src/pazpar2.c

   1 /* $Id: pazpar2.c,v 1.15 2007-01-05 20:33:05 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 #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         xmlFreeDoc(xdoc);
 435         return 0;
 436     }
 437
 438     res = nmem_malloc(se->nmem, sizeof(struct record));
 439     res->next_cluster = 0;
 440     res->target_offset = -1;
 441     res->term_frequency_vec = 0;
 442     res->title = "Unknown";
 443     res->relevance = 0;
 444
 445     mergekey_norm = nmem_strdup(se->nmem, (char*) mergekey);
 446     xmlFree(mergekey);
 447     res->merge_key = normalize_mergekey(mergekey_norm);
 448
 449     head = reclist_insert(se->reclist, res);
 450     if (!head)
 451     {
 452         /* no room for record */
 453         xmlFreeDoc(xdoc);
 454         return 0;
 455     }
 456     relevance_newrec(se->relevance, head);
 457
 458     for (n = root->children; n; n = n->next)
 459     {
 460         if (n->type != XML_ELEMENT_NODE)
 461             continue;
 462         if (!strcmp(n->name, "facet"))
 463         {
 464             xmlChar *type = xmlGetProp(n, "type");
 465             xmlChar *value = xmlNodeListGetString(xdoc, n->children, 0);
 466             if (type && value)
 467             {
 468                 add_facet(se, type, value);
 469                 relevance_countwords(se->relevance, head, value, 1);
 470             }
 471             xmlFree(type);
 472             xmlFree(value);
 473         }
 474         else if (!strcmp(n->name, "metadata"))
 475         {
 476             xmlChar *type = xmlGetProp(n, "type");
 477             if (type && !strcmp(type, "title"))
 478             {
 479                 xmlChar *value = xmlNodeListGetString(xdoc, n->children, 0);
 480                 if (value)
 481                 {
 482                     res->title = nmem_strdup(se->nmem, value);
 483                     relevance_countwords(se->relevance, head, value, 4);
 484                     xmlFree(value);
 485                 }
 486             }
 487             xmlFree(type);
 488         }
 489         else
 490             yaz_log(YLOG_WARN, "Unexpected element %s in internal record", n->name);
 491     }
 492
 493     xmlFreeDoc(xdoc);
 494
 495     relevance_donerecord(se->relevance, head);
 496     se->total_records++;
 497
 498     return res;
 499 }
 500
 501 static void ingest_records(struct client *cl, Z_Records *r)
 502 {
 503 #if USE_TIMING
 504     yaz_timing_t t = yaz_timing_create();
 505 #endif
 506     struct record *rec;
 507     struct session *s = cl->session;
 508     Z_NamePlusRecordList *rlist;
 509     int i;
 510
 511     if (r->which != Z_Records_DBOSD)
 512         return;
 513     rlist = r->u.databaseOrSurDiagnostics;
 514     for (i = 0; i < rlist->num_records; i++)
 515     {
 516         Z_NamePlusRecord *npr = rlist->records[i];
 517
 518         if (npr->which != Z_NamePlusRecord_databaseRecord)
 519         {
 520             yaz_log(YLOG_WARN, "Unexpected record type, probably diagnostic");
 521             continue;
 522         }
 523
 524         rec = ingest_record(cl, npr->u.databaseRecord);
 525         if (!rec)
 526             continue;
 527     }
 528     if (s->watchlist[SESSION_WATCH_RECORDS].fun && rlist->num_records)
 529         session_alert_watch(s, SESSION_WATCH_RECORDS);
 530
 531 #if USE_TIMING
 532     yaz_timing_stop(t);
 533     yaz_log(YLOG_LOG, "ingest_records %6.5f %3.2f %3.2f",
 534             yaz_timing_get_real(t), yaz_timing_get_user(t),
 535             yaz_timing_get_sys(t));
 536     yaz_timing_destroy(&t);
 537 #endif
 538 }
 539
 540 static void do_presentResponse(IOCHAN i, Z_APDU *a)
 541 {
 542     struct connection *co = iochan_getdata(i);
 543     struct client *cl = co->client;
 544     Z_PresentResponse *r = a->u.presentResponse;
 545
 546     if (r->records) {
 547         Z_Records *recs = r->records;
 548         if (recs->which == Z_Records_NSD)
 549         {
 550             yaz_log(YLOG_WARN, "Non-surrogate diagnostic");
 551             cl->diagnostic = *recs->u.nonSurrogateDiagnostic->condition;
 552             cl->state = Client_Error;
 553         }
 554     }
 555
 556     if (!*r->presentStatus && cl->state != Client_Error)
 557     {
 558         yaz_log(YLOG_DEBUG, "Good Present response");
 559         cl->records += *r->numberOfRecordsReturned;
 560         ingest_records(cl, r->records);
 561         cl->state = Client_Idle;
 562     }
 563     else if (*r->presentStatus)
 564     {
 565         yaz_log(YLOG_WARN, "Bad Present response");
 566         cl->state = Client_Error;
 567     }
 568 }
 569
 570 static void handler(IOCHAN i, int event)
 571 {
 572     struct connection *co = iochan_getdata(i);
 573     struct client *cl = co->client;
 574     struct session *se = 0;
 575
 576     if (cl)
 577         se = cl->session;
 578     else
 579     {
 580         yaz_log(YLOG_WARN, "Destroying orphan connection");
 581         connection_destroy(co);
 582         return;
 583     }
 584
 585     if (co->state == Conn_Connecting && event & EVENT_OUTPUT)
 586     {
 587         int errcode;
 588         socklen_t errlen = sizeof(errcode);
 589
 590         if (getsockopt(cs_fileno(co->link), SOL_SOCKET, SO_ERROR, &errcode,
 591             &errlen) < 0 || errcode != 0)
 592         {
 593             client_fatal(cl);
 594             return;
 595         }
 596         else
 597         {
 598             yaz_log(YLOG_DEBUG, "Connect OK");
 599             co->state = Conn_Open;
 600             if (cl)
 601                 cl->state = Client_Connected;
 602         }
 603     }
 604
 605     else if (event & EVENT_INPUT)
 606     {
 607         int len = cs_get(co->link, &co->ibuf, &co->ibufsize);
 608
 609         if (len < 0)
 610         {
 611             yaz_log(YLOG_WARN|YLOG_ERRNO, "Error reading from Z server");
 612             connection_destroy(co);
 613             return;
 614         }
 615         else if (len == 0)
 616         {
 617             yaz_log(YLOG_WARN, "EOF reading from Z server");
 618             connection_destroy(co);
 619             return;
 620         }
 621         else if (len > 1) // We discard input if we have no connection
 622         {
 623             co->state = Conn_Open;
 624
 625             if (cl && (cl->requestid == se->requestid || cl->state == Client_Initializing))
 626             {
 627                 Z_APDU *a;
 628
 629                 odr_reset(global_parameters.odr_in);
 630                 odr_setbuf(global_parameters.odr_in, co->ibuf, len, 0);
 631                 if (!z_APDU(global_parameters.odr_in, &a, 0, 0))
 632                 {
 633                     client_fatal(cl);
 634                     return;
 635                 }
 636                 switch (a->which)
 637                 {
 638                     case Z_APDU_initResponse:
 639                         do_initResponse(i, a);
 640                         break;
 641                     case Z_APDU_searchResponse:
 642                         do_searchResponse(i, a);
 643                         break;
 644                     case Z_APDU_presentResponse:
 645                         do_presentResponse(i, a);
 646                         break;
 647                     default:
 648                         yaz_log(YLOG_WARN, "Unexpected result from server");
 649                         client_fatal(cl);
 650                         return;
 651                 }
 652                 // We aren't expecting staggered output from target
 653                 // if (cs_more(t->link))
 654                 //    iochan_setevent(i, EVENT_INPUT);
 655             }
 656             else  // we throw away response and go to idle mode
 657             {
 658                 yaz_log(YLOG_DEBUG, "Ignoring result of expired operation");
 659                 cl->state = Client_Idle;
 660             }
 661         }
 662         /* if len==1 we do nothing but wait for more input */
 663     }
 664
 665     if (cl->state == Client_Connected) {
 666         send_init(i);
 667     }
 668
 669     if (cl->state == Client_Idle)
 670     {
 671         if (cl->requestid != se->requestid && *se->query) {
 672             send_search(i);
 673         }
 674         else if (cl->hits > 0 && cl->records < global_parameters.toget &&
 675             cl->records < cl->hits) {
 676             send_present(i);
 677         }
 678     }
 679 }
 680
 681 // Disassociate connection from client
 682 static void connection_release(struct connection *co)
 683 {
 684     struct client *cl = co->client;
 685
 686     yaz_log(YLOG_DEBUG, "Connection release %s", co->host->hostport);
 687     if (!cl)
 688         return;
 689     cl->connection = 0;
 690     co->client = 0;
 691 }
 692
 693 // Close connection and recycle structure
 694 static void connection_destroy(struct connection *co)
 695 {
 696     struct host *h = co->host;
 697     cs_close(co->link);
 698     iochan_destroy(co->iochan);
 699
 700     yaz_log(YLOG_DEBUG, "Connection destroy %s", co->host->hostport);
 701     if (h->connections == co)
 702         h->connections = co->next;
 703     else
 704     {
 705         struct connection *pco;
 706         for (pco = h->connections; pco && pco->next != co; pco = pco->next)
 707             ;
 708         if (pco)
 709             pco->next = co->next;
 710         else
 711             abort();
 712     }
 713     if (co->client)
 714     {
 715         if (co->client->state != Client_Idle)
 716             co->client->state = Client_Disconnected;
 717         co->client->connection = 0;
 718     }
 719     co->next = connection_freelist;
 720     connection_freelist = co;
 721 }
 722
 723 // Creates a new connection for client, associated with the host of
 724 // client's database
 725 static struct connection *connection_create(struct client *cl)
 726 {
 727     struct connection *new;
 728     COMSTACK link;
 729     int res;
 730     void *addr;
 731
 732     yaz_log(YLOG_DEBUG, "Connection create %s", cl->database->url);
 733     if (!(link = cs_create(tcpip_type, 0, PROTO_Z3950)))
 734     {
 735         yaz_log(YLOG_FATAL|YLOG_ERRNO, "Failed to create comstack");
 736         exit(1);
 737     }
 738
 739     if (!(addr = cs_straddr(link, cl->database->host->ipport)))
 740     {
 741         yaz_log(YLOG_WARN|YLOG_ERRNO, "Lookup of IP address failed?");
 742         return 0;
 743     }
 744
 745     res = cs_connect(link, addr);
 746     if (res < 0)
 747     {
 748         yaz_log(YLOG_WARN|YLOG_ERRNO, "cs_connect %s", cl->database->url);
 749         return 0;
 750     }
 751
 752     if ((new = connection_freelist))
 753         connection_freelist = new->next;
 754     else
 755     {
 756         new = xmalloc(sizeof (struct connection));
 757         new->ibuf = 0;
 758         new->ibufsize = 0;
 759     }
 760     new->state = Conn_Connecting;
 761     new->host = cl->database->host;
 762     new->next = new->host->connections;
 763     new->host->connections = new;
 764     new->client = cl;
 765     cl->connection = new;
 766     new->link = link;
 767
 768     new->iochan = iochan_create(cs_fileno(link), handler, 0);
 769     iochan_setdata(new->iochan, new);
 770     new->iochan->next = channel_list;
 771     channel_list = new->iochan;
 772     return new;
 773 }
 774
 775 // Close connection and set state to error
 776 static void client_fatal(struct client *cl)
 777 {
 778     yaz_log(YLOG_WARN, "Fatal error from %s", cl->database->url);
 779     connection_destroy(cl->connection);
 780     cl->state = Client_Error;
 781 }
 782
 783 // Ensure that client has a connection associated
 784 static int client_prep_connection(struct client *cl)
 785 {
 786     struct connection *co;
 787     struct session *se = cl->session;
 788     struct host *host = cl->database->host;
 789
 790     co = cl->connection;
 791
 792     yaz_log(YLOG_DEBUG, "Client prep %s", cl->database->url);
 793
 794     if (!co)
 795     {
 796         // See if someone else has an idle connection
 797         // We should look at timestamps here to select the longest-idle connection
 798         for (co = host->connections; co; co = co->next)
 799             if (co->state == Conn_Open && (!co->client || co->client->session != se))
 800                 break;
 801         if (co)
 802         {
 803             connection_release(co);
 804             cl->connection = co;
 805             co->client = cl;
 806         }
 807         else
 808             co = connection_create(cl);
 809     }
 810     if (co)
 811     {
 812         if (co->state == Conn_Connecting)
 813         {
 814             cl->state = Client_Connecting;
 815             iochan_setflag(co->iochan, EVENT_OUTPUT);
 816         }
 817         else if (co->state == Conn_Open)
 818         {
 819             if (cl->state == Client_Error || cl->state == Client_Disconnected)
 820                 cl->state = Client_Idle;
 821             iochan_setflag(co->iochan, EVENT_OUTPUT);
 822         }
 823         return 1;
 824     }
 825     else
 826         return 0;
 827 }
 828
 829 // This function will most likely vanish when a proper target profile mechanism is
 830 // introduced.
 831 void load_simpletargets(const char *fn)
 832 {
 833     FILE *f = fopen(fn, "r");
 834     char line[256];
 835
 836     if (!f)
 837     {
 838         yaz_log(YLOG_WARN|YLOG_ERRNO, "open %s", fn);
 839         exit(1);
 840     }
 841
 842     while (fgets(line, 255, f))
 843     {
 844         char *url, *db;
 845         struct host *host;
 846         struct database *database;
 847
 848         if (strncmp(line, "target ", 7))
 849             continue;
 850         url = line + 7;
 851         url[strlen(url) - 1] = '\0';
 852         yaz_log(YLOG_DEBUG, "Target: %s", url);
 853         if ((db = strchr(url, '/')))
 854             *(db++) = '\0';
 855         else
 856             db = "Default";
 857
 858         for (host = hosts; host; host = host->next)
 859             if (!strcmp(url, host->hostport))
 860                 break;
 861         if (!host)
 862         {
 863             struct addrinfo *addrinfo, hints;
 864             char *port;
 865             char ipport[128];
 866             unsigned char addrbuf[4];
 867             int res;
 868
 869             host = xmalloc(sizeof(struct host));
 870             host->hostport = xstrdup(url);
 871             host->connections = 0;
 872
 873             if ((port = strchr(url, ':')))
 874                 *(port++) = '\0';
 875             else
 876                 port = "210";
 877
 878             hints.ai_flags = 0;
 879             hints.ai_family = PF_INET;
 880             hints.ai_socktype = SOCK_STREAM;
 881             hints.ai_protocol = IPPROTO_TCP;
 882             hints.ai_addrlen = 0;
 883             hints.ai_addr = 0;
 884             hints.ai_canonname = 0;
 885             hints.ai_next = 0;
 886             // This is not robust code. It assumes that getaddrinfo returns AF_INET
 887             // address.
 888             if ((res = getaddrinfo(url, port, &hints, &addrinfo)))
 889             {
 890                 yaz_log(YLOG_WARN, "Failed to resolve %s: %s", url, gai_strerror(res));
 891                 xfree(host->hostport);
 892                 xfree(host);
 893                 continue;
 894             }
 895             assert(addrinfo->ai_family == PF_INET);
 896             memcpy(addrbuf, &((struct sockaddr_in*)addrinfo->ai_addr)->sin_addr.s_addr, 4);
 897             sprintf(ipport, "%hhd.%hhd.%hhd.%hhd:%s",
 898                     addrbuf[0], addrbuf[1], addrbuf[2], addrbuf[3], port);
 899             host->ipport = xstrdup(ipport);
 900             freeaddrinfo(addrinfo);
 901             host->next = hosts;
 902             hosts = host;
 903         }
 904         database = xmalloc(sizeof(struct database));
 905         database->host = host;
 906         database->url = xmalloc(strlen(url) + strlen(db) + 2);
 907         strcpy(database->url, url);
 908         strcat(database->url, "/");
 909         strcat(database->url, db);
 910
 911         database->databases = xmalloc(2 * sizeof(char *));
 912         database->databases[0] = xstrdup(db);
 913         database->databases[1] = 0;
 914         database->errors = 0;
 915         database->qprofile = 0;
 916         database->rprofile = database_retrieval_profile(database);
 917         database->next = databases;
 918         databases = database;
 919
 920     }
 921     fclose(f);
 922 }
 923
 924 static void pull_terms(NMEM nmem, struct ccl_rpn_node *n, char **termlist, int *num)
 925 {
 926     switch (n->kind)
 927     {
 928         case CCL_RPN_AND:
 929         case CCL_RPN_OR:
 930         case CCL_RPN_NOT:
 931         case CCL_RPN_PROX:
 932             pull_terms(nmem, n->u.p[0], termlist, num);
 933             pull_terms(nmem, n->u.p[1], termlist, num);
 934             break;
 935         case CCL_RPN_TERM:
 936             termlist[(*num)++] = nmem_strdup(nmem, n->u.t.term);
 937             break;
 938         default: // NOOP
 939             break;
 940     }
 941 }
 942
 943 // Extract terms from query into null-terminated termlist
 944 static int extract_terms(NMEM nmem, char *query, char **termlist)
 945 {
 946     int error, pos;
 947     struct ccl_rpn_node *n;
 948     int num = 0;
 949
 950     n = ccl_find_str(global_parameters.ccl_filter, query, &error, &pos);
 951     if (!n)
 952         return -1;
 953     pull_terms(nmem, n, termlist, &num);
 954     termlist[num] = 0;
 955     ccl_rpn_delete(n);
 956     return 0;
 957 }
 958
 959 static struct client *client_create(void)
 960 {
 961     struct client *r;
 962     if (client_freelist)
 963     {
 964         r = client_freelist;
 965         client_freelist = client_freelist->next;
 966     }
 967     else
 968         r = xmalloc(sizeof(struct client));
 969     r->database = 0;
 970     r->connection = 0;
 971     r->session = 0;
 972     r->hits = 0;
 973     r->records = 0;
 974     r->setno = 0;
 975     r->requestid = -1;
 976     r->diagnostic = 0;
 977     r->state = Client_Disconnected;
 978     r->next = 0;
 979     return r;
 980 }
 981
 982 void client_destroy(struct client *c)
 983 {
 984     struct session *se = c->session;
 985     if (c == se->clients)
 986         se->clients = c->next;
 987     else
 988     {
 989         struct client *cc;
 990         for (cc = se->clients; cc && cc->next != c; cc = cc->next)
 991             ;
 992         if (cc)
 993             cc->next = c->next;
 994     }
 995     if (c->connection)
 996         connection_release(c->connection);
 997     c->next = client_freelist;
 998     client_freelist = c;
 999 }
1000
1001 void session_set_watch(struct session *s, int what, session_watchfun fun, void *data)
1002 {
1003     s->watchlist[what].fun = fun;
1004     s->watchlist[what].data = data;
1005 }
1006
1007 void session_alert_watch(struct session *s, int what)
1008 {
1009     if (!s->watchlist[what].fun)
1010         return;
1011     (*s->watchlist[what].fun)(s->watchlist[what].data);
1012     s->watchlist[what].fun = 0;
1013     s->watchlist[what].data = 0;
1014 }
1015
1016 // This needs to be extended with selection criteria
1017 static struct conf_retrievalprofile *database_retrieval_profile(struct database *db)
1018 {
1019     if (!config)
1020     {
1021         yaz_log(YLOG_FATAL, "Must load configuration (-f)");
1022         exit(1);
1023     }
1024     if (!config->retrievalprofiles)
1025     {
1026         yaz_log(YLOG_FATAL, "No retrieval profiles defined");
1027     }
1028     return config->retrievalprofiles;
1029 }
1030
1031 // This should be extended with parameters to control selection criteria
1032 // Associates a set of clients with a session;
1033 int select_targets(struct session *se)
1034 {
1035     struct database *db;
1036     int c = 0;
1037
1038     while (se->clients)
1039         client_destroy(se->clients);
1040     for (db = databases; db; db = db->next)
1041     {
1042         struct client *cl = client_create();
1043         cl->database = db;
1044         cl->session = se;
1045         cl->next = se->clients;
1046         se->clients = cl;
1047         c++;
1048     }
1049     return c;
1050 }
1051
1052 int session_active_clients(struct session *s)
1053 {
1054     struct client *c;
1055     int res = 0;
1056
1057     for (c = s->clients; c; c = c->next)
1058         if (c->connection && (c->state == Client_Connecting ||
1059                     c->state == Client_Initializing ||
1060                     c->state == Client_Searching ||
1061                     c->state == Client_Presenting))
1062             res++;
1063
1064     return res;
1065 }
1066
1067 char *search(struct session *se, char *query)
1068 {
1069     int live_channels = 0;
1070     struct client *cl;
1071
1072     yaz_log(YLOG_DEBUG, "Search");
1073
1074     strcpy(se->query, query);
1075     se->requestid++;
1076     nmem_reset(se->nmem);
1077     for (cl = se->clients; cl; cl = cl->next)
1078     {
1079         cl->hits = -1;
1080         cl->records = 0;
1081         cl->diagnostic = 0;
1082
1083         if (client_prep_connection(cl))
1084             live_channels++;
1085     }
1086     if (live_channels)
1087     {
1088         char *p[512];
1089         int maxrecs = live_channels * global_parameters.toget;
1090         se->num_termlists = 0;
1091         se->reclist = reclist_create(se->nmem, maxrecs);
1092         extract_terms(se->nmem, query, p);
1093         se->relevance = relevance_create(se->nmem, (const char **) p, maxrecs);
1094         se->total_records = se->total_hits = 0;
1095         se->expected_maxrecs = maxrecs;
1096     }
1097     else
1098         return "NOTARGETS";
1099
1100     return 0;
1101 }
1102
1103 void destroy_session(struct session *s)
1104 {
1105     yaz_log(YLOG_LOG, "Destroying session");
1106     while (s->clients)
1107         client_destroy(s->clients);
1108     nmem_destroy(s->nmem);
1109     wrbuf_free(s->wrbuf, 1);
1110 }
1111
1112 struct session *new_session()
1113 {
1114     int i;
1115     struct session *session = xmalloc(sizeof(*session));
1116
1117     yaz_log(YLOG_DEBUG, "New pazpar2 session");
1118
1119     session->total_hits = 0;
1120     session->total_records = 0;
1121     session->num_termlists = 0;
1122     session->reclist = 0;
1123     session->requestid = -1;
1124     session->clients = 0;
1125     session->expected_maxrecs = 0;
1126     session->query[0] = '\0';
1127     session->nmem = nmem_create();
1128     session->wrbuf = wrbuf_alloc();
1129     for (i = 0; i <= SESSION_WATCH_MAX; i++)
1130     {
1131         session->watchlist[i].data = 0;
1132         session->watchlist[i].fun = 0;
1133     }
1134
1135     select_targets(session);
1136
1137     return session;
1138 }
1139
1140 struct hitsbytarget *hitsbytarget(struct session *se, int *count)
1141 {
1142     static struct hitsbytarget res[1000]; // FIXME MM
1143     struct client *cl;
1144
1145     *count = 0;
1146     for (cl = se->clients; cl; cl = cl->next)
1147     {
1148         strcpy(res[*count].id, cl->database->host->hostport);
1149         res[*count].hits = cl->hits;
1150         res[*count].records = cl->records;
1151         res[*count].diagnostic = cl->diagnostic;
1152         res[*count].state = client_states[cl->state];
1153         res[*count].connected  = cl->connection ? 1 : 0;
1154         (*count)++;
1155     }
1156
1157     return res;
1158 }
1159
1160 struct termlist_score **termlist(struct session *s, const char *name, int *num)
1161 {
1162     int i;
1163
1164     for (i = 0; i < s->num_termlists; i++)
1165         if (!strcmp(s->termlists[i].name, name))
1166             return termlist_highscore(s->termlists[i].termlist, num);
1167     return 0;
1168 }
1169
1170 #ifdef MISSING_HEADERS
1171 void report_nmem_stats(void)
1172 {
1173     size_t in_use, is_free;
1174
1175     nmem_get_memory_in_use(&in_use);
1176     nmem_get_memory_free(&is_free);
1177
1178     yaz_log(YLOG_LOG, "nmem stat: use=%ld free=%ld",
1179             (long) in_use, (long) is_free);
1180 }
1181 #endif
1182
1183 struct record **show(struct session *s, int start, int *num, int *total,
1184                      int *sumhits, NMEM nmem_show)
1185 {
1186     struct record **recs = nmem_malloc(nmem_show, *num
1187                                        * sizeof(struct record *));
1188     int i;
1189 #if USE_TIMING
1190     yaz_timing_t t = yaz_timing_create();
1191 #endif
1192     relevance_prepare_read(s->relevance, s->reclist);
1193
1194     *total = s->reclist->num_records;
1195     *sumhits = s->total_hits;
1196
1197     for (i = 0; i < start; i++)
1198         if (!reclist_read_record(s->reclist))
1199         {
1200             *num = 0;
1201             recs = 0;
1202             break;
1203         }
1204
1205     for (i = 0; i < *num; i++)
1206     {
1207         struct record *r = reclist_read_record(s->reclist);
1208         if (!r)
1209         {
1210             *num = i;
1211             break;
1212         }
1213         recs[i] = r;
1214     }
1215 #if USE_TIMING
1216     yaz_timing_stop(t);
1217     yaz_log(YLOG_LOG, "show %6.5f %3.2f %3.2f",
1218             yaz_timing_get_real(t), yaz_timing_get_user(t),
1219             yaz_timing_get_sys(t));
1220     yaz_timing_destroy(&t);
1221 #endif
1222     return recs;
1223 }
1224
1225 void statistics(struct session *se, struct statistics *stat)
1226 {
1227     struct client *cl;
1228     int count = 0;
1229
1230     bzero(stat, sizeof(*stat));
1231     for (cl = se->clients; cl; cl = cl->next)
1232     {
1233         if (!cl->connection)
1234             stat->num_no_connection++;
1235         switch (cl->state)
1236         {
1237             case Client_Connecting: stat->num_connecting++; break;
1238             case Client_Initializing: stat->num_initializing++; break;
1239             case Client_Searching: stat->num_searching++; break;
1240             case Client_Presenting: stat->num_presenting++; break;
1241             case Client_Idle: stat->num_idle++; break;
1242             case Client_Failed: stat->num_failed++; break;
1243             case Client_Error: stat->num_error++; break;
1244             default: break;
1245         }
1246         count++;
1247     }
1248     stat->num_hits = se->total_hits;
1249     stat->num_records = se->total_records;
1250
1251     stat->num_clients = count;
1252 }
1253
1254 static CCL_bibset load_cclfile(const char *fn)
1255 {
1256     CCL_bibset res = ccl_qual_mk();
1257     if (ccl_qual_fname(res, fn) < 0)
1258     {
1259         yaz_log(YLOG_FATAL|YLOG_ERRNO, "%s", fn);
1260         exit(1);
1261     }
1262     return res;
1263 }
1264
1265 int main(int argc, char **argv)
1266 {
1267     int ret;
1268     char *arg;
1269     int setport = 0;
1270
1271     if (signal(SIGPIPE, SIG_IGN) < 0)
1272         yaz_log(YLOG_WARN|YLOG_ERRNO, "signal");
1273
1274     yaz_log_init(YLOG_DEFAULT_LEVEL, "pazpar2", 0);
1275
1276     while ((ret = options("f:x:c:h:p:C:s:d", argv, argc, &arg)) != -2)
1277     {
1278         switch (ret) {
1279             case 'f':
1280                 if (!read_config(arg))
1281                     exit(1);
1282                 break;
1283             case 'c':
1284                 command_init(atoi(arg));
1285                 setport++;
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 cmdport              (telnet-style)\n"
1308                         "    -C cclconfig\n"
1309                         "    -s simpletargetfile\n"
1310                         "    -p hostname[:portno]    (HTTP proxy)\n");
1311                 exit(1);
1312         }
1313     }
1314
1315     if (!setport)
1316     {
1317         fprintf(stderr, "Set command port with -h or -c\n");
1318         exit(1);
1319     }
1320
1321     global_parameters.ccl_filter = load_cclfile("../etc/default.bib");
1322     global_parameters.yaz_marc = yaz_marc_create();
1323     yaz_marc_subfield_str(global_parameters.yaz_marc, "\t");
1324     global_parameters.odr_in = odr_createmem(ODR_DECODE);
1325     global_parameters.odr_out = odr_createmem(ODR_ENCODE);
1326
1327     event_loop(&channel_list);
1328
1329     return 0;
1330 }
1331
1332 /*
1333  * Local variables:
1334  * c-basic-offset: 4
1335  * indent-tabs-mode: nil
1336  * End:
1337  * vim: shiftwidth=4 tabstop=8 expandtab
1338  */