src/pazpar2.c

   1 /* $Id: pazpar2.c,v 1.14 2007-01-04 22:04:25 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/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 #define USE_TIMING 0
  24 #if USE_TIMING
  25 #include <yaz/timing.h>
  26 #endif
  27
  28 #include "pazpar2.h"
  29 #include "eventl.h"
  30 #include "command.h"
  31 #include "http.h"
  32 #include "termlists.h"
  33 #include "reclists.h"
  34 #include "relevance.h"
  35 #include "config.h"
  36
  37 #define PAZPAR2_VERSION "0.1"
  38 #define MAX_CHUNK 15
  39
  40 static void client_fatal(struct client *cl);
  41 static void connection_destroy(struct connection *co);
  42 static int client_prep_connection(struct client *cl);
  43 static void ingest_records(struct client *cl, Z_Records *r);
  44 static struct conf_retrievalprofile *database_retrieval_profile(struct database *db);
  45 void session_alert_watch(struct session *s, int what);
  46
  47 IOCHAN channel_list = 0;  // Master list of connections we're handling events to
  48
  49 static struct connection *connection_freelist = 0;
  50 static struct client *client_freelist = 0;
  51
  52 static struct host *hosts = 0;  // The hosts we know about
  53 static struct database *databases = 0; // The databases we know about
  54
  55 static char *client_states[] = {
  56     "Client_Connecting",
  57     "Client_Connected",
  58     "Client_Idle",
  59     "Client_Initializing",
  60     "Client_Searching",
  61     "Client_Presenting",
  62     "Client_Error",
  63     "Client_Failed",
  64     "Client_Disconnected",
  65     "Client_Stopped"
  66 };
  67
  68 struct parameters global_parameters =
  69 {
  70     0,
  71     30,
  72     "81",
  73     "Index Data PazPar2 (MasterKey)",
  74     PAZPAR2_VERSION,
  75     600, // 10 minutes
  76     60,
  77     100,
  78     MAX_CHUNK,
  79     0,
  80     0,
  81     0,
  82     0
  83 };
  84
  85 static int send_apdu(struct client *c, Z_APDU *a)
  86 {
  87     struct connection *co = c->connection;
  88     char *buf;
  89     int len, r;
  90
  91     if (!z_APDU(global_parameters.odr_out, &a, 0, 0))
  92     {
  93         odr_perror(global_parameters.odr_out, "Encoding APDU");
  94         abort();
  95     }
  96     buf = odr_getbuf(global_parameters.odr_out, &len, 0);
  97     r = cs_put(co->link, buf, len);
  98     if (r < 0)
  99     {
 100         yaz_log(YLOG_WARN, "cs_put: %s", cs_errmsg(cs_errno(co->link)));
 101         return -1;
 102     }
 103     else if (r == 1)
 104     {
 105         fprintf(stderr, "cs_put incomplete (ParaZ does not handle that)\n");
 106         exit(1);
 107     }
 108     odr_reset(global_parameters.odr_out); /* release the APDU structure  */
 109     co->state = Conn_Waiting;
 110     return 0;
 111 }
 112
 113
 114 static void send_init(IOCHAN i)
 115 {
 116     struct connection *co = iochan_getdata(i);
 117     struct client *cl = co->client;
 118     Z_APDU *a = zget_APDU(global_parameters.odr_out, Z_APDU_initRequest);
 119
 120     a->u.initRequest->implementationId = global_parameters.implementationId;
 121     a->u.initRequest->implementationName = global_parameters.implementationName;
 122     a->u.initRequest->implementationVersion =
 123         global_parameters.implementationVersion;
 124     ODR_MASK_SET(a->u.initRequest->options, Z_Options_search);
 125     ODR_MASK_SET(a->u.initRequest->options, Z_Options_present);
 126     ODR_MASK_SET(a->u.initRequest->options, Z_Options_namedResultSets);
 127
 128     ODR_MASK_SET(a->u.initRequest->protocolVersion, Z_ProtocolVersion_1);
 129     ODR_MASK_SET(a->u.initRequest->protocolVersion, Z_ProtocolVersion_2);
 130     ODR_MASK_SET(a->u.initRequest->protocolVersion, Z_ProtocolVersion_3);
 131     if (send_apdu(cl, a) >= 0)
 132     {
 133         iochan_setflags(i, EVENT_INPUT);
 134         cl->state = Client_Initializing;
 135     }
 136     else
 137         cl->state = Client_Error;
 138     odr_reset(global_parameters.odr_out);
 139 }
 140
 141 static void send_search(IOCHAN i)
 142 {
 143     struct connection *co = iochan_getdata(i);
 144     struct client *cl = co->client;
 145     struct session *se = cl->session;
 146     struct database *db = cl->database;
 147     Z_APDU *a = zget_APDU(global_parameters.odr_out, Z_APDU_searchRequest);
 148     int ndb, cerror, cpos;
 149     char **databaselist;
 150     Z_Query *zquery;
 151     struct ccl_rpn_node *cn;
 152     int ssub = 0, lslb = 100000, mspn = 10;
 153
 154     yaz_log(YLOG_DEBUG, "Sending search");
 155
 156     cn = ccl_find_str(global_parameters.ccl_filter, se->query, &cerror, &cpos);
 157     if (!cn)
 158         return;
 159     a->u.searchRequest->query = zquery = odr_malloc(global_parameters.odr_out,
 160             sizeof(Z_Query));
 161     zquery->which = Z_Query_type_1;
 162     zquery->u.type_1 = ccl_rpn_query(global_parameters.odr_out, cn);
 163     ccl_rpn_delete(cn);
 164
 165     for (ndb = 0; db->databases[ndb]; ndb++)
 166         ;
 167     databaselist = odr_malloc(global_parameters.odr_out, sizeof(char*) * ndb);
 168     for (ndb = 0; db->databases[ndb]; ndb++)
 169         databaselist[ndb] = db->databases[ndb];
 170
 171     a->u.presentRequest->preferredRecordSyntax =
 172             yaz_oidval_to_z3950oid(global_parameters.odr_out,
 173             CLASS_RECSYN, VAL_USMARC);
 174     a->u.searchRequest->smallSetUpperBound = &ssub;
 175     a->u.searchRequest->largeSetLowerBound = &lslb;
 176     a->u.searchRequest->mediumSetPresentNumber = &mspn;
 177     a->u.searchRequest->resultSetName = "Default";
 178     a->u.searchRequest->databaseNames = databaselist;
 179     a->u.searchRequest->num_databaseNames = ndb;
 180
 181     if (send_apdu(cl, a) >= 0)
 182     {
 183         iochan_setflags(i, EVENT_INPUT);
 184         cl->state = Client_Searching;
 185         cl->requestid = se->requestid;
 186     }
 187     else
 188         cl->state = Client_Error;
 189
 190     odr_reset(global_parameters.odr_out);
 191 }
 192
 193 static void send_present(IOCHAN i)
 194 {
 195     struct connection *co = iochan_getdata(i);
 196     struct client *cl = co->client;
 197     Z_APDU *a = zget_APDU(global_parameters.odr_out, Z_APDU_presentRequest);
 198     int toget;
 199     int start = cl->records + 1;
 200
 201     toget = global_parameters.chunk;
 202     if (toget > cl->hits - cl->records)
 203         toget = cl->hits - cl->records;
 204
 205     yaz_log(YLOG_DEBUG, "Trying to present %d records\n", toget);
 206
 207     a->u.presentRequest->resultSetStartPoint = &start;
 208     a->u.presentRequest->numberOfRecordsRequested = &toget;
 209
 210     a->u.presentRequest->resultSetId = "Default";
 211
 212     a->u.presentRequest->preferredRecordSyntax =
 213             yaz_oidval_to_z3950oid(global_parameters.odr_out,
 214             CLASS_RECSYN, VAL_USMARC);
 215
 216     if (send_apdu(cl, a) >= 0)
 217     {
 218         iochan_setflags(i, EVENT_INPUT);
 219         cl->state = Client_Presenting;
 220     }
 221     else
 222         cl->state = Client_Error;
 223     odr_reset(global_parameters.odr_out);
 224 }
 225
 226 static void do_initResponse(IOCHAN i, Z_APDU *a)
 227 {
 228     struct connection *co = iochan_getdata(i);
 229     struct client *cl = co->client;
 230     Z_InitResponse *r = a->u.initResponse;
 231
 232     yaz_log(YLOG_DEBUG, "Received init response");
 233
 234     if (*r->result)
 235     {
 236         cl->state = Client_Idle;
 237     }
 238     else
 239         cl->state = Client_Failed; // FIXME need to do something to the connection
 240 }
 241
 242 static void do_searchResponse(IOCHAN i, Z_APDU *a)
 243 {
 244     struct connection *co = iochan_getdata(i);
 245     struct client *cl = co->client;
 246     struct session *se = cl->session;
 247     Z_SearchResponse *r = a->u.searchResponse;
 248
 249     yaz_log(YLOG_DEBUG, "Searchresponse (status=%d)", *r->searchStatus);
 250
 251     if (*r->searchStatus)
 252     {
 253         cl->hits = *r->resultCount;
 254         se->total_hits += cl->hits;
 255         if (r->presentStatus && !*r->presentStatus && r->records)
 256         {
 257             yaz_log(YLOG_DEBUG, "Records in search response");
 258             cl->records += *r->numberOfRecordsReturned;
 259             ingest_records(cl, r->records);
 260         }
 261         cl->state = Client_Idle;
 262     }
 263     else
 264     {          /*"FAILED"*/
 265         cl->hits = 0;
 266         cl->state = Client_Error;
 267         if (r->records) {
 268             Z_Records *recs = r->records;
 269             if (recs->which == Z_Records_NSD)
 270             {
 271                 yaz_log(YLOG_WARN, "Non-surrogate diagnostic");
 272                 cl->diagnostic = *recs->u.nonSurrogateDiagnostic->condition;
 273                 cl->state = Client_Error;
 274             }
 275         }
 276     }
 277 }
 278
 279 char *normalize_mergekey(char *buf)
 280 {
 281     char *p = buf, *pout = buf;
 282
 283     while (*p)
 284     {
 285         while (*p && !isalnum(*p))
 286             p++;
 287         while (isalnum(*p))
 288             *(pout++) = tolower(*(p++));
 289         if (*p)
 290             *(pout++) = ' ';
 291         while (*p && !isalnum(*p))
 292             p++;
 293     }
 294     if (buf != pout)
 295         *pout = '\0';
 296
 297     return buf;
 298 }
 299
 300
 301 #ifdef GAGA
 302 // FIXME needs to be generalized. Should flexibly generate X lists per search
 303 static void extract_subject(struct session *s, const char *rec)
 304 {
 305     const char *field, *subfield;
 306
 307     while ((field = find_field(rec, "650")))
 308     {
 309         rec = field;
 310         if ((subfield = find_subfield(field, 'a')))
 311         {
 312             char *e, *ef;
 313             char buf[1024];
 314             int len;
 315
 316             ef = index(subfield, '\n');
 317             if (!ef)
 318                 return;
 319             if ((e = index(subfield, '\t')) && e < ef)
 320                 ef = e;
 321             while (ef > subfield && !isalpha(*(ef - 1)) && *(ef - 1) != ')')
 322                 ef--;
 323             len = ef - subfield;
 324             assert(len < 1023);
 325             memcpy(buf, subfield, len);
 326             buf[len] = '\0';
 327 #ifdef FIXME
 328             if (*buf)
 329                 termlist_insert(s->termlist, buf);
 330 #endif
 331         }
 332     }
 333 }
 334 #endif
 335
 336 static void add_facet(struct session *s, const char *type, const char *value)
 337 {
 338     int i;
 339
 340     for (i = 0; i < s->num_termlists; i++)
 341         if (!strcmp(s->termlists[i].name, type))
 342             break;
 343     if (i == s->num_termlists)
 344     {
 345         if (i == SESSION_MAX_TERMLISTS)
 346         {
 347             yaz_log(YLOG_FATAL, "Too many termlists");
 348             exit(1);
 349         }
 350         s->termlists[i].name = nmem_strdup(s->nmem, type);
 351         s->termlists[i].termlist = termlist_create(s->nmem, s->expected_maxrecs, 15);
 352         s->num_termlists = i + 1;
 353     }
 354     termlist_insert(s->termlists[i].termlist, value);
 355 }
 356
 357 static xmlDoc *normalize_record(struct client *cl, Z_External *rec)
 358 {
 359     struct conf_retrievalprofile *rprofile = cl->database->rprofile;
 360     struct conf_retrievalmap *m;
 361     xmlNode *res;
 362     xmlDoc *rdoc;
 363
 364     // First normalize to XML
 365     if (rprofile->native_syntax == Nativesyn_iso2709)
 366     {
 367         char *buf;
 368         int len;
 369         if (rec->which != Z_External_octet)
 370         {
 371             yaz_log(YLOG_WARN, "Unexpected external branch, probably BER");
 372             return 0;
 373         }
 374         buf = (char*) rec->u.octet_aligned->buf;
 375         len = rec->u.octet_aligned->len;
 376         if (yaz_marc_read_iso2709(rprofile->yaz_marc, buf, len) < 0)
 377         {
 378             yaz_log(YLOG_WARN, "Failed to decode MARC");
 379             return 0;
 380         }
 381         if (yaz_marc_write_xml(rprofile->yaz_marc, &res,
 382                     "http://www.loc.gov/MARC21/slim", 0, 0) < 0)
 383         {
 384             yaz_log(YLOG_WARN, "Failed to encode as XML");
 385             return 0;
 386         }
 387         rdoc = xmlNewDoc("1.0");
 388         xmlDocSetRootElement(rdoc, res);
 389     }
 390     else
 391     {
 392         yaz_log(YLOG_FATAL, "Unknown native_syntax in normalize_record");
 393         exit(1);
 394     }
 395     for (m = rprofile->maplist; m; m = m->next)
 396     {
 397         xmlDoc *new;
 398         if (m->type != Map_xslt)
 399         {
 400             yaz_log(YLOG_WARN, "Unknown map type");
 401             return 0;
 402         }
 403         if (!(new = xsltApplyStylesheet(m->stylesheet, rdoc, 0)))
 404         {
 405             yaz_log(YLOG_WARN, "XSLT transformation failed");
 406             return 0;
 407         }
 408         xmlFreeDoc(rdoc);
 409         rdoc = new;
 410     }
 411     if (global_parameters.dump_records)
 412     {
 413         fprintf(stderr, "Record:\n----------------\n");
 414         xmlDocFormatDump(stderr, rdoc, 1);
 415     }
 416     return rdoc;
 417 }
 418
 419 static struct record *ingest_record(struct client *cl, Z_External *rec)
 420 {
 421     xmlDoc *xdoc = normalize_record(cl, rec);
 422     xmlNode *root, *n;
 423     struct record *res, *head;
 424     struct session *se = cl->session;
 425     xmlChar *mergekey, *mergekey_norm;
 426
 427     if (!xdoc)
 428         return 0;
 429
 430     root = xmlDocGetRootElement(xdoc);
 431     if (!(mergekey = xmlGetProp(root, "mergekey")))
 432     {
 433         yaz_log(YLOG_WARN, "No mergekey found in record");
 434         return 0;
 435     }
 436
 437     res = nmem_malloc(se->nmem, sizeof(struct record));
 438     res->next_cluster = 0;
 439     res->target_offset = -1;
 440     res->term_frequency_vec = 0;
 441     res->title = "Unknown";
 442     res->relevance = 0;
 443
 444     mergekey_norm = nmem_strdup(se->nmem, (char*) mergekey);
 445     xmlFree(mergekey);
 446     res->merge_key = normalize_mergekey(mergekey_norm);
 447
 448     head = reclist_insert(se->reclist, res);
 449     relevance_newrec(se->relevance, head);
 450
 451     for (n = root->children; n; n = n->next)
 452     {
 453         if (n->type != XML_ELEMENT_NODE)
 454             continue;
 455         if (!strcmp(n->name, "facet"))
 456         {
 457             xmlChar *type = xmlGetProp(n, "type");
 458             xmlChar *value = xmlNodeListGetString(xdoc, n->children, 0);
 459             if (type && value)
 460             {
 461                 add_facet(se, type, value);
 462                 relevance_countwords(se->relevance, head, value, 1);
 463             }
 464             xmlFree(type);
 465             xmlFree(value);
 466         }
 467         else if (!strcmp(n->name, "metadata"))
 468         {
 469             xmlChar *type = xmlGetProp(n, "type");
 470             if (type && !strcmp(type, "title"))
 471             {
 472                 xmlChar *value = xmlNodeListGetString(xdoc, n->children, 0);
 473                 if (value)
 474                 {
 475                     res->title = nmem_strdup(se->nmem, value);
 476                     relevance_countwords(se->relevance, head, value, 4);
 477                     xmlFree(value);
 478                 }
 479             }
 480             xmlFree(type);
 481         }
 482         else
 483             yaz_log(YLOG_WARN, "Unexpected element %s in internal record", n->name);
 484     }
 485
 486     xmlFreeDoc(xdoc);
 487
 488     relevance_donerecord(se->relevance, head);
 489     se->total_records++;
 490
 491     return res;
 492 }
 493
 494 static void ingest_records(struct client *cl, Z_Records *r)
 495 {
 496 #if USE_TIMING
 497     yaz_timing_t t = yaz_timing_create();
 498 #endif
 499     struct record *rec;
 500     struct session *s = cl->session;
 501     Z_NamePlusRecordList *rlist;
 502     int i;
 503
 504     if (r->which != Z_Records_DBOSD)
 505         return;
 506     rlist = r->u.databaseOrSurDiagnostics;
 507     for (i = 0; i < rlist->num_records; i++)
 508     {
 509         Z_NamePlusRecord *npr = rlist->records[i];
 510
 511         if (npr->which != Z_NamePlusRecord_databaseRecord)
 512         {
 513             yaz_log(YLOG_WARN, "Unexpected record type, probably diagnostic");
 514             continue;
 515         }
 516
 517         rec = ingest_record(cl, npr->u.databaseRecord);
 518         if (!rec)
 519             continue;
 520     }
 521     if (s->watchlist[SESSION_WATCH_RECORDS].fun && rlist->num_records)
 522         session_alert_watch(s, SESSION_WATCH_RECORDS);
 523
 524 #if USE_TIMING
 525     yaz_timing_stop(t);
 526     yaz_log(YLOG_LOG, "ingest_records %6.5f %3.2f %3.2f",
 527             yaz_timing_get_real(t), yaz_timing_get_user(t),
 528             yaz_timing_get_sys(t));
 529     yaz_timing_destroy(&t);
 530 #endif
 531 }
 532
 533 static void do_presentResponse(IOCHAN i, Z_APDU *a)
 534 {
 535     struct connection *co = iochan_getdata(i);
 536     struct client *cl = co->client;
 537     Z_PresentResponse *r = a->u.presentResponse;
 538
 539     if (r->records) {
 540         Z_Records *recs = r->records;
 541         if (recs->which == Z_Records_NSD)
 542         {
 543             yaz_log(YLOG_WARN, "Non-surrogate diagnostic");
 544             cl->diagnostic = *recs->u.nonSurrogateDiagnostic->condition;
 545             cl->state = Client_Error;
 546         }
 547     }
 548
 549     if (!*r->presentStatus && cl->state != Client_Error)
 550     {
 551         yaz_log(YLOG_DEBUG, "Good Present response");
 552         cl->records += *r->numberOfRecordsReturned;
 553         ingest_records(cl, r->records);
 554         cl->state = Client_Idle;
 555     }
 556     else if (*r->presentStatus)
 557     {
 558         yaz_log(YLOG_WARN, "Bad Present response");
 559         cl->state = Client_Error;
 560     }
 561 }
 562
 563 static void handler(IOCHAN i, int event)
 564 {
 565     struct connection *co = iochan_getdata(i);
 566     struct client *cl = co->client;
 567     struct session *se = 0;
 568
 569     if (cl)
 570         se = cl->session;
 571     else
 572     {
 573         yaz_log(YLOG_WARN, "Destroying orphan connection");
 574         connection_destroy(co);
 575         return;
 576     }
 577
 578     if (co->state == Conn_Connecting && event & EVENT_OUTPUT)
 579     {
 580         int errcode;
 581         socklen_t errlen = sizeof(errcode);
 582
 583         if (getsockopt(cs_fileno(co->link), SOL_SOCKET, SO_ERROR, &errcode,
 584             &errlen) < 0 || errcode != 0)
 585         {
 586             client_fatal(cl);
 587             return;
 588         }
 589         else
 590         {
 591             yaz_log(YLOG_DEBUG, "Connect OK");
 592             co->state = Conn_Open;
 593             if (cl)
 594                 cl->state = Client_Connected;
 595         }
 596     }
 597
 598     else if (event & EVENT_INPUT)
 599     {
 600         int len = cs_get(co->link, &co->ibuf, &co->ibufsize);
 601
 602         if (len < 0)
 603         {
 604             yaz_log(YLOG_WARN|YLOG_ERRNO, "Error reading from Z server");
 605             connection_destroy(co);
 606             return;
 607         }
 608         else if (len == 0)
 609         {
 610             yaz_log(YLOG_WARN, "EOF reading from Z server");
 611             connection_destroy(co);
 612             return;
 613         }
 614         else if (len > 1) // We discard input if we have no connection
 615         {
 616             co->state = Conn_Open;
 617
 618             if (cl && (cl->requestid == se->requestid || cl->state == Client_Initializing))
 619             {
 620                 Z_APDU *a;
 621
 622                 odr_reset(global_parameters.odr_in);
 623                 odr_setbuf(global_parameters.odr_in, co->ibuf, len, 0);
 624                 if (!z_APDU(global_parameters.odr_in, &a, 0, 0))
 625                 {
 626                     client_fatal(cl);
 627                     return;
 628                 }
 629                 switch (a->which)
 630                 {
 631                     case Z_APDU_initResponse:
 632                         do_initResponse(i, a);
 633                         break;
 634                     case Z_APDU_searchResponse:
 635                         do_searchResponse(i, a);
 636                         break;
 637                     case Z_APDU_presentResponse:
 638                         do_presentResponse(i, a);
 639                         break;
 640                     default:
 641                         yaz_log(YLOG_WARN, "Unexpected result from server");
 642                         client_fatal(cl);
 643                         return;
 644                 }
 645                 // We aren't expecting staggered output from target
 646                 // if (cs_more(t->link))
 647                 //    iochan_setevent(i, EVENT_INPUT);
 648             }
 649             else  // we throw away response and go to idle mode
 650             {
 651                 yaz_log(YLOG_DEBUG, "Ignoring result of expired operation");
 652                 cl->state = Client_Idle;
 653             }
 654         }
 655         /* if len==1 we do nothing but wait for more input */
 656     }
 657
 658     if (cl->state == Client_Connected) {
 659         send_init(i);
 660     }
 661
 662     if (cl->state == Client_Idle)
 663     {
 664         if (cl->requestid != se->requestid && *se->query) {
 665             send_search(i);
 666         }
 667         else if (cl->hits > 0 && cl->records < global_parameters.toget &&
 668             cl->records < cl->hits) {
 669             send_present(i);
 670         }
 671     }
 672 }
 673
 674 // Disassociate connection from client
 675 static void connection_release(struct connection *co)
 676 {
 677     struct client *cl = co->client;
 678
 679     yaz_log(YLOG_DEBUG, "Connection release %s", co->host->hostport);
 680     if (!cl)
 681         return;
 682     cl->connection = 0;
 683     co->client = 0;
 684 }
 685
 686 // Close connection and recycle structure
 687 static void connection_destroy(struct connection *co)
 688 {
 689     struct host *h = co->host;
 690     cs_close(co->link);
 691     iochan_destroy(co->iochan);
 692
 693     yaz_log(YLOG_DEBUG, "Connection destroy %s", co->host->hostport);
 694     if (h->connections == co)
 695         h->connections = co->next;
 696     else
 697     {
 698         struct connection *pco;
 699         for (pco = h->connections; pco && pco->next != co; pco = pco->next)
 700             ;
 701         if (pco)
 702             pco->next = co->next;
 703         else
 704             abort();
 705     }
 706     if (co->client)
 707     {
 708         if (co->client->state != Client_Idle)
 709             co->client->state = Client_Disconnected;
 710         co->client->connection = 0;
 711     }
 712     co->next = connection_freelist;
 713     connection_freelist = co;
 714 }
 715
 716 // Creates a new connection for client, associated with the host of
 717 // client's database
 718 static struct connection *connection_create(struct client *cl)
 719 {
 720     struct connection *new;
 721     COMSTACK link;
 722     int res;
 723     void *addr;
 724
 725     yaz_log(YLOG_DEBUG, "Connection create %s", cl->database->url);
 726     if (!(link = cs_create(tcpip_type, 0, PROTO_Z3950)))
 727     {
 728         yaz_log(YLOG_FATAL|YLOG_ERRNO, "Failed to create comstack");
 729         exit(1);
 730     }
 731
 732     if (!(addr = cs_straddr(link, cl->database->host->ipport)))
 733     {
 734         yaz_log(YLOG_WARN|YLOG_ERRNO, "Lookup of IP address failed?");
 735         return 0;
 736     }
 737
 738     res = cs_connect(link, addr);
 739     if (res < 0)
 740     {
 741         yaz_log(YLOG_WARN|YLOG_ERRNO, "cs_connect %s", cl->database->url);
 742         return 0;
 743     }
 744
 745     if ((new = connection_freelist))
 746         connection_freelist = new->next;
 747     else
 748     {
 749         new = xmalloc(sizeof (struct connection));
 750         new->ibuf = 0;
 751         new->ibufsize = 0;
 752     }
 753     new->state = Conn_Connecting;
 754     new->host = cl->database->host;
 755     new->next = new->host->connections;
 756     new->host->connections = new;
 757     new->client = cl;
 758     cl->connection = new;
 759     new->link = link;
 760
 761     new->iochan = iochan_create(cs_fileno(link), handler, 0);
 762     iochan_setdata(new->iochan, new);
 763     new->iochan->next = channel_list;
 764     channel_list = new->iochan;
 765     return new;
 766 }
 767
 768 // Close connection and set state to error
 769 static void client_fatal(struct client *cl)
 770 {
 771     yaz_log(YLOG_WARN, "Fatal error from %s", cl->database->url);
 772     connection_destroy(cl->connection);
 773     cl->state = Client_Error;
 774 }
 775
 776 // Ensure that client has a connection associated
 777 static int client_prep_connection(struct client *cl)
 778 {
 779     struct connection *co;
 780     struct session *se = cl->session;
 781     struct host *host = cl->database->host;
 782
 783     co = cl->connection;
 784
 785     yaz_log(YLOG_DEBUG, "Client prep %s", cl->database->url);
 786
 787     if (!co)
 788     {
 789         // See if someone else has an idle connection
 790         // We should look at timestamps here to select the longest-idle connection
 791         for (co = host->connections; co; co = co->next)
 792             if (co->state == Conn_Open && (!co->client || co->client->session != se))
 793                 break;
 794         if (co)
 795         {
 796             connection_release(co);
 797             cl->connection = co;
 798             co->client = cl;
 799         }
 800         else
 801             co = connection_create(cl);
 802     }
 803     if (co)
 804     {
 805         if (co->state == Conn_Connecting)
 806         {
 807             cl->state = Client_Connecting;
 808             iochan_setflag(co->iochan, EVENT_OUTPUT);
 809         }
 810         else if (co->state == Conn_Open)
 811         {
 812             if (cl->state == Client_Error || cl->state == Client_Disconnected)
 813                 cl->state = Client_Idle;
 814             iochan_setflag(co->iochan, EVENT_OUTPUT);
 815         }
 816         return 1;
 817     }
 818     else
 819         return 0;
 820 }
 821
 822 // This function will most likely vanish when a proper target profile mechanism is
 823 // introduced.
 824 void load_simpletargets(const char *fn)
 825 {
 826     FILE *f = fopen(fn, "r");
 827     char line[256];
 828
 829     if (!f)
 830     {
 831         yaz_log(YLOG_WARN|YLOG_ERRNO, "open %s", fn);
 832         exit(1);
 833     }
 834
 835     while (fgets(line, 255, f))
 836     {
 837         char *url, *db;
 838         struct host *host;
 839         struct database *database;
 840
 841         if (strncmp(line, "target ", 7))
 842             continue;
 843         url = line + 7;
 844         url[strlen(url) - 1] = '\0';
 845         yaz_log(YLOG_DEBUG, "Target: %s", url);
 846         if ((db = strchr(url, '/')))
 847             *(db++) = '\0';
 848         else
 849             db = "Default";
 850
 851         for (host = hosts; host; host = host->next)
 852             if (!strcmp(url, host->hostport))
 853                 break;
 854         if (!host)
 855         {
 856             struct addrinfo *addrinfo, hints;
 857             char *port;
 858             char ipport[128];
 859             unsigned char addrbuf[4];
 860             int res;
 861
 862             host = xmalloc(sizeof(struct host));
 863             host->hostport = xstrdup(url);
 864             host->connections = 0;
 865
 866             if ((port = strchr(url, ':')))
 867                 *(port++) = '\0';
 868             else
 869                 port = "210";
 870
 871             hints.ai_flags = 0;
 872             hints.ai_family = PF_INET;
 873             hints.ai_socktype = SOCK_STREAM;
 874             hints.ai_protocol = IPPROTO_TCP;
 875             hints.ai_addrlen = 0;
 876             hints.ai_addr = 0;
 877             hints.ai_canonname = 0;
 878             hints.ai_next = 0;
 879             // This is not robust code. It assumes that getaddrinfo returns AF_INET
 880             // address.
 881             if ((res = getaddrinfo(url, port, &hints, &addrinfo)))
 882             {
 883                 yaz_log(YLOG_WARN, "Failed to resolve %s: %s", url, gai_strerror(res));
 884                 xfree(host->hostport);
 885                 xfree(host);
 886                 continue;
 887             }
 888             assert(addrinfo->ai_family == PF_INET);
 889             memcpy(addrbuf, &((struct sockaddr_in*)addrinfo->ai_addr)->sin_addr.s_addr, 4);
 890             sprintf(ipport, "%hhd.%hhd.%hhd.%hhd:%s",
 891                     addrbuf[0], addrbuf[1], addrbuf[2], addrbuf[3], port);
 892             host->ipport = xstrdup(ipport);
 893             freeaddrinfo(addrinfo);
 894             host->next = hosts;
 895             hosts = host;
 896         }
 897         database = xmalloc(sizeof(struct database));
 898         database->host = host;
 899         database->url = xmalloc(strlen(url) + strlen(db) + 2);
 900         strcpy(database->url, url);
 901         strcat(database->url, "/");
 902         strcat(database->url, db);
 903
 904         database->databases = xmalloc(2 * sizeof(char *));
 905         database->databases[0] = xstrdup(db);
 906         database->databases[1] = 0;
 907         database->errors = 0;
 908         database->qprofile = 0;
 909         database->rprofile = database_retrieval_profile(database);
 910         database->next = databases;
 911         databases = database;
 912
 913     }
 914     fclose(f);
 915 }
 916
 917 static void pull_terms(NMEM nmem, struct ccl_rpn_node *n, char **termlist, int *num)
 918 {
 919     switch (n->kind)
 920     {
 921         case CCL_RPN_AND:
 922         case CCL_RPN_OR:
 923         case CCL_RPN_NOT:
 924         case CCL_RPN_PROX:
 925             pull_terms(nmem, n->u.p[0], termlist, num);
 926             pull_terms(nmem, n->u.p[1], termlist, num);
 927             break;
 928         case CCL_RPN_TERM:
 929             termlist[(*num)++] = nmem_strdup(nmem, n->u.t.term);
 930             break;
 931         default: // NOOP
 932             break;
 933     }
 934 }
 935
 936 // Extract terms from query into null-terminated termlist
 937 static int extract_terms(NMEM nmem, char *query, char **termlist)
 938 {
 939     int error, pos;
 940     struct ccl_rpn_node *n;
 941     int num = 0;
 942
 943     n = ccl_find_str(global_parameters.ccl_filter, query, &error, &pos);
 944     if (!n)
 945         return -1;
 946     pull_terms(nmem, n, termlist, &num);
 947     termlist[num] = 0;
 948     ccl_rpn_delete(n);
 949     return 0;
 950 }
 951
 952 static struct client *client_create(void)
 953 {
 954     struct client *r;
 955     if (client_freelist)
 956     {
 957         r = client_freelist;
 958         client_freelist = client_freelist->next;
 959     }
 960     else
 961         r = xmalloc(sizeof(struct client));
 962     r->database = 0;
 963     r->connection = 0;
 964     r->session = 0;
 965     r->hits = 0;
 966     r->records = 0;
 967     r->setno = 0;
 968     r->requestid = -1;
 969     r->diagnostic = 0;
 970     r->state = Client_Disconnected;
 971     r->next = 0;
 972     return r;
 973 }
 974
 975 void client_destroy(struct client *c)
 976 {
 977     struct session *se = c->session;
 978     if (c == se->clients)
 979         se->clients = c->next;
 980     else
 981     {
 982         struct client *cc;
 983         for (cc = se->clients; cc && cc->next != c; cc = cc->next)
 984             ;
 985         if (cc)
 986             cc->next = c->next;
 987     }
 988     if (c->connection)
 989         connection_release(c->connection);
 990     c->next = client_freelist;
 991     client_freelist = c;
 992 }
 993
 994 void session_set_watch(struct session *s, int what, session_watchfun fun, void *data)
 995 {
 996     s->watchlist[what].fun = fun;
 997     s->watchlist[what].data = data;
 998 }
 999
1000 void session_alert_watch(struct session *s, int what)
1001 {
1002     if (!s->watchlist[what].fun)
1003         return;
1004     (*s->watchlist[what].fun)(s->watchlist[what].data);
1005     s->watchlist[what].fun = 0;
1006     s->watchlist[what].data = 0;
1007 }
1008
1009 // This needs to be extended with selection criteria
1010 static struct conf_retrievalprofile *database_retrieval_profile(struct database *db)
1011 {
1012     if (!config)
1013     {
1014         yaz_log(YLOG_FATAL, "Must load configuration (-f)");
1015         exit(1);
1016     }
1017     if (!config->retrievalprofiles)
1018     {
1019         yaz_log(YLOG_FATAL, "No retrieval profiles defined");
1020     }
1021     return config->retrievalprofiles;
1022 }
1023
1024 // This should be extended with parameters to control selection criteria
1025 // Associates a set of clients with a session;
1026 int select_targets(struct session *se)
1027 {
1028     struct database *db;
1029     int c = 0;
1030
1031     while (se->clients)
1032         client_destroy(se->clients);
1033     for (db = databases; db; db = db->next)
1034     {
1035         struct client *cl = client_create();
1036         cl->database = db;
1037         cl->session = se;
1038         cl->next = se->clients;
1039         se->clients = cl;
1040         c++;
1041     }
1042     return c;
1043 }
1044
1045 int session_active_clients(struct session *s)
1046 {
1047     struct client *c;
1048     int res = 0;
1049
1050     for (c = s->clients; c; c = c->next)
1051         if (c->connection && (c->state == Client_Connecting ||
1052                     c->state == Client_Initializing ||
1053                     c->state == Client_Searching ||
1054                     c->state == Client_Presenting))
1055             res++;
1056
1057     return res;
1058 }
1059
1060 char *search(struct session *se, char *query)
1061 {
1062     int live_channels = 0;
1063     struct client *cl;
1064
1065     yaz_log(YLOG_DEBUG, "Search");
1066
1067     strcpy(se->query, query);
1068     se->requestid++;
1069     nmem_reset(se->nmem);
1070     for (cl = se->clients; cl; cl = cl->next)
1071     {
1072         cl->hits = -1;
1073         cl->records = 0;
1074         cl->diagnostic = 0;
1075
1076         if (client_prep_connection(cl))
1077             live_channels++;
1078     }
1079     if (live_channels)
1080     {
1081         char *p[512];
1082         int maxrecs = live_channels * global_parameters.toget;
1083         se->num_termlists = 0;
1084         se->reclist = reclist_create(se->nmem, maxrecs);
1085         extract_terms(se->nmem, query, p);
1086         se->relevance = relevance_create(se->nmem, (const char **) p, maxrecs);
1087         se->total_records = se->total_hits = 0;
1088         se->expected_maxrecs = maxrecs;
1089     }
1090     else
1091         return "NOTARGETS";
1092
1093     return 0;
1094 }
1095
1096 void destroy_session(struct session *s)
1097 {
1098     yaz_log(YLOG_LOG, "Destroying session");
1099     while (s->clients)
1100         client_destroy(s->clients);
1101     nmem_destroy(s->nmem);
1102     wrbuf_free(s->wrbuf, 1);
1103 }
1104
1105 struct session *new_session()
1106 {
1107     int i;
1108     struct session *session = xmalloc(sizeof(*session));
1109
1110     yaz_log(YLOG_DEBUG, "New pazpar2 session");
1111
1112     session->total_hits = 0;
1113     session->total_records = 0;
1114     session->num_termlists = 0;
1115     session->reclist = 0;
1116     session->requestid = -1;
1117     session->clients = 0;
1118     session->expected_maxrecs = 0;
1119     session->query[0] = '\0';
1120     session->nmem = nmem_create();
1121     session->wrbuf = wrbuf_alloc();
1122     for (i = 0; i <= SESSION_WATCH_MAX; i++)
1123     {
1124         session->watchlist[i].data = 0;
1125         session->watchlist[i].fun = 0;
1126     }
1127
1128     select_targets(session);
1129
1130     return session;
1131 }
1132
1133 struct hitsbytarget *hitsbytarget(struct session *se, int *count)
1134 {
1135     static struct hitsbytarget res[1000]; // FIXME MM
1136     struct client *cl;
1137
1138     *count = 0;
1139     for (cl = se->clients; cl; cl = cl->next)
1140     {
1141         strcpy(res[*count].id, cl->database->host->hostport);
1142         res[*count].hits = cl->hits;
1143         res[*count].records = cl->records;
1144         res[*count].diagnostic = cl->diagnostic;
1145         res[*count].state = client_states[cl->state];
1146         res[*count].connected  = cl->connection ? 1 : 0;
1147         (*count)++;
1148     }
1149
1150     return res;
1151 }
1152
1153 struct termlist_score **termlist(struct session *s, const char *name, int *num)
1154 {
1155     int i;
1156
1157     for (i = 0; i < s->num_termlists; i++)
1158         if (!strcmp(s->termlists[i].name, name))
1159             return termlist_highscore(s->termlists[i].termlist, num);
1160     return 0;
1161 }
1162
1163 #ifdef MISSING_HEADERS
1164 void report_nmem_stats(void)
1165 {
1166     size_t in_use, is_free;
1167
1168     nmem_get_memory_in_use(&in_use);
1169     nmem_get_memory_free(&is_free);
1170
1171     yaz_log(YLOG_LOG, "nmem stat: use=%ld free=%ld",
1172             (long) in_use, (long) is_free);
1173 }
1174 #endif
1175
1176 struct record **show(struct session *s, int start, int *num, int *total,
1177                      int *sumhits, NMEM nmem_show)
1178 {
1179     struct record **recs = nmem_malloc(nmem_show, *num
1180                                        * sizeof(struct record *));
1181     int i;
1182 #if USE_TIMING
1183     yaz_timing_t t = yaz_timing_create();
1184 #endif
1185     relevance_prepare_read(s->relevance, s->reclist);
1186
1187     *total = s->reclist->num_records;
1188     *sumhits = s->total_hits;
1189
1190     for (i = 0; i < start; i++)
1191         if (!reclist_read_record(s->reclist))
1192         {
1193             *num = 0;
1194             recs = 0;
1195             break;
1196         }
1197
1198     for (i = 0; i < *num; i++)
1199     {
1200         struct record *r = reclist_read_record(s->reclist);
1201         if (!r)
1202         {
1203             *num = i;
1204             break;
1205         }
1206         recs[i] = r;
1207     }
1208 #if USE_TIMING
1209     yaz_timing_stop(t);
1210     yaz_log(YLOG_LOG, "show %6.5f %3.2f %3.2f",
1211             yaz_timing_get_real(t), yaz_timing_get_user(t),
1212             yaz_timing_get_sys(t));
1213     yaz_timing_destroy(&t);
1214 #endif
1215     return recs;
1216 }
1217
1218 void statistics(struct session *se, struct statistics *stat)
1219 {
1220     struct client *cl;
1221     int count = 0;
1222
1223     bzero(stat, sizeof(*stat));
1224     for (cl = se->clients; cl; cl = cl->next)
1225     {
1226         if (!cl->connection)
1227             stat->num_no_connection++;
1228         switch (cl->state)
1229         {
1230             case Client_Connecting: stat->num_connecting++; break;
1231             case Client_Initializing: stat->num_initializing++; break;
1232             case Client_Searching: stat->num_searching++; break;
1233             case Client_Presenting: stat->num_presenting++; break;
1234             case Client_Idle: stat->num_idle++; break;
1235             case Client_Failed: stat->num_failed++; break;
1236             case Client_Error: stat->num_error++; break;
1237             default: break;
1238         }
1239         count++;
1240     }
1241     stat->num_hits = se->total_hits;
1242     stat->num_records = se->total_records;
1243
1244     stat->num_clients = count;
1245 }
1246
1247 static CCL_bibset load_cclfile(const char *fn)
1248 {
1249     CCL_bibset res = ccl_qual_mk();
1250     if (ccl_qual_fname(res, fn) < 0)
1251     {
1252         yaz_log(YLOG_FATAL|YLOG_ERRNO, "%s", fn);
1253         exit(1);
1254     }
1255     return res;
1256 }
1257
1258 int main(int argc, char **argv)
1259 {
1260     int ret;
1261     char *arg;
1262     int setport = 0;
1263
1264     if (signal(SIGPIPE, SIG_IGN) < 0)
1265         yaz_log(YLOG_WARN|YLOG_ERRNO, "signal");
1266
1267     yaz_log_init(YLOG_DEFAULT_LEVEL, "pazpar2", 0);
1268
1269     while ((ret = options("f:x:c:h:p:C:s:d", argv, argc, &arg)) != -2)
1270     {
1271         switch (ret) {
1272             case 'f':
1273                 if (!read_config(arg))
1274                     exit(1);
1275                 break;
1276             case 'c':
1277                 command_init(atoi(arg));
1278                 setport++;
1279                 break;
1280             case 'h':
1281                 http_init(arg);
1282                 setport++;
1283                 break;
1284             case 'C':
1285                 global_parameters.ccl_filter = load_cclfile(arg);
1286                 break;
1287             case 'p':
1288                 http_set_proxyaddr(arg);
1289                 break;
1290             case 's':
1291                 load_simpletargets(arg);
1292                 break;
1293             case 'd':
1294                 global_parameters.dump_records = 1;
1295                 break;
1296             default:
1297                 fprintf(stderr, "Usage: pazpar2\n"
1298                         "    -f configfile\n"
1299                         "    -h [host:]port          (REST protocol listener)\n"
1300                         "    -c cmdport              (telnet-style)\n"
1301                         "    -C cclconfig\n"
1302                         "    -s simpletargetfile\n"
1303                         "    -p hostname[:portno]    (HTTP proxy)\n");
1304                 exit(1);
1305         }
1306     }
1307
1308     if (!setport)
1309     {
1310         fprintf(stderr, "Set command port with -h or -c\n");
1311         exit(1);
1312     }
1313
1314     global_parameters.ccl_filter = load_cclfile("../etc/default.bib");
1315     global_parameters.yaz_marc = yaz_marc_create();
1316     yaz_marc_subfield_str(global_parameters.yaz_marc, "\t");
1317     global_parameters.odr_in = odr_createmem(ODR_DECODE);
1318     global_parameters.odr_out = odr_createmem(ODR_ENCODE);
1319
1320     event_loop(&channel_list);
1321
1322     return 0;
1323 }
1324
1325 /*
1326  * Local variables:
1327  * c-basic-offset: 4
1328  * indent-tabs-mode: nil
1329  * End:
1330  * vim: shiftwidth=4 tabstop=8 expandtab
1331  */