)
1258     {
1259         fprintf(stderr, "Set command port with -h or -c\n");
1260         exit(1);
1261     }
1262
1263     global_parameters.ccl_filter = load_cclfile("../etc/default.bib");
1264     global_parameters.yaz_marc = yaz_marc_create();
1265     yaz_marc_subfield_str(global_parameters.yaz_marc, "\t");
1266     global_parameters.odr_in = odr_createmem(ODR_DECODE);
1267     global_parameters.odr_out = odr_createmem(ODR_ENCODE);
1268
1269     event_loop(&channel_list);
1270
1271     return 0;
1272 }
1273
1274 /*
1275  * Local variables:
1276  * c-basic-offset: 4
1277  * indent-tabs-mode: nil
1278  * End:
1279  * vim: shiftwidth=4 tabstop=8 expandtab
1280  */