src/pazpar2.c

   1 /* $Id: pazpar2.c,v 1.18 2007-01-08 12:43:41 adam 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/yaz-util.h>
  21 #include <yaz/nmem.h>
  22
  23 #if HAVE_CONFIG_H
  24 #include "cconfig.h"
  25 #endif
  26
  27 #define USE_TIMING 0
  28 #if USE_TIMING
  29 #include <yaz/timing.h>
  30 #endif
  31
  32 #include "pazpar2.h"
  33 #include "eventl.h"
  34 #include "http.h"
  35 #include "termlists.h"
  36 #include "reclists.h"
  37 #include "relevance.h"
  38 #include "config.h"
  39
  40 #define MAX_CHUNK 15
  41
  42 static void client_fatal(struct client *cl);
  43 static void connection_destroy(struct connection *co);
  44 static int client_prep_connection(struct client *cl);
  45 static void ingest_records(struct client *cl, Z_Records *r);
  46 static struct conf_retrievalprofile *database_retrieval_profile(struct database *db);
  47 void session_alert_watch(struct session *s, int what);
  48
  49 IOCHAN channel_list = 0;  // Master list of connections we're handling events to
  50
  51 static struct connection *connection_freelist = 0;
  52 static struct client *client_freelist = 0;
  53
  54 static struct host *hosts = 0;  // The hosts we know about
  55 static struct database *databases = 0; // The databases we know about
  56
  57 static char *client_states[] = {
  58     "Client_Connecting",
  59     "Client_Connected",
  60     "Client_Idle",
  61     "Client_Initializing",
  62     "Client_Searching",
  63     "Client_Presenting",
  64     "Client_Error",
  65     "Client_Failed",
  66     "Client_Disconnected",
  67     "Client_Stopped"
  68 };
  69
  70 struct parameters global_parameters =
  71 {
  72     0,
  73     30,
  74     "81",
  75     "Index Data PazPar2 (MasterKey)",
  76     VERSION,
  77     600, // 10 minutes
  78     60,
  79     100,
  80     MAX_CHUNK,
  81     0,
  82     0,
  83     0,
  84     0
  85 };
  86
  87 static int send_apdu(struct client *c, Z_APDU *a)
  88 {
  89     struct connection *co = c->connection;
  90     char *buf;
  91     int len, r;
  92
  93     if (!z_APDU(global_parameters.odr_out, &a, 0, 0))
  94     {
  95         odr_perror(global_parameters.odr_out, "Encoding APDU");
  96         abort();
  97     }
  98     buf = odr_getbuf(global_parameters.odr_out, &len, 0);
  99     r = cs_put(co->link, buf, len);
 100     if (r < 0)
 101     {
 102         yaz_log(YLOG_WARN, "cs_put: %s", cs_errmsg(cs_errno(co->link)));
 103         return -1;
 104     }
 105     else if (r == 1)
 106     {
 107         fprintf(stderr, "cs_put incomplete (ParaZ does not handle that)\n");
 108         exit(1);
 109     }
 110     odr_reset(global_parameters.odr_out); /* release the APDU structure  */
 111     co->state = Conn_Waiting;
 112     return 0;
 113 }
 114
 115
 116 static void send_init(IOCHAN i)
 117 {
 118     struct connection *co = iochan_getdata(i);
 119     struct client *cl = co->client;
 120     Z_APDU *a = zget_APDU(global_parameters.odr_out, Z_APDU_initRequest);
 121
 122     a->u.initRequest->implementationId = global_parameters.implementationId;
 123     a->u.initRequest->implementationName = global_parameters.implementationName;
 124     a->u.initRequest->implementationVersion =
 125         global_parameters.implementationVersion;
 126     ODR_MASK_SET(a->u.initRequest->options, Z_Options_search);
 127     ODR_MASK_SET(a->u.initRequest->options, Z_Options_present);
 128     ODR_MASK_SET(a->u.initRequest->options, Z_Options_namedResultSets);
 129
 130     ODR_MASK_SET(a->u.initRequest->protocolVersion, Z_ProtocolVersion_1);
 131     ODR_MASK_SET(a->u.initRequest->protocolVersion, Z_ProtocolVersion_2);
 132     ODR_MASK_SET(a->u.initRequest->protocolVersion, Z_ProtocolVersion_3);
 133     if (send_apdu(cl, a) >= 0)
 134     {
 135         iochan_setflags(i, EVENT_INPUT);
 136         cl->state = Client_Initializing;
 137     }
 138     else
 139         cl->state = Client_Error;
 140     odr_reset(global_parameters.odr_out);
 141 }
 142
 143 static void send_search(IOCHAN i)
 144 {
 145     struct connection *co = iochan_getdata(i);
 146     struct client *cl = co->client;
 147     struct session *se = cl->session;
 148     struct database *db = cl->database;
 149     Z_APDU *a = zget_APDU(global_parameters.odr_out, Z_APDU_searchRequest);
 150     int ndb, cerror, cpos;
 151     char **databaselist;
 152     Z_Query *zquery;
 153     struct ccl_rpn_node *cn;
 154     int ssub = 0, lslb = 100000, mspn = 10;
 155
 156     yaz_log(YLOG_DEBUG, "Sending search");
 157
 158     cn = ccl_find_str(global_parameters.ccl_filter, se->query, &cerror, &cpos);
 159     if (!cn)
 160         return;
 161     a->u.searchRequest->query = zquery = odr_malloc(global_parameters.odr_out,
 162             sizeof(Z_Query));
 163     zquery->which = Z_Query_type_1;
 164     zquery->u.type_1 = ccl_rpn_query(global_parameters.odr_out, cn);
 165     ccl_rpn_delete(cn);
 166
 167     for (ndb = 0; db->databases[ndb]; ndb++)
 168         ;
 169     databaselist = odr_malloc(global_parameters.odr_out, sizeof(char*) * ndb);
 170     for (ndb = 0; db->databases[ndb]; ndb++)
 171         databaselist[ndb] = db->databases[ndb];
 172
 173     a->u.presentRequest->preferredRecordSyntax =
 174             yaz_oidval_to_z3950oid(global_parameters.odr_out,
 175             CLASS_RECSYN, VAL_USMARC);
 176     a->u.searchRequest->smallSetUpperBound = &ssub;
 177     a->u.searchRequest->largeSetLowerBound = &lslb;
 178     a->u.searchRequest->mediumSetPresentNumber = &mspn;
 179     a->u.searchRequest->resultSetName = "Default";
 180     a->u.searchRequest->databaseNames = databaselist;
 181     a->u.searchRequest->num_databaseNames = ndb;
 182
 183     if (send_apdu(cl, a) >= 0)
 184     {
 185         iochan_setflags(i, EVENT_INPUT);
 186         cl->state = Client_Searching;
 187         cl->requestid = se->requestid;
 188     }
 189     else
 190         cl->state = Client_Error;
 191
 192     odr_reset(global_parameters.odr_out);
 193 }
 194
 195 static void send_present(IOCHAN i)
 196 {
 197     struct connection *co = iochan_getdata(i);
 198     struct client *cl = co->client;
 199     Z_APDU *a = zget_APDU(global_parameters.odr_out, Z_APDU_presentRequest);
 200     int toget;
 201     int start = cl->records + 1;
 202
 203     toget = global_parameters.chunk;
 204     if (toget > global_parameters.toget - cl->records)
 205         toget = global_parameters.toget - cl->records;
 206     if (toget > cl->hits - cl->records)
 207         toget = cl->hits - cl->records;
 208
 209     yaz_log(YLOG_DEBUG, "Trying to present %d records\n", toget);
 210
 211     a->u.presentRequest->resultSetStartPoint = &start;
 212     a->u.presentRequest->numberOfRecordsRequested = &toget;
 213
 214     a->u.presentRequest->resultSetId = "Default";
 215
 216     a->u.presentRequest->preferredRecordSyntax =
 217             yaz_oidval_to_z3950oid(global_parameters.odr_out,
 218             CLASS_RECSYN, VAL_USMARC);
 219
 220     if (send_apdu(cl, a) >= 0)
 221     {
 222         iochan_setflags(i, EVENT_INPUT);
 223         cl->state = Client_Presenting;
 224     }
 225     else
 226         cl->state = Client_Error;
 227     odr_reset(global_parameters.odr_out);
 228 }
 229
 230 static void do_initResponse(IOCHAN i, Z_APDU *a)
 231 {
 232     struct connection *co = iochan_getdata(i);
 233     struct client *cl = co->client;
 234     Z_InitResponse *r = a->u.initResponse;
 235
 236     yaz_log(YLOG_DEBUG, "Received init response");
 237
 238     if (*r->result)
 239     {
 240         cl->state = Client_Idle;
 241     }
 242     else
 243         cl->state = Client_Failed; // FIXME need to do something to the connection
 244 }
 245
 246 static void do_searchResponse(IOCHAN i, Z_APDU *a)
 247 {
 248     struct connection *co = iochan_getdata(i);
 249     struct client *cl = co->client;
 250     struct session *se = cl->session;
 251     Z_SearchResponse *r = a->u.searchResponse;
 252
 253     yaz_log(YLOG_DEBUG, "Searchresponse (status=%d)", *r->searchStatus);
 254
 255     if (*r->searchStatus)
 256     {
 257         cl->hits = *r->resultCount;
 258         se->total_hits += cl->hits;
 259         if (r->presentStatus && !*r->presentStatus && r->records)
 260         {
 261             yaz_log(YLOG_DEBUG, "Records in search response");
 262             cl->records += *r->numberOfRecordsReturned;
 263             ingest_records(cl, r->records);
 264         }
 265         cl->state = Client_Idle;
 266     }
 267     else
 268     {          /*"FAILED"*/
 269         cl->hits = 0;
 270         cl->state = Client_Error;
 271         if (r->records) {
 272             Z_Records *recs = r->records;
 273             if (recs->which == Z_Records_NSD)
 274             {
 275                 yaz_log(YLOG_WARN, "Non-surrogate diagnostic");
 276                 cl->diagnostic = *recs->u.nonSurrogateDiagnostic->condition;
 277                 cl->state = Client_Error;
 278             }
 279         }
 280     }
 281 }
 282
 283 char *normalize_mergekey(char *buf)
 284 {
 285     char *p = buf, *pout = buf;
 286
 287     while (*p)
 288     {
 289         while (*p && !isalnum(*p))
 290             p++;
 291         while (isalnum(*p))
 292             *(pout++) = tolower(*(p++));
 293         if (*p)
 294             *(pout++) = ' ';
 295         while (*p && !isalnum(*p))
 296             p++;
 297     }
 298     if (buf != pout)
 299         *pout = '\0';
 300
 301     return buf;
 302 }
 303
 304
 305 #ifdef GAGA
 306 // FIXME needs to be generalized. Should flexibly generate X lists per search
 307 static void extract_subject(struct session *s, const char *rec)
 308 {
 309     const char *field, *subfield;
 310
 311     while ((field = find_field(rec, "650")))
 312     {
 313         rec = field;
 314         if ((subfield = find_subfield(field, 'a')))
 315         {
 316             char *e, *ef;
 317             char buf[1024];
 318             int len;
 319
 320             ef = index(subfield, '\n');
 321             if (!ef)
 322                 return;
 323             if ((e = index(subfield, '\t')) && e < ef)
 324                 ef = e;
 325             while (ef > subfield && !isalpha(*(ef - 1)) && *(ef - 1) != ')')
 326                 ef--;
 327             len = ef - subfield;
 328             assert(len < 1023);
 329             memcpy(buf, subfield, len);
 330             buf[len] = '\0';
 331 #ifdef FIXME
 332             if (*buf)
 333                 termlist_insert(s->termlist, buf);
 334 #endif
 335         }
 336     }
 337 }
 338 #endif
 339
 340 static void add_facet(struct session *s, const char *type, const char *value)
 341 {
 342     int i;
 343
 344     for (i = 0; i < s->num_termlists; i++)
 345         if (!strcmp(s->termlists[i].name, type))
 346             break;
 347     if (i == s->num_termlists)
 348     {
 349         if (i == SESSION_MAX_TERMLISTS)
 350         {
 351             yaz_log(YLOG_FATAL, "Too many termlists");
 352             exit(1);
 353         }
 354         s->termlists[i].name = nmem_strdup(s->nmem, type);
 355         s->termlists[i].termlist = termlist_create(s->nmem, s->expected_maxrecs, 15);
 356         s->num_termlists = i + 1;
 357     }
 358     termlist_insert(s->termlists[i].termlist, value);
 359 }
 360
 361 static xmlDoc *normalize_record(struct client *cl, Z_External *rec)
 362 {
 363     struct conf_retrievalprofile *rprofile = cl->database->rprofile;
 364     struct conf_retrievalmap *m;
 365     xmlNode *res;
 366     xmlDoc *rdoc;
 367
 368     // First normalize to XML
 369     if (rprofile->native_syntax == Nativesyn_iso2709)
 370     {
 371         char *buf;
 372         int len;
 373         if (rec->which != Z_External_octet)
 374         {
 375             yaz_log(YLOG_WARN, "Unexpected external branch, probably BER");
 376             return 0;
 377         }
 378         buf = (char*) rec->u.octet_aligned->buf;
 379         len = rec->u.octet_aligned->len;
 380         if (yaz_marc_read_iso2709(rprofile->yaz_marc, buf, len) < 0)
 381         {
 382             yaz_log(YLOG_WARN, "Failed to decode MARC");
 383             return 0;
 384         }
 385         if (yaz_marc_write_xml(rprofile->yaz_marc, &res,
 386                     "http://www.loc.gov/MARC21/slim", 0, 0) < 0)
 387         {
 388             yaz_log(YLOG_WARN, "Failed to encode as XML");
 389             return 0;
 390         }
 391         rdoc = xmlNewDoc("1.0");
 392         xmlDocSetRootElement(rdoc, res);
 393     }
 394     else
 395     {
 396         yaz_log(YLOG_FATAL, "Unknown native_syntax in normalize_record");
 397         exit(1);
 398     }
 399     for (m = rprofile->maplist; m; m = m->next)
 400     {
 401         xmlDoc *new;
 402         if (m->type != Map_xslt)
 403         {
 404             yaz_log(YLOG_WARN, "Unknown map type");
 405             return 0;
 406         }
 407         if (!(new = xsltApplyStylesheet(m->stylesheet, rdoc, 0)))
 408         {
 409             yaz_log(YLOG_WARN, "XSLT transformation failed");
 410             return 0;
 411         }
 412         xmlFreeDoc(rdoc);
 413         rdoc = new;
 414     }
 415     if (global_parameters.dump_records)
 416     {
 417         fprintf(stderr, "Record:\n----------------\n");
 418         xmlDocFormatDump(stderr, rdoc, 1);
 419     }
 420     return rdoc;
 421 }
 422
 423 static struct record *ingest_record(struct client *cl, Z_External *rec)
 424 {
 425     xmlDoc *xdoc = normalize_record(cl, rec);
 426     xmlNode *root, *n;
 427     struct record *res, *head;
 428     struct session *se = cl->session;
 429     xmlChar *mergekey, *mergekey_norm;
 430
 431     if (!xdoc)
 432         return 0;
 433
 434     root = xmlDocGetRootElement(xdoc);
 435     if (!(mergekey = xmlGetProp(root, "mergekey")))
 436     {
 437         yaz_log(YLOG_WARN, "No mergekey found in record");
 438         xmlFreeDoc(xdoc);
 439         return 0;
 440     }
 441
 442     res = nmem_malloc(se->nmem, sizeof(struct record));
 443     res->next_cluster = 0;
 444     res->target_offset = -1;
 445     res->term_frequency_vec = 0;
 446     res->title = "Unknown";
 447     res->relevance = 0;
 448
 449     mergekey_norm = nmem_strdup(se->nmem, (char*) mergekey);
 450     xmlFree(mergekey);
 451     res->merge_key = normalize_mergekey(mergekey_norm);
 452
 453     head = reclist_insert(se->reclist, res);
 454     if (!head)
 455     {
 456         /* no room for record */
 457         xmlFreeDoc(xdoc);
 458         return 0;
 459     }
 460     relevance_newrec(se->relevance, head);
 461
 462     for (n = root->children; n; n = n->next)
 463     {
 464         if (n->type != XML_ELEMENT_NODE)
 465             continue;
 466         if (!strcmp(n->name, "facet"))
 467         {
 468             xmlChar *type = xmlGetProp(n, "type");
 469             xmlChar *value = xmlNodeListGetString(xdoc, n->children, 0);
 470             if (type && value)
 471             {
 472                 add_facet(se, type, value);
 473                 relevance_countwords(se->relevance, head, value, 1);
 474             }
 475             xmlFree(type);
 476             xmlFree(value);
 477         }
 478         else if (!strcmp(n->name, "metadata"))
 479         {
 480             xmlChar *type = xmlGetProp(n, "type");
 481             if (type && !strcmp(type, "title"))
 482             {
 483                 xmlChar *value = xmlNodeListGetString(xdoc, n->children, 0);
 484                 if (value)
 485                 {
 486                     res->title = nmem_strdup(se->nmem, value);
 487                     relevance_countwords(se->relevance, head, value, 4);
 488                     xmlFree(value);
 489                 }
 490             }
 491             xmlFree(type);
 492         }
 493         else
 494             yaz_log(YLOG_WARN, "Unexpected element %s in internal record", n->name);
 495     }
 496
 497     xmlFreeDoc(xdoc);
 498
 499     relevance_donerecord(se->relevance, head);
 500     se->total_records++;
 501
 502     return res;
 503 }
 504
 505 static void ingest_records(struct client *cl, Z_Records *r)
 506 {
 507 #if USE_TIMING
 508     yaz_timing_t t = yaz_timing_create();
 509 #endif
 510     struct record *rec;
 511     struct session *s = cl->session;
 512     Z_NamePlusRecordList *rlist;
 513     int i;
 514
 515     if (r->which != Z_Records_DBOSD)
 516         return;
 517     rlist = r->u.databaseOrSurDiagnostics;
 518     for (i = 0; i < rlist->num_records; i++)
 519     {
 520         Z_NamePlusRecord *npr = rlist->records[i];
 521
 522         if (npr->which != Z_NamePlusRecord_databaseRecord)
 523         {
 524             yaz_log(YLOG_WARN, "Unexpected record type, probably diagnostic");
 525             continue;
 526         }
 527
 528         rec = ingest_record(cl, npr->u.databaseRecord);
 529         if (!rec)
 530             continue;
 531     }
 532     if (s->watchlist[SESSION_WATCH_RECORDS].fun && rlist->num_records)
 533         session_alert_watch(s, SESSION_WATCH_RECORDS);
 534
 535 #if USE_TIMING
 536     yaz_timing_stop(t);
 537     yaz_log(YLOG_LOG, "ingest_records %6.5f %3.2f %3.2f",
 538             yaz_timing_get_real(t), yaz_timing_get_user(t),
 539             yaz_timing_get_sys(t));
 540     yaz_timing_destroy(&t);
 541 #endif
 542 }
 543
 544 static void do_presentResponse(IOCHAN i, Z_APDU *a)
 545 {
 546     struct connection *co = iochan_getdata(i);
 547     struct client *cl = co->client;
 548     Z_PresentResponse *r = a->u.presentResponse;
 549
 550     if (r->records) {
 551         Z_Records *recs = r->records;
 552         if (recs->which == Z_Records_NSD)
 553         {
 554             yaz_log(YLOG_WARN, "Non-surrogate diagnostic");
 555             cl->diagnostic = *recs->u.nonSurrogateDiagnostic->condition;
 556             cl->state = Client_Error;
 557         }
 558     }
 559
 560     if (!*r->presentStatus && cl->state != Client_Error)
 561     {
 562         yaz_log(YLOG_DEBUG, "Good Present response");
 563         cl->records += *r->numberOfRecordsReturned;
 564         ingest_records(cl, r->records);
 565         cl->state = Client_Idle;
 566     }
 567     else if (*r->presentStatus)
 568     {
 569         yaz_log(YLOG_WARN, "Bad Present response");
 570         cl->state = Client_Error;
 571     }
 572 }
 573
 574 static void handler(IOCHAN i, int event)
 575 {
 576     struct connection *co = iochan_getdata(i);
 577     struct client *cl = co->client;
 578     struct session *se = 0;
 579
 580     if (cl)
 581         se = cl->session;
 582     else
 583     {
 584         yaz_log(YLOG_WARN, "Destroying orphan connection");
 585         connection_destroy(co);
 586         return;
 587     }
 588
 589     if (co->state == Conn_Connecting && event & EVENT_OUTPUT)
 590     {
 591         int errcode;
 592         socklen_t errlen = sizeof(errcode);
 593
 594         if (getsockopt(cs_fileno(co->link), SOL_SOCKET, SO_ERROR, &errcode,
 595             &errlen) < 0 || errcode != 0)
 596         {
 597             client_fatal(cl);
 598             return;
 599         }
 600         else
 601         {
 602             yaz_log(YLOG_DEBUG, "Connect OK");
 603             co->state = Conn_Open;
 604             if (cl)
 605                 cl->state = Client_Connected;
 606         }
 607     }
 608
 609     else if (event & EVENT_INPUT)
 610     {
 611         int len = cs_get(co->link, &co->ibuf, &co->ibufsize);
 612
 613         if (len < 0)
 614         {
 615             yaz_log(YLOG_WARN|YLOG_ERRNO, "Error reading from Z server");
 616             connection_destroy(co);
 617             return;
 618         }
 619         else if (len == 0)
 620         {
 621             yaz_log(YLOG_WARN, "EOF reading from Z server");
 622             connection_destroy(co);
 623             return;
 624         }
 625         else if (len > 1) // We discard input if we have no connection
 626         {
 627             co->state = Conn_Open;
 628
 629             if (cl && (cl->requestid == se->requestid || cl->state == Client_Initializing))
 630             {
 631                 Z_APDU *a;
 632
 633                 odr_reset(global_parameters.odr_in);
 634                 odr_setbuf(global_parameters.odr_in, co->ibuf, len, 0);
 635                 if (!z_APDU(global_parameters.odr_in, &a, 0, 0))
 636                 {
 637                     client_fatal(cl);
 638                     return;
 639                 }
 640                 switch (a->which)
 641                 {
 642                     case Z_APDU_initResponse:
 643                         do_initResponse(i, a);
 644                         break;
 645                     case Z_APDU_searchResponse:
 646                         do_searchResponse(i, a);
 647                         break;
 648                     case Z_APDU_presentResponse:
 649                         do_presentResponse(i, a);
 650                         break;
 651                     default:
 652                         yaz_log(YLOG_WARN, "Unexpected result from server");
 653                         client_fatal(cl);
 654                         return;
 655                 }
 656                 // We aren't expecting staggered output from target
 657                 // if (cs_more(t->link))
 658                 //    iochan_setevent(i, EVENT_INPUT);
 659             }
 660             else  // we throw away response and go to idle mode
 661             {
 662                 yaz_log(YLOG_DEBUG, "Ignoring result of expired operation");
 663                 cl->state = Client_Idle;
 664             }
 665         }
 666         /* if len==1 we do nothing but wait for more input */
 667     }
 668
 669     if (cl->state == Client_Connected) {
 670         send_init(i);
 671     }
 672
 673     if (cl->state == Client_Idle)
 674     {
 675         if (cl->requestid != se->requestid && *se->query) {
 676             send_search(i);
 677         }
 678         else if (cl->hits > 0 && cl->records < global_parameters.toget &&
 679             cl->records < cl->hits) {
 680             send_present(i);
 681         }
 682     }
 683 }
 684
 685 // Disassociate connection from client
 686 static void connection_release(struct connection *co)
 687 {
 688     struct client *cl = co->client;
 689
 690     yaz_log(YLOG_DEBUG, "Connection release %s", co->host->hostport);
 691     if (!cl)
 692         return;
 693     cl->connection = 0;
 694     co->client = 0;
 695 }
 696
 697 // Close connection and recycle structure
 698 static void connection_destroy(struct connection *co)
 699 {
 700     struct host *h = co->host;
 701     cs_close(co->link);
 702     iochan_destroy(co->iochan);
 703
 704     yaz_log(YLOG_DEBUG, "Connection destroy %s", co->host->hostport);
 705     if (h->connections == co)
 706         h->connections = co->next;
 707     else
 708     {
 709         struct connection *pco;
 710         for (pco = h->connections; pco && pco->next != co; pco = pco->next)
 711             ;
 712         if (pco)
 713             pco->next = co->next;
 714         else
 715             abort();
 716     }
 717     if (co->client)
 718     {
 719         if (co->client->state != Client_Idle)
 720             co->client->state = Client_Disconnected;
 721         co->client->connection = 0;
 722     }
 723     co->next = connection_freelist;
 724     connection_freelist = co;
 725 }
 726
 727 // Creates a new connection for client, associated with the host of
 728 // client's database
 729 static struct connection *connection_create(struct client *cl)
 730 {
 731     struct connection *new;
 732     COMSTACK link;
 733     int res;
 734     void *addr;
 735
 736     yaz_log(YLOG_DEBUG, "Connection create %s", cl->database->url);
 737     if (!(link = cs_create(tcpip_type, 0, PROTO_Z3950)))
 738     {
 739         yaz_log(YLOG_FATAL|YLOG_ERRNO, "Failed to create comstack");
 740         exit(1);
 741     }
 742
 743     if (!(addr = cs_straddr(link, cl->database->host->ipport)))
 744     {
 745         yaz_log(YLOG_WARN|YLOG_ERRNO, "Lookup of IP address failed?");
 746         return 0;
 747     }
 748
 749     res = cs_connect(link, addr);
 750     if (res < 0)
 751     {
 752         yaz_log(YLOG_WARN|YLOG_ERRNO, "cs_connect %s", cl->database->url);
 753         return 0;
 754     }
 755
 756     if ((new = connection_freelist))
 757         connection_freelist = new->next;
 758     else
 759     {
 760         new = xmalloc(sizeof (struct connection));
 761         new->ibuf = 0;
 762         new->ibufsize = 0;
 763     }
 764     new->state = Conn_Connecting;
 765     new->host = cl->database->host;
 766     new->next = new->host->connections;
 767     new->host->connections = new;
 768     new->client = cl;
 769     cl->connection = new;
 770     new->link = link;
 771
 772     new->iochan = iochan_create(cs_fileno(link), handler, 0);
 773     iochan_setdata(new->iochan, new);
 774     new->iochan->next = channel_list;
 775     channel_list = new->iochan;
 776     return new;
 777 }
 778
 779 // Close connection and set state to error
 780 static void client_fatal(struct client *cl)
 781 {
 782     yaz_log(YLOG_WARN, "Fatal error from %s", cl->database->url);
 783     connection_destroy(cl->connection);
 784     cl->state = Client_Error;
 785 }
 786
 787 // Ensure that client has a connection associated
 788 static int client_prep_connection(struct client *cl)
 789 {
 790     struct connection *co;
 791     struct session *se = cl->session;
 792     struct host *host = cl->database->host;
 793
 794     co = cl->connection;
 795
 796     yaz_log(YLOG_DEBUG, "Client prep %s", cl->database->url);
 797
 798     if (!co)
 799     {
 800         // See if someone else has an idle connection
 801         // We should look at timestamps here to select the longest-idle connection
 802         for (co = host->connections; co; co = co->next)
 803             if (co->state == Conn_Open && (!co->client || co->client->session != se))
 804                 break;
 805         if (co)
 806         {
 807             connection_release(co);
 808             cl->connection = co;
 809             co->client = cl;
 810         }
 811         else
 812             co = connection_create(cl);
 813     }
 814     if (co)
 815     {
 816         if (co->state == Conn_Connecting)
 817         {
 818             cl->state = Client_Connecting;
 819             iochan_setflag(co->iochan, EVENT_OUTPUT);
 820         }
 821         else if (co->state == Conn_Open)
 822         {
 823             if (cl->state == Client_Error || cl->state == Client_Disconnected)
 824                 cl->state = Client_Idle;
 825             iochan_setflag(co->iochan, EVENT_OUTPUT);
 826         }
 827         return 1;
 828     }
 829     else
 830         return 0;
 831 }
 832
 833 // This function will most likely vanish when a proper target profile mechanism is
 834 // introduced.
 835 void load_simpletargets(const char *fn)
 836 {
 837     FILE *f = fopen(fn, "r");
 838     char line[256];
 839
 840     if (!f)
 841     {
 842         yaz_log(YLOG_WARN|YLOG_ERRNO, "open %s", fn);
 843         exit(1);
 844     }
 845
 846     while (fgets(line, 255, f))
 847     {
 848         char *url, *db;
 849         struct host *host;
 850         struct database *database;
 851
 852         if (strncmp(line, "target ", 7))
 853             continue;
 854         url = line + 7;
 855         url[strlen(url) - 1] = '\0';
 856         yaz_log(YLOG_DEBUG, "Target: %s", url);
 857         if ((db = strchr(url, '/')))
 858             *(db++) = '\0';
 859         else
 860             db = "Default";
 861
 862         for (host = hosts; host; host = host->next)
 863             if (!strcmp(url, host->hostport))
 864                 break;
 865         if (!host)
 866         {
 867             struct addrinfo *addrinfo, hints;
 868             char *port;
 869             char ipport[128];
 870             unsigned char addrbuf[4];
 871             int res;
 872
 873             host = xmalloc(sizeof(struct host));
 874             host->hostport = xstrdup(url);
 875             host->connections = 0;
 876
 877             if ((port = strchr(url, ':')))
 878                 *(port++) = '\0';
 879             else
 880                 port = "210";
 881
 882             hints.ai_flags = 0;
 883             hints.ai_family = PF_INET;
 884             hints.ai_socktype = SOCK_STREAM;
 885             hints.ai_protocol = IPPROTO_TCP;
 886             hints.ai_addrlen = 0;
 887             hints.ai_addr = 0;
 888             hints.ai_canonname = 0;
 889             hints.ai_next = 0;
 890             // This is not robust code. It assumes that getaddrinfo returns AF_INET
 891             // address.
 892             if ((res = getaddrinfo(url, port, &hints, &addrinfo)))
 893             {
 894                 yaz_log(YLOG_WARN, "Failed to resolve %s: %s", url, gai_strerror(res));
 895                 xfree(host->hostport);
 896                 xfree(host);
 897                 continue;
 898             }
 899             assert(addrinfo->ai_family == PF_INET);
 900             memcpy(addrbuf, &((struct sockaddr_in*)addrinfo->ai_addr)->sin_addr.s_addr, 4);
 901             sprintf(ipport, "%hhd.%hhd.%hhd.%hhd:%s",
 902                     addrbuf[0], addrbuf[1], addrbuf[2], addrbuf[3], port);
 903             host->ipport = xstrdup(ipport);
 904             freeaddrinfo(addrinfo);
 905             host->next = hosts;
 906             hosts = host;
 907         }
 908         database = xmalloc(sizeof(struct database));
 909         database->host = host;
 910         database->url = xmalloc(strlen(url) + strlen(db) + 2);
 911         strcpy(database->url, url);
 912         strcat(database->url, "/");
 913         strcat(database->url, db);
 914
 915         database->databases = xmalloc(2 * sizeof(char *));
 916         database->databases[0] = xstrdup(db);
 917         database->databases[1] = 0;
 918         database->errors = 0;
 919         database->qprofile = 0;
 920         database->rprofile = database_retrieval_profile(database);
 921         database->next = databases;
 922         databases = database;
 923
 924     }
 925     fclose(f);
 926 }
 927
 928 static void pull_terms(NMEM nmem, struct ccl_rpn_node *n, char **termlist, int *num)
 929 {
 930     switch (n->kind)
 931     {
 932         case CCL_RPN_AND:
 933         case CCL_RPN_OR:
 934         case CCL_RPN_NOT:
 935         case CCL_RPN_PROX:
 936             pull_terms(nmem, n->u.p[0], termlist, num);
 937             pull_terms(nmem, n->u.p[1], termlist, num);
 938             break;
 939         case CCL_RPN_TERM:
 940             termlist[(*num)++] = nmem_strdup(nmem, n->u.t.term);
 941             break;
 942         default: // NOOP
 943             break;
 944     }
 945 }
 946
 947 // Extract terms from query into null-terminated termlist
 948 static int extract_terms(NMEM nmem, char *query, char **termlist)
 949 {
 950     int error, pos;
 951     struct ccl_rpn_node *n;
 952     int num = 0;
 953
 954     n = ccl_find_str(global_parameters.ccl_filter, query, &error, &pos);
 955     if (!n)
 956         return -1;
 957     pull_terms(nmem, n, termlist, &num);
 958     termlist[num] = 0;
 959     ccl_rpn_delete(n);
 960     return 0;
 961 }
 962
 963 static struct client *client_create(void)
 964 {
 965     struct client *r;
 966     if (client_freelist)
 967     {
 968         r = client_freelist;
 969         client_freelist = client_freelist->next;
 970     }
 971     else
 972         r = xmalloc(sizeof(struct client));
 973     r->database = 0;
 974     r->connection = 0;
 975     r->session = 0;
 976     r->hits = 0;
 977     r->records = 0;
 978     r->setno = 0;
 979     r->requestid = -1;
 980     r->diagnostic = 0;
 981     r->state = Client_Disconnected;
 982     r->next = 0;
 983     return r;
 984 }
 985
 986 void client_destroy(struct client *c)
 987 {
 988     struct session *se = c->session;
 989     if (c == se->clients)
 990         se->clients = c->next;
 991     else
 992     {
 993         struct client *cc;
 994         for (cc = se->clients; cc && cc->next != c; cc = cc->next)
 995             ;
 996         if (cc)
 997             cc->next = c->next;
 998     }
 999     if (c->connection)
1000         connection_release(c->connection);
1001     c->next = client_freelist;
1002     client_freelist = c;
1003 }
1004
1005 void session_set_watch(struct session *s, int what, session_watchfun fun, void *data)
1006 {
1007     s->watchlist[what].fun = fun;
1008     s->watchlist[what].data = data;
1009 }
1010
1011 void session_alert_watch(struct session *s, int what)
1012 {
1013     if (!s->watchlist[what].fun)
1014         return;
1015     (*s->watchlist[what].fun)(s->watchlist[what].data);
1016     s->watchlist[what].fun = 0;
1017     s->watchlist[what].data = 0;
1018 }
1019
1020 // This needs to be extended with selection criteria
1021 static struct conf_retrievalprofile *database_retrieval_profile(struct database *db)
1022 {
1023     if (!config)
1024     {
1025         yaz_log(YLOG_FATAL, "Must load configuration (-f)");
1026         exit(1);
1027     }
1028     if (!config->retrievalprofiles)
1029     {
1030         yaz_log(YLOG_FATAL, "No retrieval profiles defined");
1031     }
1032     return config->retrievalprofiles;
1033 }
1034
1035 // This should be extended with parameters to control selection criteria
1036 // Associates a set of clients with a session;
1037 int select_targets(struct session *se)
1038 {
1039     struct database *db;
1040     int c = 0;
1041
1042     while (se->clients)
1043         client_destroy(se->clients);
1044     for (db = databases; db; db = db->next)
1045     {
1046         struct client *cl = client_create();
1047         cl->database = db;
1048         cl->session = se;
1049         cl->next = se->clients;
1050         se->clients = cl;
1051         c++;
1052     }
1053     return c;
1054 }
1055
1056 int session_active_clients(struct session *s)
1057 {
1058     struct client *c;
1059     int res = 0;
1060
1061     for (c = s->clients; c; c = c->next)
1062         if (c->connection && (c->state == Client_Connecting ||
1063                     c->state == Client_Initializing ||
1064                     c->state == Client_Searching ||
1065                     c->state == Client_Presenting))
1066             res++;
1067
1068     return res;
1069 }
1070
1071 char *search(struct session *se, char *query)
1072 {
1073     int live_channels = 0;
1074     struct client *cl;
1075
1076     yaz_log(YLOG_DEBUG, "Search");
1077
1078     strcpy(se->query, query);
1079     se->requestid++;
1080     nmem_reset(se->nmem);
1081     for (cl = se->clients; cl; cl = cl->next)
1082     {
1083         cl->hits = -1;
1084         cl->records = 0;
1085         cl->diagnostic = 0;
1086
1087         if (client_prep_connection(cl))
1088             live_channels++;
1089     }
1090     if (live_channels)
1091     {
1092         char *p[512];
1093         int maxrecs = live_channels * global_parameters.toget;
1094         se->num_termlists = 0;
1095         se->reclist = reclist_create(se->nmem, maxrecs);
1096         extract_terms(se->nmem, query, p);
1097         se->relevance = relevance_create(se->nmem, (const char **) p, maxrecs);
1098         se->total_records = se->total_hits = 0;
1099         se->expected_maxrecs = maxrecs;
1100     }
1101     else
1102         return "NOTARGETS";
1103
1104     return 0;
1105 }
1106
1107 void destroy_session(struct session *s)
1108 {
1109     yaz_log(YLOG_LOG, "Destroying session");
1110     while (s->clients)
1111         client_destroy(s->clients);
1112     nmem_destroy(s->nmem);
1113     wrbuf_free(s->wrbuf, 1);
1114 }
1115
1116 struct session *new_session()
1117 {
1118     int i;
1119     struct session *session = xmalloc(sizeof(*session));
1120
1121     yaz_log(YLOG_DEBUG, "New pazpar2 session");
1122
1123     session->total_hits = 0;
1124     session->total_records = 0;
1125     session->num_termlists = 0;
1126     session->reclist = 0;
1127     session->requestid = -1;
1128     session->clients = 0;
1129     session->expected_maxrecs = 0;
1130     session->query[0] = '\0';
1131     session->nmem = nmem_create();
1132     session->wrbuf = wrbuf_alloc();
1133     for (i = 0; i <= SESSION_WATCH_MAX; i++)
1134     {
1135         session->watchlist[i].data = 0;
1136         session->watchlist[i].fun = 0;
1137     }
1138
1139     select_targets(session);
1140
1141     return session;
1142 }
1143
1144 struct hitsbytarget *hitsbytarget(struct session *se, int *count)
1145 {
1146     static struct hitsbytarget res[1000]; // FIXME MM
1147     struct client *cl;
1148
1149     *count = 0;
1150     for (cl = se->clients; cl; cl = cl->next)
1151     {
1152         strcpy(res[*count].id, cl->database->host->hostport);
1153         res[*count].hits = cl->hits;
1154         res[*count].records = cl->records;
1155         res[*count].diagnostic = cl->diagnostic;
1156         res[*count].state = client_states[cl->state];
1157         res[*count].connected  = cl->connection ? 1 : 0;
1158         (*count)++;
1159     }
1160
1161     return res;
1162 }
1163
1164 struct termlist_score **termlist(struct session *s, const char *name, int *num)
1165 {
1166     int i;
1167
1168     for (i = 0; i < s->num_termlists; i++)
1169         if (!strcmp(s->termlists[i].name, name))
1170             return termlist_highscore(s->termlists[i].termlist, num);
1171     return 0;
1172 }
1173
1174 #ifdef MISSING_HEADERS
1175 void report_nmem_stats(void)
1176 {
1177     size_t in_use, is_free;
1178
1179     nmem_get_memory_in_use(&in_use);
1180     nmem_get_memory_free(&is_free);
1181
1182     yaz_log(YLOG_LOG, "nmem stat: use=%ld free=%ld",
1183             (long) in_use, (long) is_free);
1184 }
1185 #endif
1186
1187 struct record **show(struct session *s, int start, int *num, int *total,
1188                      int *sumhits, NMEM nmem_show)
1189 {
1190     struct record **recs = nmem_malloc(nmem_show, *num
1191                                        * sizeof(struct record *));
1192     int i;
1193 #if USE_TIMING
1194     yaz_timing_t t = yaz_timing_create();
1195 #endif
1196     relevance_prepare_read(s->relevance, s->reclist);
1197
1198     *total = s->reclist->num_records;
1199     *sumhits = s->total_hits;
1200
1201     for (i = 0; i < start; i++)
1202         if (!reclist_read_record(s->reclist))
1203         {
1204             *num = 0;
1205             recs = 0;
1206             break;
1207         }
1208
1209     for (i = 0; i < *num; i++)
1210     {
1211         struct record *r = reclist_read_record(s->reclist);
1212         if (!r)
1213         {
1214             *num = i;
1215             break;
1216         }
1217         recs[i] = r;
1218     }
1219 #if USE_TIMING
1220     yaz_timing_stop(t);
1221     yaz_log(YLOG_LOG, "show %6.5f %3.2f %3.2f",
1222             yaz_timing_get_real(t), yaz_timing_get_user(t),
1223             yaz_timing_get_sys(t));
1224     yaz_timing_destroy(&t);
1225 #endif
1226     return recs;
1227 }
1228
1229 void statistics(struct session *se, struct statistics *stat)
1230 {
1231     struct client *cl;
1232     int count = 0;
1233
1234     bzero(stat, sizeof(*stat));
1235     for (cl = se->clients; cl; cl = cl->next)
1236     {
1237         if (!cl->connection)
1238             stat->num_no_connection++;
1239         switch (cl->state)
1240         {
1241             case Client_Connecting: stat->num_connecting++; break;
1242             case Client_Initializing: stat->num_initializing++; break;
1243             case Client_Searching: stat->num_searching++; break;
1244             case Client_Presenting: stat->num_presenting++; break;
1245             case Client_Idle: stat->num_idle++; break;
1246             case Client_Failed: stat->num_failed++; break;
1247             case Client_Error: stat->num_error++; break;
1248             default: break;
1249         }
1250         count++;
1251     }
1252     stat->num_hits = se->total_hits;
1253     stat->num_records = se->total_records;
1254
1255     stat->num_clients = count;
1256 }
1257
1258 static CCL_bibset load_cclfile(const char *fn)
1259 {
1260     CCL_bibset res = ccl_qual_mk();
1261     if (ccl_qual_fname(res, fn) < 0)
1262     {
1263         yaz_log(YLOG_FATAL|YLOG_ERRNO, "%s", fn);
1264         exit(1);
1265     }
1266     return res;
1267 }
1268
1269 int main(int argc, char **argv)
1270 {
1271     int ret;
1272     char *arg;
1273     int setport = 0;
1274
1275     if (signal(SIGPIPE, SIG_IGN) < 0)
1276         yaz_log(YLOG_WARN|YLOG_ERRNO, "signal");
1277
1278     yaz_log_init(YLOG_DEFAULT_LEVEL, "pazpar2", 0);
1279
1280     while ((ret = options("f:x:h:p:C:s:d", argv, argc, &arg)) != -2)
1281     {
1282         switch (ret) {
1283             case 'f':
1284                 if (!read_config(arg))
1285                     exit(1);
1286                 break;
1287             case 'h':
1288                 http_init(arg);
1289                 setport++;
1290                 break;
1291             case 'C':
1292                 global_parameters.ccl_filter = load_cclfile(arg);
1293                 break;
1294             case 'p':
1295                 http_set_proxyaddr(arg);
1296                 break;
1297             case 's':
1298                 load_simpletargets(arg);
1299                 break;
1300             case 'd':
1301                 global_parameters.dump_records = 1;
1302                 break;
1303             default:
1304                 fprintf(stderr, "Usage: pazpar2\n"
1305                         "    -f configfile\n"
1306                         "    -h [host:]port          (REST protocol listener)\n"
1307                         "    -C cclconfig\n"
1308                         "    -s simpletargetfile\n"
1309                         "    -p hostname[:portno]    (HTTP proxy)\n");
1310                 exit(1);
1311         }
1312     }
1313
1314     if (!setport)
1315     {
1316         fprintf(stderr, "Set command port with -h\n");
1317         exit(1);
1318     }
1319
1320     global_parameters.ccl_filter = load_cclfile("../etc/default.bib");
1321     global_parameters.yaz_marc = yaz_marc_create();
1322     yaz_marc_subfield_str(global_parameters.yaz_marc, "\t");
1323     global_parameters.odr_in = odr_createmem(ODR_DECODE);
1324     global_parameters.odr_out = odr_createmem(ODR_ENCODE);
1325
1326     event_loop(&channel_list);
1327
1328     return 0;
1329 }
1330
1331 /*
1332  * Local variables:
1333  * c-basic-offset: 4
1334  * indent-tabs-mode: nil
1335  * End:
1336  * vim: shiftwidth=4 tabstop=8 expandtab
1337  */