src/pazpar2.c

   1 /* $Id: pazpar2.c,v 1.13 2007-01-04 20:00:58 quinn Exp $ */;
   2
   3 #include <stdlib.h>
   4 #include <stdio.h>
   5 #include <string.h>
   6 #include <sys/time.h>
   7 #include <unistd.h>
   8 #include <sys/socket.h>
   9 #include <netdb.h>
  10 #include <signal.h>
  11 #include <ctype.h>
  12 #include <assert.h>
  13
  14 #include <yaz/comstack.h>
  15 #include <yaz/tcpip.h>
  16 #include <yaz/proto.h>
  17 #include <yaz/readconf.h>
  18 #include <yaz/pquery.h>
  19 #include <yaz/yaz-util.h>
  20 #include <yaz/nmem.h>
  21
  22 #define USE_TIMING 0
  23 #if USE_TIMING
  24 #include <yaz/timing.h>
  25 #endif
  26
  27 #include "pazpar2.h"
  28 #include "eventl.h"
  29 #include "command.h"
  30 #include "http.h"
  31 #include "termlists.h"
  32 #include "reclists.h"
  33 #include "relevance.h"
  34 #include "config.h"
  35
  36 #define PAZPAR2_VERSION "0.1"
  37 #define MAX_CHUNK 15
  38
  39 static void client_fatal(struct client *cl);
  40 static void connection_destroy(struct connection *co);
  41 static int client_prep_connection(struct client *cl);
  42 static void ingest_records(struct client *cl, Z_Records *r);
  43 static struct conf_retrievalprofile *database_retrieval_profile(struct database *db);
  44 void session_alert_watch(struct session *s, int what);
  45
  46 IOCHAN channel_list = 0;  // Master list of connections we're handling events to
  47
  48 static struct connection *connection_freelist = 0;
  49 static struct client *client_freelist = 0;
  50
  51 static struct host *hosts = 0;  // The hosts we know about
  52 static struct database *databases = 0; // The databases we know about
  53
  54 static char *client_states[] = {
  55     "Client_Connecting",
  56     "Client_Connected",
  57     "Client_Idle",
  58     "Client_Initializing",
  59     "Client_Searching",
  60     "Client_Presenting",
  61     "Client_Error",
  62     "Client_Failed",
  63     "Client_Disconnected",
  64     "Client_Stopped"
  65 };
  66
  67 struct parameters global_parameters =
  68 {
  69     0,
  70     30,
  71     "81",
  72     "Index Data PazPar2 (MasterKey)",
  73     PAZPAR2_VERSION,
  74     600, // 10 minutes
  75     60,
  76     100,
  77     MAX_CHUNK,
  78     0,
  79     0,
  80     0,
  81     0
  82 };
  83
  84 static int send_apdu(struct client *c, Z_APDU *a)
  85 {
  86     struct connection *co = c->connection;
  87     char *buf;
  88     int len, r;
  89
  90     if (!z_APDU(global_parameters.odr_out, &a, 0, 0))
  91     {
  92         odr_perror(global_parameters.odr_out, "Encoding APDU");
  93         abort();
  94     }
  95     buf = odr_getbuf(global_parameters.odr_out, &len, 0);
  96     r = cs_put(co->link, buf, len);
  97     if (r < 0)
  98     {
  99         yaz_log(YLOG_WARN, "cs_put: %s", cs_errmsg(cs_errno(co->link)));
 100         return -1;
 101     }
 102     else if (r == 1)
 103     {
 104         fprintf(stderr, "cs_put incomplete (ParaZ does not handle that)\n");
 105         exit(1);
 106     }
 107     odr_reset(global_parameters.odr_out); /* release the APDU structure  */
 108     co->state = Conn_Waiting;
 109     return 0;
 110 }
 111
 112
 113 static void send_init(IOCHAN i)
 114 {
 115     struct connection *co = iochan_getdata(i);
 116     struct client *cl = co->client;
 117     Z_APDU *a = zget_APDU(global_parameters.odr_out, Z_APDU_initRequest);
 118
 119     a->u.initRequest->implementationId = global_parameters.implementationId;
 120     a->u.initRequest->implementationName = global_parameters.implementationName;
 121     a->u.initRequest->implementationVersion =
 122         global_parameters.implementationVersion;
 123     ODR_MASK_SET(a->u.initRequest->options, Z_Options_search);
 124     ODR_MASK_SET(a->u.initRequest->options, Z_Options_present);
 125     ODR_MASK_SET(a->u.initRequest->options, Z_Options_namedResultSets);
 126
 127     ODR_MASK_SET(a->u.initRequest->protocolVersion, Z_ProtocolVersion_1);
 128     ODR_MASK_SET(a->u.initRequest->protocolVersion, Z_ProtocolVersion_2);
 129     ODR_MASK_SET(a->u.initRequest->protocolVersion, Z_ProtocolVersion_3);
 130     if (send_apdu(cl, a) >= 0)
 131     {
 132         iochan_setflags(i, EVENT_INPUT);
 133         cl->state = Client_Initializing;
 134     }
 135     else
 136         cl->state = Client_Error;
 137     odr_reset(global_parameters.odr_out);
 138 }
 139
 140 static void send_search(IOCHAN i)
 141 {
 142     struct connection *co = iochan_getdata(i);
 143     struct client *cl = co->client;
 144     struct session *se = cl->session;
 145     struct database *db = cl->database;
 146     Z_APDU *a = zget_APDU(global_parameters.odr_out, Z_APDU_searchRequest);
 147     int ndb, cerror, cpos;
 148     char **databaselist;
 149     Z_Query *zquery;
 150     struct ccl_rpn_node *cn;
 151     int ssub = 0, lslb = 100000, mspn = 10;
 152
 153     yaz_log(YLOG_DEBUG, "Sending search");
 154
 155     cn = ccl_find_str(global_parameters.ccl_filter, se->query, &cerror, &cpos);
 156     if (!cn)
 157         return;
 158     a->u.searchRequest->query = zquery = odr_malloc(global_parameters.odr_out,
 159             sizeof(Z_Query));
 160     zquery->which = Z_Query_type_1;
 161     zquery->u.type_1 = ccl_rpn_query(global_parameters.odr_out, cn);
 162     ccl_rpn_delete(cn);
 163
 164     for (ndb = 0; db->databases[ndb]; ndb++)
 165         ;
 166     databaselist = odr_malloc(global_parameters.odr_out, sizeof(char*) * ndb);
 167     for (ndb = 0; db->databases[ndb]; ndb++)
 168         databaselist[ndb] = db->databases[ndb];
 169
 170     a->u.presentRequest->preferredRecordSyntax =
 171             yaz_oidval_to_z3950oid(global_parameters.odr_out,
 172             CLASS_RECSYN, VAL_USMARC);
 173     a->u.searchRequest->smallSetUpperBound = &ssub;
 174     a->u.searchRequest->largeSetLowerBound = &lslb;
 175     a->u.searchRequest->mediumSetPresentNumber = &mspn;
 176     a->u.searchRequest->resultSetName = "Default";
 177     a->u.searchRequest->databaseNames = databaselist;
 178     a->u.searchRequest->num_databaseNames = ndb;
 179
 180     if (send_apdu(cl, a) >= 0)
 181     {
 182         iochan_setflags(i, EVENT_INPUT);
 183         cl->state = Client_Searching;
 184         cl->requestid = se->requestid;
 185     }
 186     else
 187         cl->state = Client_Error;
 188
 189     odr_reset(global_parameters.odr_out);
 190 }
 191
 192 static void send_present(IOCHAN i)
 193 {
 194     struct connection *co = iochan_getdata(i);
 195     struct client *cl = co->client;
 196     Z_APDU *a = zget_APDU(global_parameters.odr_out, Z_APDU_presentRequest);
 197     int toget;
 198     int start = cl->records + 1;
 199
 200     toget = global_parameters.chunk;
 201     if (toget > cl->hits - cl->records)
 202         toget = cl->hits - cl->records;
 203
 204     yaz_log(YLOG_DEBUG, "Trying to present %d records\n", toget);
 205
 206     a->u.presentRequest->resultSetStartPoint = &start;
 207     a->u.presentRequest->numberOfRecordsRequested = &toget;
 208
 209     a->u.presentRequest->resultSetId = "Default";
 210
 211     a->u.presentRequest->preferredRecordSyntax =
 212             yaz_oidval_to_z3950oid(global_parameters.odr_out,
 213             CLASS_RECSYN, VAL_USMARC);
 214
 215     if (send_apdu(cl, a) >= 0)
 216     {
 217         iochan_setflags(i, EVENT_INPUT);
 218         cl->state = Client_Presenting;
 219     }
 220     else
 221         cl->state = Client_Error;
 222     odr_reset(global_parameters.odr_out);
 223 }
 224
 225 static void do_initResponse(IOCHAN i, Z_APDU *a)
 226 {
 227     struct connection *co = iochan_getdata(i);
 228     struct client *cl = co->client;
 229     Z_InitResponse *r = a->u.initResponse;
 230
 231     yaz_log(YLOG_DEBUG, "Received init response");
 232
 233     if (*r->result)
 234     {
 235         cl->state = Client_Idle;
 236     }
 237     else
 238         cl->state = Client_Failed; // FIXME need to do something to the connection
 239 }
 240
 241 static void do_searchResponse(IOCHAN i, Z_APDU *a)
 242 {
 243     struct connection *co = iochan_getdata(i);
 244     struct client *cl = co->client;
 245     struct session *se = cl->session;
 246     Z_SearchResponse *r = a->u.searchResponse;
 247
 248     yaz_log(YLOG_DEBUG, "Searchresponse (status=%d)", *r->searchStatus);
 249
 250     if (*r->searchStatus)
 251     {
 252         cl->hits = *r->resultCount;
 253         se->total_hits += cl->hits;
 254         if (r->presentStatus && !*r->presentStatus && r->records)
 255         {
 256             yaz_log(YLOG_DEBUG, "Records in search response");
 257             cl->records += *r->numberOfRecordsReturned;
 258             ingest_records(cl, r->records);
 259         }
 260         cl->state = Client_Idle;
 261     }
 262     else
 263     {          /*"FAILED"*/
 264         cl->hits = 0;
 265         cl->state = Client_Error;
 266         if (r->records) {
 267             Z_Records *recs = r->records;
 268             if (recs->which == Z_Records_NSD)
 269             {
 270                 yaz_log(YLOG_WARN, "Non-surrogate diagnostic");
 271                 cl->diagnostic = *recs->u.nonSurrogateDiagnostic->condition;
 272                 cl->state = Client_Error;
 273             }
 274         }
 275     }
 276 }
 277
 278 char *normalize_mergekey(char *buf)
 279 {
 280     char *p = buf, *pout = buf;
 281
 282     while (*p)
 283     {
 284         while (*p && !isalnum(*p))
 285             p++;
 286         while (isalnum(*p))
 287             *(pout++) = tolower(*(p++));
 288         if (*p)
 289             *(pout++) = ' ';
 290         while (*p && !isalnum(*p))
 291             p++;
 292     }
 293     if (buf != pout)
 294         *pout = '\0';
 295
 296     return buf;
 297 }
 298
 299
 300 #ifdef GAGA
 301 // FIXME needs to be generalized. Should flexibly generate X lists per search
 302 static void extract_subject(struct session *s, const char *rec)
 303 {
 304     const char *field, *subfield;
 305
 306     while ((field = find_field(rec, "650")))
 307     {
 308         rec = field;
 309         if ((subfield = find_subfield(field, 'a')))
 310         {
 311             char *e, *ef;
 312             char buf[1024];
 313             int len;
 314
 315             ef = index(subfield, '\n');
 316             if (!ef)
 317                 return;
 318             if ((e = index(subfield, '\t')) && e < ef)
 319                 ef = e;
 320             while (ef > subfield && !isalpha(*(ef - 1)) && *(ef - 1) != ')')
 321                 ef--;
 322             len = ef - subfield;
 323             assert(len < 1023);
 324             memcpy(buf, subfield, len);
 325             buf[len] = '\0';
 326 #ifdef FIXME
 327             if (*buf)
 328                 termlist_insert(s->termlist, buf);
 329 #endif
 330         }
 331     }
 332 }
 333 #endif
 334
 335 static void add_facet(struct session *s, const char *type, const char *value)
 336 {
 337     int i;
 338
 339     for (i = 0; i < s->num_termlists; i++)
 340         if (!strcmp(s->termlists[i].name, type))
 341             break;
 342     if (i == s->num_termlists)
 343     {
 344         if (i == SESSION_MAX_TERMLISTS)
 345         {
 346             yaz_log(YLOG_FATAL, "Too many termlists");
 347             exit(1);
 348         }
 349         s->termlists[i].name = nmem_strdup(s->nmem, type);
 350         s->termlists[i].termlist = termlist_create(s->nmem, s->expected_maxrecs, 15);
 351         s->num_termlists = i + 1;
 352     }
 353     termlist_insert(s->termlists[i].termlist, value);
 354 }
 355
 356 static xmlDoc *normalize_record(struct client *cl, Z_External *rec)
 357 {
 358     struct conf_retrievalprofile *rprofile = cl->database->rprofile;
 359     struct conf_retrievalmap *m;
 360     xmlNode *res;
 361     xmlDoc *rdoc;
 362
 363     // First normalize to XML
 364     if (rprofile->native_syntax == Nativesyn_iso2709)
 365     {
 366         char *buf;
 367         int len;
 368         if (rec->which != Z_External_octet)
 369         {
 370             yaz_log(YLOG_WARN, "Unexpected external branch, probably BER");
 371             return 0;
 372         }
 373         buf = (char*) rec->u.octet_aligned->buf;
 374         len = rec->u.octet_aligned->len;
 375         if (yaz_marc_read_iso2709(rprofile->yaz_marc, buf, len) < 0)
 376         {
 377             yaz_log(YLOG_WARN, "Failed to decode MARC");
 378             return 0;
 379         }
 380         if (yaz_marc_write_xml(rprofile->yaz_marc, &res,
 381                     "http://www.loc.gov/MARC21/slim", 0, 0) < 0)
 382         {
 383             yaz_log(YLOG_WARN, "Failed to encode as XML");
 384             return 0;
 385         }
 386         rdoc = xmlNewDoc("1.0");
 387         xmlDocSetRootElement(rdoc, res);
 388     }
 389     else
 390     {
 391         yaz_log(YLOG_FATAL, "Unknown native_syntax in normalize_record");
 392         exit(1);
 393     }
 394     for (m = rprofile->maplist; m; m = m->next)
 395     {
 396         xmlDoc *new;
 397         if (m->type != Map_xslt)
 398         {
 399             yaz_log(YLOG_WARN, "Unknown map type");
 400             return 0;
 401         }
 402         if (!(new = xsltApplyStylesheet(m->stylesheet, rdoc, 0)))
 403         {
 404             yaz_log(YLOG_WARN, "XSLT transformation failed");
 405             return 0;
 406         }
 407         xmlFreeDoc(rdoc);
 408         rdoc = new;
 409     }
 410     if (global_parameters.dump_records)
 411     {
 412         fprintf(stderr, "Record:\n----------------\n");
 413         xmlDocFormatDump(stderr, rdoc, 1);
 414     }
 415     return rdoc;
 416 }
 417
 418 static struct record *ingest_record(struct client *cl, Z_External *rec)
 419 {
 420     xmlDoc *xdoc = normalize_record(cl, rec);
 421     xmlNode *root, *n;
 422     struct record *res, *head;
 423     struct session *se = cl->session;
 424     xmlChar *mergekey, *mergekey_norm;
 425
 426     if (!xdoc)
 427         return 0;
 428
 429     root = xmlDocGetRootElement(xdoc);
 430     if (!(mergekey = xmlGetProp(root, "mergekey")))
 431     {
 432         yaz_log(YLOG_WARN, "No mergekey found in record");
 433         return 0;
 434     }
 435
 436     res = nmem_malloc(se->nmem, sizeof(struct record));
 437     res->next_cluster = 0;
 438     res->target_offset = -1;
 439     res->term_frequency_vec = 0;
 440     res->title = "Unknown";
 441     res->relevance = 0;
 442
 443     mergekey_norm = nmem_strdup(se->nmem, (char*) mergekey);
 444     xmlFree(mergekey);
 445     res->merge_key = normalize_mergekey(mergekey_norm);
 446
 447     head = reclist_insert(se->reclist, res);
 448     relevance_newrec(se->relevance, head);
 449
 450     for (n = root->children; n; n = n->next)
 451     {
 452         if (n->type != XML_ELEMENT_NODE)
 453             continue;
 454         if (!strcmp(n->name, "facet"))
 455         {
 456             xmlChar *type = xmlGetProp(n, "type");
 457             xmlChar *value = xmlNodeListGetString(xdoc, n->children, 0);
 458             if (type && value)
 459             {
 460                 add_facet(se, type, value);
 461                 relevance_countwords(se->relevance, head, value, 1);
 462             }
 463             xmlFree(type);
 464             xmlFree(value);
 465         }
 466         else if (!strcmp(n->name, "metadata"))
 467         {
 468             xmlChar *type = xmlGetProp(n, "type");
 469             if (type && !strcmp(type, "title"))
 470             {
 471                 xmlChar *value = xmlNodeListGetString(xdoc, n->children, 0);
 472                 if (value)
 473                 {
 474                     res->title = nmem_strdup(se->nmem, value);
 475                     relevance_countwords(se->relevance, head, value, 4);
 476                     xmlFree(value);
 477                 }
 478             }
 479             xmlFree(type);
 480         }
 481         else
 482             yaz_log(YLOG_WARN, "Unexpected element %s in internal record", n->name);
 483     }
 484
 485     xmlFreeDoc(xdoc);
 486
 487     relevance_donerecord(se->relevance, head);
 488     se->total_records++;
 489
 490     return res;
 491 }
 492
 493 static void ingest_records(struct client *cl, Z_Records *r)
 494 {
 495 #if USE_TIMING
 496     yaz_timing_t t = yaz_timing_create();
 497 #endif
 498     struct record *rec;
 499     struct session *s = cl->session;
 500     Z_NamePlusRecordList *rlist;
 501     int i;
 502
 503     if (r->which != Z_Records_DBOSD)
 504         return;
 505     rlist = r->u.databaseOrSurDiagnostics;
 506     for (i = 0; i < rlist->num_records; i++)
 507     {
 508         Z_NamePlusRecord *npr = rlist->records[i];
 509
 510         if (npr->which != Z_NamePlusRecord_databaseRecord)
 511         {
 512             yaz_log(YLOG_WARN, "Unexpected record type, probably diagnostic");
 513             continue;
 514         }
 515
 516         rec = ingest_record(cl, npr->u.databaseRecord);
 517         if (!rec)
 518             continue;
 519     }
 520     if (s->watchlist[SESSION_WATCH_RECORDS].fun && rlist->num_records)
 521         session_alert_watch(s, SESSION_WATCH_RECORDS);
 522
 523 #if USE_TIMING
 524     yaz_timing_stop(t);
 525     yaz_log(YLOG_LOG, "ingest_records %6.5f %3.2f %3.2f",
 526             yaz_timing_get_real(t), yaz_timing_get_user(t),
 527             yaz_timing_get_sys(t));
 528     yaz_timing_destroy(&t);
 529 #endif
 530 }
 531
 532 static void do_presentResponse(IOCHAN i, Z_APDU *a)
 533 {
 534     struct connection *co = iochan_getdata(i);
 535     struct client *cl = co->client;
 536     Z_PresentResponse *r = a->u.presentResponse;
 537
 538     if (r->records) {
 539         Z_Records *recs = r->records;
 540         if (recs->which == Z_Records_NSD)
 541         {
 542             yaz_log(YLOG_WARN, "Non-surrogate diagnostic");
 543             cl->diagnostic = *recs->u.nonSurrogateDiagnostic->condition;
 544             cl->state = Client_Error;
 545         }
 546     }
 547
 548     if (!*r->presentStatus && cl->state != Client_Error)
 549     {
 550         yaz_log(YLOG_DEBUG, "Good Present response");
 551         cl->records += *r->numberOfRecordsReturned;
 552         ingest_records(cl, r->records);
 553         cl->state = Client_Idle;
 554     }
 555     else if (*r->presentStatus)
 556     {
 557         yaz_log(YLOG_WARN, "Bad Present response");
 558         cl->state = Client_Error;
 559     }
 560 }
 561
 562 static void handler(IOCHAN i, int event)
 563 {
 564     struct connection *co = iochan_getdata(i);
 565     struct client *cl = co->client;
 566     struct session *se = 0;
 567
 568     if (cl)
 569         se = cl->session;
 570     else
 571     {
 572         yaz_log(YLOG_WARN, "Destroying orphan connection");
 573         connection_destroy(co);
 574         return;
 575     }
 576
 577     if (co->state == Conn_Connecting && event & EVENT_OUTPUT)
 578     {
 579         int errcode;
 580         socklen_t errlen = sizeof(errcode);
 581
 582         if (getsockopt(cs_fileno(co->link), SOL_SOCKET, SO_ERROR, &errcode,
 583             &errlen) < 0 || errcode != 0)
 584         {
 585             client_fatal(cl);
 586             return;
 587         }
 588         else
 589         {
 590             yaz_log(YLOG_DEBUG, "Connect OK");
 591             co->state = Conn_Open;
 592             if (cl)
 593                 cl->state = Client_Connected;
 594         }
 595     }
 596
 597     else if (event & EVENT_INPUT)
 598     {
 599         int len = cs_get(co->link, &co->ibuf, &co->ibufsize);
 600
 601         if (len < 0)
 602         {
 603             yaz_log(YLOG_WARN|YLOG_ERRNO, "Error reading from Z server");
 604             connection_destroy(co);
 605             return;
 606         }
 607         else if (len == 0)
 608         {
 609             yaz_log(YLOG_WARN, "EOF reading from Z server");
 610             connection_destroy(co);
 611             return;
 612         }
 613         else if (len > 1) // We discard input if we have no connection
 614         {
 615             co->state = Conn_Open;
 616
 617             if (cl && (cl->requestid == se->requestid || cl->state == Client_Initializing))
 618             {
 619                 Z_APDU *a;
 620
 621                 odr_reset(global_parameters.odr_in);
 622                 odr_setbuf(global_parameters.odr_in, co->ibuf, len, 0);
 623                 if (!z_APDU(global_parameters.odr_in, &a, 0, 0))
 624                 {
 625                     client_fatal(cl);
 626                     return;
 627                 }
 628                 switch (a->which)
 629                 {
 630                     case Z_APDU_initResponse:
 631                         do_initResponse(i, a);
 632                         break;
 633                     case Z_APDU_searchResponse:
 634                         do_searchResponse(i, a);
 635                         break;
 636                     case Z_APDU_presentResponse:
 637                         do_presentResponse(i, a);
 638                         break;
 639                     default:
 640                         yaz_log(YLOG_WARN, "Unexpected result from server");
 641                         client_fatal(cl);
 642                         return;
 643                 }
 644                 // We aren't expecting staggered output from target
 645                 // if (cs_more(t->link))
 646                 //    iochan_setevent(i, EVENT_INPUT);
 647             }
 648             else  // we throw away response and go to idle mode
 649             {
 650                 yaz_log(YLOG_DEBUG, "Ignoring result of expired operation");
 651                 cl->state = Client_Idle;
 652             }
 653         }
 654         /* if len==1 we do nothing but wait for more input */
 655     }
 656
 657     if (cl->state == Client_Connected) {
 658         send_init(i);
 659     }
 660
 661     if (cl->state == Client_Idle)
 662     {
 663         if (cl->requestid != se->requestid && *se->query) {
 664             send_search(i);
 665         }
 666         else if (cl->hits > 0 && cl->records < global_parameters.toget &&
 667             cl->records < cl->hits) {
 668             send_present(i);
 669         }
 670     }
 671 }
 672
 673 // Disassociate connection from client
 674 static void connection_release(struct connection *co)
 675 {
 676     struct client *cl = co->client;
 677
 678     yaz_log(YLOG_DEBUG, "Connection release %s", co->host->hostport);
 679     if (!cl)
 680         return;
 681     cl->connection = 0;
 682     co->client = 0;
 683 }
 684
 685 // Close connection and recycle structure
 686 static void connection_destroy(struct connection *co)
 687 {
 688     struct host *h = co->host;
 689     cs_close(co->link);
 690     iochan_destroy(co->iochan);
 691
 692     yaz_log(YLOG_DEBUG, "Connection destroy %s", co->host->hostport);
 693     if (h->connections == co)
 694         h->connections = co->next;
 695     else
 696     {
 697         struct connection *pco;
 698         for (pco = h->connections; pco && pco->next != co; pco = pco->next)
 699             ;
 700         if (pco)
 701             pco->next = co->next;
 702         else
 703             abort();
 704     }
 705     if (co->client)
 706     {
 707         if (co->client->state != Client_Idle)
 708             co->client->state = Client_Disconnected;
 709         co->client->connection = 0;
 710     }
 711     co->next = connection_freelist;
 712     connection_freelist = co;
 713 }
 714
 715 // Creates a new connection for client, associated with the host of
 716 // client's database
 717 static struct connection *connection_create(struct client *cl)
 718 {
 719     struct connection *new;
 720     COMSTACK link;
 721     int res;
 722     void *addr;
 723
 724     yaz_log(YLOG_DEBUG, "Connection create %s", cl->database->url);
 725     if (!(link = cs_create(tcpip_type, 0, PROTO_Z3950)))
 726     {
 727         yaz_log(YLOG_FATAL|YLOG_ERRNO, "Failed to create comstack");
 728         exit(1);
 729     }
 730
 731     if (!(addr = cs_straddr(link, cl->database->host->ipport)))
 732     {
 733         yaz_log(YLOG_WARN|YLOG_ERRNO, "Lookup of IP address failed?");
 734         return 0;
 735     }
 736
 737     res = cs_connect(link, addr);
 738     if (res < 0)
 739     {
 740         yaz_log(YLOG_WARN|YLOG_ERRNO, "cs_connect %s", cl->database->url);
 741         return 0;
 742     }
 743
 744     if ((new = connection_freelist))
 745         connection_freelist = new->next;
 746     else
 747     {
 748         new = xmalloc(sizeof (struct connection));
 749         new->ibuf = 0;
 750         new->ibufsize = 0;
 751     }
 752     new->state = Conn_Connecting;
 753     new->host = cl->database->host;
 754     new->next = new->host->connections;
 755     new->host->connections = new;
 756     new->client = cl;
 757     cl->connection = new;
 758     new->link = link;
 759
 760     new->iochan = iochan_create(cs_fileno(link), handler, 0);
 761     iochan_setdata(new->iochan, new);
 762     new->iochan->next = channel_list;
 763     channel_list = new->iochan;
 764     return new;
 765 }
 766
 767 // Close connection and set state to error
 768 static void client_fatal(struct client *cl)
 769 {
 770     yaz_log(YLOG_WARN, "Fatal error from %s", cl->database->url);
 771     connection_destroy(cl->connection);
 772     cl->state = Client_Error;
 773 }
 774
 775 // Ensure that client has a connection associated
 776 static int client_prep_connection(struct client *cl)
 777 {
 778     struct connection *co;
 779     struct session *se = cl->session;
 780     struct host *host = cl->database->host;
 781
 782     co = cl->connection;
 783
 784     yaz_log(YLOG_DEBUG, "Client prep %s", cl->database->url);
 785
 786     if (!co)
 787     {
 788         // See if someone else has an idle connection
 789         // We should look at timestamps here to select the longest-idle connection
 790         for (co = host->connections; co; co = co->next)
 791             if (co->state == Conn_Open && (!co->client || co->client->session != se))
 792                 break;
 793         if (co)
 794         {
 795             connection_release(co);
 796             cl->connection = co;
 797             co->client = cl;
 798         }
 799         else
 800             co = connection_create(cl);
 801     }
 802     if (co)
 803     {
 804         if (co->state == Conn_Connecting)
 805         {
 806             cl->state = Client_Connecting;
 807             iochan_setflag(co->iochan, EVENT_OUTPUT);
 808         }
 809         else if (co->state == Conn_Open)
 810         {
 811             if (cl->state == Client_Error || cl->state == Client_Disconnected)
 812                 cl->state = Client_Idle;
 813             iochan_setflag(co->iochan, EVENT_OUTPUT);
 814         }
 815         return 1;
 816     }
 817     else
 818         return 0;
 819 }
 820
 821 // This function will most likely vanish when a proper target profile mechanism is
 822 // introduced.
 823 void load_simpletargets(const char *fn)
 824 {
 825     FILE *f = fopen(fn, "r");
 826     char line[256];
 827
 828     if (!f)
 829     {
 830         yaz_log(YLOG_WARN|YLOG_ERRNO, "open %s", fn);
 831         exit(1);
 832     }
 833
 834     while (fgets(line, 255, f))
 835     {
 836         char *url, *db;
 837         struct host *host;
 838         struct database *database;
 839
 840         if (strncmp(line, "target ", 7))
 841             continue;
 842         url = line + 7;
 843         url[strlen(url) - 1] = '\0';
 844         yaz_log(YLOG_DEBUG, "Target: %s", url);
 845         if ((db = strchr(url, '/')))
 846             *(db++) = '\0';
 847         else
 848             db = "Default";
 849
 850         for (host = hosts; host; host = host->next)
 851             if (!strcmp(url, host->hostport))
 852                 break;
 853         if (!host)
 854         {
 855             struct addrinfo *addrinfo, hints;
 856             char *port;
 857             char ipport[128];
 858             unsigned char addrbuf[4];
 859             int res;
 860
 861             host = xmalloc(sizeof(struct host));
 862             host->hostport = xstrdup(url);
 863             host->connections = 0;
 864
 865             if ((port = strchr(url, ':')))
 866                 *(port++) = '\0';
 867             else
 868                 port = "210";
 869
 870             hints.ai_flags = 0;
 871             hints.ai_family = PF_INET;
 872             hints.ai_socktype = SOCK_STREAM;
 873             hints.ai_protocol = IPPROTO_TCP;
 874             hints.ai_addrlen = 0;
 875             hints.ai_addr = 0;
 876             hints.ai_canonname = 0;
 877             hints.ai_next = 0;
 878             // This is not robust code. It assumes that getaddrinfo returns AF_INET
 879             // address.
 880             if ((res = getaddrinfo(url, port, &hints, &addrinfo)))
 881             {
 882                 yaz_log(YLOG_WARN, "Failed to resolve %s: %s", url, gai_strerror(res));
 883                 xfree(host->hostport);
 884                 xfree(host);
 885                 continue;
 886             }
 887             assert(addrinfo->ai_family == PF_INET);
 888             memcpy(addrbuf, &((struct sockaddr_in*)addrinfo->ai_addr)->sin_addr.s_addr, 4);
 889             sprintf(ipport, "%hhd.%hhd.%hhd.%hhd:%s",
 890                     addrbuf[0], addrbuf[1], addrbuf[2], addrbuf[3], port);
 891             host->ipport = xstrdup(ipport);
 892             freeaddrinfo(addrinfo);
 893             host->next = hosts;
 894             hosts = host;
 895         }
 896         database = xmalloc(sizeof(struct database));
 897         database->host = host;
 898         database->url = xmalloc(strlen(url) + strlen(db) + 2);
 899         strcpy(database->url, url);
 900         strcat(database->url, "/");
 901         strcat(database->url, db);
 902
 903         database->databases = xmalloc(2 * sizeof(char *));
 904         database->databases[0] = xstrdup(db);
 905         database->databases[1] = 0;
 906         database->errors = 0;
 907         database->qprofile = 0;
 908         database->rprofile = database_retrieval_profile(database);
 909         database->next = databases;
 910         databases = database;
 911
 912     }
 913     fclose(f);
 914 }
 915
 916 static void pull_terms(NMEM nmem, struct ccl_rpn_node *n, char **termlist, int *num)
 917 {
 918     switch (n->kind)
 919     {
 920         case CCL_RPN_AND:
 921         case CCL_RPN_OR:
 922         case CCL_RPN_NOT:
 923         case CCL_RPN_PROX:
 924             pull_terms(nmem, n->u.p[0], termlist, num);
 925             pull_terms(nmem, n->u.p[1], termlist, num);
 926             break;
 927         case CCL_RPN_TERM:
 928             termlist[(*num)++] = nmem_strdup(nmem, n->u.t.term);
 929             break;
 930         default: // NOOP
 931             break;
 932     }
 933 }
 934
 935 // Extract terms from query into null-terminated termlist
 936 static int extract_terms(NMEM nmem, char *query, char **termlist)
 937 {
 938     int error, pos;
 939     struct ccl_rpn_node *n;
 940     int num = 0;
 941
 942     n = ccl_find_str(global_parameters.ccl_filter, query, &error, &pos);
 943     if (!n)
 944         return -1;
 945     pull_terms(nmem, n, termlist, &num);
 946     termlist[num] = 0;
 947     ccl_rpn_delete(n);
 948     return 0;
 949 }
 950
 951 static struct client *client_create(void)
 952 {
 953     struct client *r;
 954     if (client_freelist)
 955     {
 956         r = client_freelist;
 957         client_freelist = client_freelist->next;
 958     }
 959     else
 960         r = xmalloc(sizeof(struct client));
 961     r->database = 0;
 962     r->connection = 0;
 963     r->session = 0;
 964     r->hits = 0;
 965     r->records = 0;
 966     r->setno = 0;
 967     r->requestid = -1;
 968     r->diagnostic = 0;
 969     r->state = Client_Disconnected;
 970     r->next = 0;
 971     return r;
 972 }
 973
 974 void client_destroy(struct client *c)
 975 {
 976     struct session *se = c->session;
 977     if (c == se->clients)
 978         se->clients = c->next;
 979     else
 980     {
 981         struct client *cc;
 982         for (cc = se->clients; cc && cc->next != c; cc = cc->next)
 983             ;
 984         if (cc)
 985             cc->next = c->next;
 986     }
 987     if (c->connection)
 988         connection_release(c->connection);
 989     c->next = client_freelist;
 990     client_freelist = c;
 991 }
 992
 993 void session_set_watch(struct session *s, int what, session_watchfun fun, void *data)
 994 {
 995     s->watchlist[what].fun = fun;
 996     s->watchlist[what].data = data;
 997 }
 998
 999 void session_alert_watch(struct session *s, int what)
1000 {
1001     if (!s->watchlist[what].fun)
1002         return;
1003     (*s->watchlist[what].fun)(s->watchlist[what].data);
1004     s->watchlist[what].fun = 0;
1005     s->watchlist[what].data = 0;
1006 }
1007
1008 // This needs to be extended with selection criteria
1009 static struct conf_retrievalprofile *database_retrieval_profile(struct database *db)
1010 {
1011     if (!config)
1012     {
1013         yaz_log(YLOG_FATAL, "Must load configuration (-f)");
1014         exit(1);
1015     }
1016     if (!config->retrievalprofiles)
1017     {
1018         yaz_log(YLOG_FATAL, "No retrieval profiles defined");
1019     }
1020     return config->retrievalprofiles;
1021 }
1022
1023 // This should be extended with parameters to control selection criteria
1024 // Associates a set of clients with a session;
1025 int select_targets(struct session *se)
1026 {
1027     struct database *db;
1028     int c = 0;
1029
1030     while (se->clients)
1031         client_destroy(se->clients);
1032     for (db = databases; db; db = db->next)
1033     {
1034         struct client *cl = client_create();
1035         cl->database = db;
1036         cl->session = se;
1037         cl->next = se->clients;
1038         se->clients = cl;
1039         c++;
1040     }
1041     return c;
1042 }
1043
1044 int session_active_clients(struct session *s)
1045 {
1046     struct client *c;
1047     int res = 0;
1048
1049     for (c = s->clients; c; c = c->next)
1050         if (c->connection && (c->state == Client_Connecting ||
1051                     c->state == Client_Initializing ||
1052                     c->state == Client_Searching ||
1053                     c->state == Client_Presenting))
1054             res++;
1055
1056     return res;
1057 }
1058
1059 char *search(struct session *se, char *query)
1060 {
1061     int live_channels = 0;
1062     struct client *cl;
1063
1064     yaz_log(YLOG_DEBUG, "Search");
1065
1066     strcpy(se->query, query);
1067     se->requestid++;
1068     nmem_reset(se->nmem);
1069     for (cl = se->clients; cl; cl = cl->next)
1070     {
1071         cl->hits = -1;
1072         cl->records = 0;
1073         cl->diagnostic = 0;
1074
1075         if (client_prep_connection(cl))
1076             live_channels++;
1077     }
1078     if (live_channels)
1079     {
1080         char *p[512];
1081         int maxrecs = live_channels * global_parameters.toget;
1082         se->num_termlists = 0;
1083         se->reclist = reclist_create(se->nmem, maxrecs);
1084         extract_terms(se->nmem, query, p);
1085         se->relevance = relevance_create(se->nmem, (const char **) p, maxrecs);
1086         se->total_records = se->total_hits = 0;
1087         se->expected_maxrecs = maxrecs;
1088     }
1089     else
1090         return "NOTARGETS";
1091
1092     return 0;
1093 }
1094
1095 void destroy_session(struct session *s)
1096 {
1097     yaz_log(YLOG_LOG, "Destroying session");
1098     while (s->clients)
1099         client_destroy(s->clients);
1100     nmem_destroy(s->nmem);
1101     wrbuf_free(s->wrbuf, 1);
1102 }
1103
1104 struct session *new_session()
1105 {
1106     int i;
1107     struct session *session = xmalloc(sizeof(*session));
1108
1109     yaz_log(YLOG_DEBUG, "New pazpar2 session");
1110
1111     session->total_hits = 0;
1112     session->total_records = 0;
1113     session->num_termlists = 0;
1114     session->reclist = 0;
1115     session->requestid = -1;
1116     session->clients = 0;
1117     session->expected_maxrecs = 0;
1118     session->query[0] = '\0';
1119     session->nmem = nmem_create();
1120     session->wrbuf = wrbuf_alloc();
1121     for (i = 0; i <= SESSION_WATCH_MAX; i++)
1122     {
1123         session->watchlist[i].data = 0;
1124         session->watchlist[i].fun = 0;
1125     }
1126
1127     select_targets(session);
1128
1129     return session;
1130 }
1131
1132 struct hitsbytarget *hitsbytarget(struct session *se, int *count)
1133 {
1134     static struct hitsbytarget res[1000]; // FIXME MM
1135     struct client *cl;
1136
1137     *count = 0;
1138     for (cl = se->clients; cl; cl = cl->next)
1139     {
1140         strcpy(res[*count].id, cl->database->host->hostport);
1141         res[*count].hits = cl->hits;
1142         res[*count].records = cl->records;
1143         res[*count].diagnostic = cl->diagnostic;
1144         res[*count].state = client_states[cl->state];
1145         res[*count].connected  = cl->connection ? 1 : 0;
1146         (*count)++;
1147     }
1148
1149     return res;
1150 }
1151
1152 struct termlist_score **termlist(struct session *s, const char *name, int *num)
1153 {
1154     int i;
1155
1156     for (i = 0; i < s->num_termlists; i++)
1157         if (!strcmp(s->termlists[i].name, name))
1158             return termlist_highscore(s->termlists[i].termlist, num);
1159     return 0;
1160 }
1161
1162 #ifdef REPORT_NMEM
1163 // conditional compilation by SH: This lead to a warning with currently installed
1164 // YAZ header files on us1
1165 void report_nmem_stats(void)
1166 {
1167     size_t in_use, is_free;
1168
1169     nmem_get_memory_in_use(&in_use);
1170     nmem_get_memory_free(&is_free);
1171
1172     yaz_log(YLOG_LOG, "nmem stat: use=%ld free=%ld",
1173             (long) in_use, (long) is_free);
1174 }
1175 #endif
1176
1177 struct record **show(struct session *s, int start, int *num, int *total,
1178                      int *sumhits, NMEM nmem_show)
1179 {
1180     struct record **recs = nmem_malloc(nmem_show, *num
1181                                        * sizeof(struct record *));
1182     int i;
1183 #if USE_TIMING
1184     yaz_timing_t t = yaz_timing_create();
1185 #endif
1186     relevance_prepare_read(s->relevance, s->reclist);
1187
1188     *total = s->reclist->num_records;
1189     *sumhits = s->total_hits;
1190
1191     for (i = 0; i < start; i++)
1192         if (!reclist_read_record(s->reclist))
1193         {
1194             *num = 0;
1195             recs = 0;
1196             break;
1197         }
1198
1199     for (i = 0; i < *num; i++)
1200     {
1201         struct record *r = reclist_read_record(s->reclist);
1202         if (!r)
1203         {
1204             *num = i;
1205             break;
1206         }
1207         recs[i] = r;
1208     }
1209 #if USE_TIMING
1210     yaz_timing_stop(t);
1211     yaz_log(YLOG_LOG, "show %6.5f %3.2f %3.2f",
1212             yaz_timing_get_real(t), yaz_timing_get_user(t),
1213             yaz_timing_get_sys(t));
1214     yaz_timing_destroy(&t);
1215 #endif
1216     return recs;
1217 }
1218
1219 void statistics(struct session *se, struct statistics *stat)
1220 {
1221     struct client *cl;
1222     int count = 0;
1223
1224     bzero(stat, sizeof(*stat));
1225     for (cl = se->clients; cl; cl = cl->next)
1226     {
1227         if (!cl->connection)
1228             stat->num_no_connection++;
1229         switch (cl->state)
1230         {
1231             case Client_Connecting: stat->num_connecting++; break;
1232             case Client_Initializing: stat->num_initializing++; break;
1233             case Client_Searching: stat->num_searching++; break;
1234             case Client_Presenting: stat->num_presenting++; break;
1235             case Client_Idle: stat->num_idle++; break;
1236             case Client_Failed: stat->num_failed++; break;
1237             case Client_Error: stat->num_error++; break;
1238             default: break;
1239         }
1240         count++;
1241     }
1242     stat->num_hits = se->total_hits;
1243     stat->num_records = se->total_records;
1244
1245     stat->num_clients = count;
1246 }
1247
1248 static CCL_bibset load_cclfile(const char *fn)
1249 {
1250     CCL_bibset res = ccl_qual_mk();
1251     if (ccl_qual_fname(res, fn) < 0)
1252     {
1253         yaz_log(YLOG_FATAL|YLOG_ERRNO, "%s", fn);
1254         exit(1);
1255     }
1256     return res;
1257 }
1258
1259 int main(int argc, char **argv)
1260 {
1261     int ret;
1262     char *arg;
1263     int setport = 0;
1264
1265     if (signal(SIGPIPE, SIG_IGN) < 0)
1266         yaz_log(YLOG_WARN|YLOG_ERRNO, "signal");
1267
1268     yaz_log_init(YLOG_DEFAULT_LEVEL, "pazpar2", 0);
1269
1270     while ((ret = options("f:x:c:h:p:C:s:d", argv, argc, &arg)) != -2)
1271     {
1272         switch (ret) {
1273             case 'f':
1274                 if (!read_config(arg))
1275                     exit(1);
1276                 break;
1277             case 'c':
1278                 command_init(atoi(arg));
1279                 setport++;
1280                 break;
1281             case 'h':
1282                 http_init(arg);
1283                 setport++;
1284                 break;
1285             case 'C':
1286                 global_parameters.ccl_filter = load_cclfile(arg);
1287                 break;
1288             case 'p':
1289                 http_set_proxyaddr(arg);
1290                 break;
1291             case 's':
1292                 load_simpletargets(arg);
1293                 break;
1294             case 'd':
1295                 global_parameters.dump_records = 1;
1296                 break;
1297             default:
1298                 fprintf(stderr, "Usage: pazpar2\n"
1299                         "    -f configfile\n"
1300                         "    -h [host:]port          (REST protocol listener)\n"
1301                         "    -c cmdport              (telnet-style)\n"
1302                         "    -C cclconfig\n"
1303                         "    -s simpletargetfile\n"
1304                         "    -p hostname[:portno]    (HTTP proxy)\n");
1305                 exit(1);
1306         }
1307     }
1308
1309     if (!setport)
1310     {
1311         fprintf(stderr, "Set command port with -h or -c\n");
1312         exit(1);
1313     }
1314
1315     global_parameters.ccl_filter = load_cclfile("../etc/default.bib");
1316     global_parameters.yaz_marc = yaz_marc_create();
1317     yaz_marc_subfield_str(global_parameters.yaz_marc, "\t");
1318     global_parameters.odr_in = odr_createmem(ODR_DECODE);
1319     global_parameters.odr_out = odr_createmem(ODR_ENCODE);
1320
1321     event_loop(&channel_list);
1322
1323     return 0;
1324 }
1325
1326 /*
1327  * Local variables:
1328  * c-basic-offset: 4
1329  * indent-tabs-mode: nil
1330  * End:
1331  * vim: shiftwidth=4 tabstop=8 expandtab
1332  */