src/pazpar2.c

   1 /* $Id: pazpar2.c,v 1.11 2007-01-04 07:27:29 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/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             add_facet(se, type, value);
 459             relevance_countwords(se->relevance, head, value, 1);
 460             xmlFree(type);
 461             xmlFree(value);
 462         }
 463         else if (!strcmp(n->name, "metadata"))
 464         {
 465             xmlChar *type = xmlGetProp(n, "type"), *value;
 466             if (!strcmp(type, "title"))
 467                 res->title = nmem_strdup(se->nmem,
 468                         value = xmlNodeListGetString(xdoc, n->children, 0));
 469
 470             relevance_countwords(se->relevance, head, value, 4);
 471             xmlFree(type);
 472             xmlFree(value);
 473         }
 474         else
 475             yaz_log(YLOG_WARN, "Unexpected element %s in internal record", n->name);
 476     }
 477
 478     xmlFreeDoc(xdoc);
 479
 480     relevance_donerecord(se->relevance, head);
 481     se->total_records++;
 482
 483     return res;
 484 }
 485
 486 static void ingest_records(struct client *cl, Z_Records *r)
 487 {
 488 #if USE_TIMING
 489     yaz_timing_t t = yaz_timing_create();
 490 #endif
 491     struct record *rec;
 492     struct session *s = cl->session;
 493     Z_NamePlusRecordList *rlist;
 494     int i;
 495
 496     if (r->which != Z_Records_DBOSD)
 497         return;
 498     rlist = r->u.databaseOrSurDiagnostics;
 499     for (i = 0; i < rlist->num_records; i++)
 500     {
 501         Z_NamePlusRecord *npr = rlist->records[i];
 502
 503         if (npr->which != Z_NamePlusRecord_databaseRecord)
 504         {
 505             yaz_log(YLOG_WARN, "Unexpected record type, probably diagnostic");
 506             continue;
 507         }
 508
 509         rec = ingest_record(cl, npr->u.databaseRecord);
 510         if (!rec)
 511             continue;
 512     }
 513     if (s->watchlist[SESSION_WATCH_RECORDS].fun && rlist->num_records)
 514         session_alert_watch(s, SESSION_WATCH_RECORDS);
 515
 516 #if USE_TIMING
 517     yaz_timing_stop(t);
 518     yaz_log(YLOG_LOG, "ingest_records %6.5f %3.2f %3.2f",
 519             yaz_timing_get_real(t), yaz_timing_get_user(t),
 520             yaz_timing_get_sys(t));
 521     yaz_timing_destroy(&t);
 522 #endif
 523 }
 524
 525 static void do_presentResponse(IOCHAN i, Z_APDU *a)
 526 {
 527     struct connection *co = iochan_getdata(i);
 528     struct client *cl = co->client;
 529     Z_PresentResponse *r = a->u.presentResponse;
 530
 531     if (r->records) {
 532         Z_Records *recs = r->records;
 533         if (recs->which == Z_Records_NSD)
 534         {
 535             yaz_log(YLOG_WARN, "Non-surrogate diagnostic");
 536             cl->diagnostic = *recs->u.nonSurrogateDiagnostic->condition;
 537             cl->state = Client_Error;
 538         }
 539     }
 540
 541     if (!*r->presentStatus && cl->state != Client_Error)
 542     {
 543         yaz_log(YLOG_DEBUG, "Good Present response");
 544         cl->records += *r->numberOfRecordsReturned;
 545         ingest_records(cl, r->records);
 546         cl->state = Client_Idle;
 547     }
 548     else if (*r->presentStatus)
 549     {
 550         yaz_log(YLOG_WARN, "Bad Present response");
 551         cl->state = Client_Error;
 552     }
 553 }
 554
 555 static void handler(IOCHAN i, int event)
 556 {
 557     struct connection *co = iochan_getdata(i);
 558     struct client *cl = co->client;
 559     struct session *se = 0;
 560
 561     if (cl)
 562         se = cl->session;
 563     else
 564     {
 565         yaz_log(YLOG_WARN, "Destroying orphan connection");
 566         connection_destroy(co);
 567         return;
 568     }
 569
 570     if (co->state == Conn_Connecting && event & EVENT_OUTPUT)
 571     {
 572         int errcode;
 573         socklen_t errlen = sizeof(errcode);
 574
 575         if (getsockopt(cs_fileno(co->link), SOL_SOCKET, SO_ERROR, &errcode,
 576             &errlen) < 0 || errcode != 0)
 577         {
 578             client_fatal(cl);
 579             return;
 580         }
 581         else
 582         {
 583             yaz_log(YLOG_DEBUG, "Connect OK");
 584             co->state = Conn_Open;
 585             if (cl)
 586                 cl->state = Client_Connected;
 587         }
 588     }
 589
 590     else if (event & EVENT_INPUT)
 591     {
 592         int len = cs_get(co->link, &co->ibuf, &co->ibufsize);
 593
 594         if (len < 0)
 595         {
 596             yaz_log(YLOG_WARN|YLOG_ERRNO, "Error reading from Z server");
 597             connection_destroy(co);
 598             return;
 599         }
 600         else if (len == 0)
 601         {
 602             yaz_log(YLOG_WARN, "EOF reading from Z server");
 603             connection_destroy(co);
 604             return;
 605         }
 606         else if (len > 1) // We discard input if we have no connection
 607         {
 608             co->state = Conn_Open;
 609
 610             if (cl && (cl->requestid == se->requestid || cl->state == Client_Initializing))
 611             {
 612                 Z_APDU *a;
 613
 614                 odr_reset(global_parameters.odr_in);
 615                 odr_setbuf(global_parameters.odr_in, co->ibuf, len, 0);
 616                 if (!z_APDU(global_parameters.odr_in, &a, 0, 0))
 617                 {
 618                     client_fatal(cl);
 619                     return;
 620                 }
 621                 switch (a->which)
 622                 {
 623                     case Z_APDU_initResponse:
 624                         do_initResponse(i, a);
 625                         break;
 626                     case Z_APDU_searchResponse:
 627                         do_searchResponse(i, a);
 628                         break;
 629                     case Z_APDU_presentResponse:
 630                         do_presentResponse(i, a);
 631                         break;
 632                     default:
 633                         yaz_log(YLOG_WARN, "Unexpected result from server");
 634                         client_fatal(cl);
 635                         return;
 636                 }
 637                 // We aren't expecting staggered output from target
 638                 // if (cs_more(t->link))
 639                 //    iochan_setevent(i, EVENT_INPUT);
 640             }
 641             else  // we throw away response and go to idle mode
 642             {
 643                 yaz_log(YLOG_DEBUG, "Ignoring result of expired operation");
 644                 cl->state = Client_Idle;
 645             }
 646         }
 647         /* if len==1 we do nothing but wait for more input */
 648     }
 649
 650     if (cl->state == Client_Connected) {
 651         send_init(i);
 652     }
 653
 654     if (cl->state == Client_Idle)
 655     {
 656         if (cl->requestid != se->requestid && *se->query) {
 657             send_search(i);
 658         }
 659         else if (cl->hits > 0 && cl->records < global_parameters.toget &&
 660             cl->records < cl->hits) {
 661             send_present(i);
 662         }
 663     }
 664 }
 665
 666 // Disassociate connection from client
 667 static void connection_release(struct connection *co)
 668 {
 669     struct client *cl = co->client;
 670
 671     yaz_log(YLOG_DEBUG, "Connection release %s", co->host->hostport);
 672     if (!cl)
 673         return;
 674     cl->connection = 0;
 675     co->client = 0;
 676 }
 677
 678 // Close connection and recycle structure
 679 static void connection_destroy(struct connection *co)
 680 {
 681     struct host *h = co->host;
 682     cs_close(co->link);
 683     iochan_destroy(co->iochan);
 684
 685     yaz_log(YLOG_DEBUG, "Connection destroy %s", co->host->hostport);
 686     if (h->connections == co)
 687         h->connections = co->next;
 688     else
 689     {
 690         struct connection *pco;
 691         for (pco = h->connections; pco && pco->next != co; pco = pco->next)
 692             ;
 693         if (pco)
 694             pco->next = co->next;
 695         else
 696             abort();
 697     }
 698     if (co->client)
 699     {
 700         if (co->client->state != Client_Idle)
 701             co->client->state = Client_Disconnected;
 702         co->client->connection = 0;
 703     }
 704     co->next = connection_freelist;
 705     connection_freelist = co;
 706 }
 707
 708 // Creates a new connection for client, associated with the host of
 709 // client's database
 710 static struct connection *connection_create(struct client *cl)
 711 {
 712     struct connection *new;
 713     COMSTACK link;
 714     int res;
 715     void *addr;
 716
 717     yaz_log(YLOG_DEBUG, "Connection create %s", cl->database->url);
 718     if (!(link = cs_create(tcpip_type, 0, PROTO_Z3950)))
 719     {
 720         yaz_log(YLOG_FATAL|YLOG_ERRNO, "Failed to create comstack");
 721         exit(1);
 722     }
 723
 724     if (!(addr = cs_straddr(link, cl->database->host->ipport)))
 725     {
 726         yaz_log(YLOG_WARN|YLOG_ERRNO, "Lookup of IP address failed?");
 727         return 0;
 728     }
 729
 730     res = cs_connect(link, addr);
 731     if (res < 0)
 732     {
 733         yaz_log(YLOG_WARN|YLOG_ERRNO, "cs_connect %s", cl->database->url);
 734         return 0;
 735     }
 736
 737     if ((new = connection_freelist))
 738         connection_freelist = new->next;
 739     else
 740     {
 741         new = xmalloc(sizeof (struct connection));
 742         new->ibuf = 0;
 743         new->ibufsize = 0;
 744     }
 745     new->state = Conn_Connecting;
 746     new->host = cl->database->host;
 747     new->next = new->host->connections;
 748     new->host->connections = new;
 749     new->client = cl;
 750     cl->connection = new;
 751     new->link = link;
 752
 753     new->iochan = iochan_create(cs_fileno(link), handler, 0);
 754     iochan_setdata(new->iochan, new);
 755     new->iochan->next = channel_list;
 756     channel_list = new->iochan;
 757     return new;
 758 }
 759
 760 // Close connection and set state to error
 761 static void client_fatal(struct client *cl)
 762 {
 763     yaz_log(YLOG_WARN, "Fatal error from %s", cl->database->url);
 764     connection_destroy(cl->connection);
 765     cl->state = Client_Error;
 766 }
 767
 768 // Ensure that client has a connection associated
 769 static int client_prep_connection(struct client *cl)
 770 {
 771     struct connection *co;
 772     struct session *se = cl->session;
 773     struct host *host = cl->database->host;
 774
 775     co = cl->connection;
 776
 777     yaz_log(YLOG_DEBUG, "Client prep %s", cl->database->url);
 778
 779     if (!co)
 780     {
 781         // See if someone else has an idle connection
 782         // We should look at timestamps here to select the longest-idle connection
 783         for (co = host->connections; co; co = co->next)
 784             if (co->state == Conn_Open && (!co->client || co->client->session != se))
 785                 break;
 786         if (co)
 787         {
 788             connection_release(co);
 789             cl->connection = co;
 790             co->client = cl;
 791         }
 792         else
 793             co = connection_create(cl);
 794     }
 795     if (co)
 796     {
 797         if (co->state == Conn_Connecting)
 798         {
 799             cl->state = Client_Connecting;
 800             iochan_setflag(co->iochan, EVENT_OUTPUT);
 801         }
 802         else if (co->state == Conn_Open)
 803         {
 804             if (cl->state == Client_Error || cl->state == Client_Disconnected)
 805                 cl->state = Client_Idle;
 806             iochan_setflag(co->iochan, EVENT_OUTPUT);
 807         }
 808         return 1;
 809     }
 810     else
 811         return 0;
 812 }
 813
 814 // This function will most likely vanish when a proper target profile mechanism is
 815 // introduced.
 816 void load_simpletargets(const char *fn)
 817 {
 818     FILE *f = fopen(fn, "r");
 819     char line[256];
 820
 821     if (!f)
 822     {
 823         yaz_log(YLOG_WARN|YLOG_ERRNO, "open %s", fn);
 824         exit(1);
 825     }
 826
 827     while (fgets(line, 255, f))
 828     {
 829         char *url, *db;
 830         struct host *host;
 831         struct database *database;
 832
 833         if (strncmp(line, "target ", 7))
 834             continue;
 835         url = line + 7;
 836         url[strlen(url) - 1] = '\0';
 837         yaz_log(YLOG_DEBUG, "Target: %s", url);
 838         if ((db = strchr(url, '/')))
 839             *(db++) = '\0';
 840         else
 841             db = "Default";
 842
 843         for (host = hosts; host; host = host->next)
 844             if (!strcmp(url, host->hostport))
 845                 break;
 846         if (!host)
 847         {
 848             struct addrinfo *addrinfo, hints;
 849             char *port;
 850             char ipport[128];
 851             unsigned char addrbuf[4];
 852             int res;
 853
 854             host = xmalloc(sizeof(struct host));
 855             host->hostport = xstrdup(url);
 856             host->connections = 0;
 857
 858             if ((port = strchr(url, ':')))
 859                 *(port++) = '\0';
 860             else
 861                 port = "210";
 862
 863             hints.ai_flags = 0;
 864             hints.ai_family = PF_INET;
 865             hints.ai_socktype = SOCK_STREAM;
 866             hints.ai_protocol = IPPROTO_TCP;
 867             hints.ai_addrlen = 0;
 868             hints.ai_addr = 0;
 869             hints.ai_canonname = 0;
 870             hints.ai_next = 0;
 871             // This is not robust code. It assumes that getaddrinfo returns AF_INET
 872             // address.
 873             if ((res = getaddrinfo(url, port, &hints, &addrinfo)))
 874             {
 875                 yaz_log(YLOG_WARN, "Failed to resolve %s: %s", url, gai_strerror(res));
 876                 xfree(host->hostport);
 877                 xfree(host);
 878                 continue;
 879             }
 880             assert(addrinfo->ai_family == PF_INET);
 881             memcpy(addrbuf, &((struct sockaddr_in*)addrinfo->ai_addr)->sin_addr.s_addr, 4);
 882             sprintf(ipport, "%hhd.%hhd.%hhd.%hhd:%s",
 883                     addrbuf[0], addrbuf[1], addrbuf[2], addrbuf[3], port);
 884             host->ipport = xstrdup(ipport);
 885             freeaddrinfo(addrinfo);
 886             host->next = hosts;
 887             hosts = host;
 888         }
 889         database = xmalloc(sizeof(struct database));
 890         database->host = host;
 891         database->url = xmalloc(strlen(url) + strlen(db) + 2);
 892         strcpy(database->url, url);
 893         strcat(database->url, "/");
 894         strcat(database->url, db);
 895
 896         database->databases = xmalloc(2 * sizeof(char *));
 897         database->databases[0] = xstrdup(db);
 898         database->databases[1] = 0;
 899         database->errors = 0;
 900         database->qprofile = 0;
 901         database->rprofile = database_retrieval_profile(database);
 902         database->next = databases;
 903         databases = database;
 904
 905     }
 906     fclose(f);
 907 }
 908
 909 static void pull_terms(NMEM nmem, struct ccl_rpn_node *n, char **termlist, int *num)
 910 {
 911     switch (n->kind)
 912     {
 913         case CCL_RPN_AND:
 914         case CCL_RPN_OR:
 915         case CCL_RPN_NOT:
 916         case CCL_RPN_PROX:
 917             pull_terms(nmem, n->u.p[0], termlist, num);
 918             pull_terms(nmem, n->u.p[1], termlist, num);
 919             break;
 920         case CCL_RPN_TERM:
 921             termlist[(*num)++] = nmem_strdup(nmem, n->u.t.term);
 922             break;
 923         default: // NOOP
 924             break;
 925     }
 926 }
 927
 928 // Extract terms from query into null-terminated termlist
 929 static int extract_terms(NMEM nmem, char *query, char **termlist)
 930 {
 931     int error, pos;
 932     struct ccl_rpn_node *n;
 933     int num = 0;
 934
 935     n = ccl_find_str(global_parameters.ccl_filter, query, &error, &pos);
 936     if (!n)
 937         return -1;
 938     pull_terms(nmem, n, termlist, &num);
 939     termlist[num] = 0;
 940     ccl_rpn_delete(n);
 941     return 0;
 942 }
 943
 944 static struct client *client_create(void)
 945 {
 946     struct client *r;
 947     if (client_freelist)
 948     {
 949         r = client_freelist;
 950         client_freelist = client_freelist->next;
 951     }
 952     else
 953         r = xmalloc(sizeof(struct client));
 954     r->database = 0;
 955     r->connection = 0;
 956     r->session = 0;
 957     r->hits = 0;
 958     r->records = 0;
 959     r->setno = 0;
 960     r->requestid = -1;
 961     r->diagnostic = 0;
 962     r->state = Client_Disconnected;
 963     r->next = 0;
 964     return r;
 965 }
 966
 967 void client_destroy(struct client *c)
 968 {
 969     struct session *se = c->session;
 970     if (c == se->clients)
 971         se->clients = c->next;
 972     else
 973     {
 974         struct client *cc;
 975         for (cc = se->clients; cc && cc->next != c; cc = cc->next)
 976             ;
 977         if (cc)
 978             cc->next = c->next;
 979     }
 980     if (c->connection)
 981         connection_release(c->connection);
 982     c->next = client_freelist;
 983     client_freelist = c;
 984 }
 985
 986 void session_set_watch(struct session *s, int what, session_watchfun fun, void *data)
 987 {
 988     s->watchlist[what].fun = fun;
 989     s->watchlist[what].data = data;
 990 }
 991
 992 void session_alert_watch(struct session *s, int what)
 993 {
 994     if (!s->watchlist[what].fun)
 995         return;
 996     (*s->watchlist[what].fun)(s->watchlist[what].data);
 997     s->watchlist[what].fun = 0;
 998     s->watchlist[what].data = 0;
 999 }
1000
1001 // This needs to be extended with selection criteria
1002 static struct conf_retrievalprofile *database_retrieval_profile(struct database *db)
1003 {
1004     if (!config)
1005     {
1006         yaz_log(YLOG_FATAL, "Must load configuration (-f)");
1007         exit(1);
1008     }
1009     if (!config->retrievalprofiles)
1010     {
1011         yaz_log(YLOG_FATAL, "No retrieval profiles defined");
1012     }
1013     return config->retrievalprofiles;
1014 }
1015
1016 // This should be extended with parameters to control selection criteria
1017 // Associates a set of clients with a session;
1018 int select_targets(struct session *se)
1019 {
1020     struct database *db;
1021     int c = 0;
1022
1023     while (se->clients)
1024         client_destroy(se->clients);
1025     for (db = databases; db; db = db->next)
1026     {
1027         struct client *cl = client_create();
1028         cl->database = db;
1029         cl->session = se;
1030         cl->next = se->clients;
1031         se->clients = cl;
1032         c++;
1033     }
1034     return c;
1035 }
1036
1037 int session_active_clients(struct session *s)
1038 {
1039     struct client *c;
1040     int res = 0;
1041
1042     for (c = s->clients; c; c = c->next)
1043         if (c->connection && (c->state == Client_Connecting ||
1044                     c->state == Client_Initializing ||
1045                     c->state == Client_Searching ||
1046                     c->state == Client_Presenting))
1047             res++;
1048
1049     return res;
1050 }
1051
1052 char *search(struct session *se, char *query)
1053 {
1054     int live_channels = 0;
1055     struct client *cl;
1056
1057     yaz_log(YLOG_DEBUG, "Search");
1058
1059     strcpy(se->query, query);
1060     se->requestid++;
1061     nmem_reset(se->nmem);
1062     for (cl = se->clients; cl; cl = cl->next)
1063     {
1064         cl->hits = -1;
1065         cl->records = 0;
1066         cl->diagnostic = 0;
1067
1068         if (client_prep_connection(cl))
1069             live_channels++;
1070     }
1071     if (live_channels)
1072     {
1073         char *p[512];
1074         int maxrecs = live_channels * global_parameters.toget;
1075         se->num_termlists = 0;
1076         se->reclist = reclist_create(se->nmem, maxrecs);
1077         extract_terms(se->nmem, query, p);
1078         se->relevance = relevance_create(se->nmem, (const char **) p, maxrecs);
1079         se->total_records = se->total_hits = 0;
1080         se->expected_maxrecs = maxrecs;
1081     }
1082     else
1083         return "NOTARGETS";
1084
1085     return 0;
1086 }
1087
1088 void destroy_session(struct session *s)
1089 {
1090     yaz_log(YLOG_LOG, "Destroying session");
1091     while (s->clients)
1092         client_destroy(s->clients);
1093     nmem_destroy(s->nmem);
1094     wrbuf_free(s->wrbuf, 1);
1095 }
1096
1097 struct session *new_session()
1098 {
1099     int i;
1100     struct session *session = xmalloc(sizeof(*session));
1101
1102     yaz_log(YLOG_DEBUG, "New pazpar2 session");
1103
1104     session->total_hits = 0;
1105     session->total_records = 0;
1106     session->num_termlists = 0;
1107     session->reclist = 0;
1108     session->requestid = -1;
1109     session->clients = 0;
1110     session->expected_maxrecs = 0;
1111     session->query[0] = '\0';
1112     session->nmem = nmem_create();
1113     session->wrbuf = wrbuf_alloc();
1114     for (i = 0; i <= SESSION_WATCH_MAX; i++)
1115     {
1116         session->watchlist[i].data = 0;
1117         session->watchlist[i].fun = 0;
1118     }
1119
1120     select_targets(session);
1121
1122     return session;
1123 }
1124
1125 struct hitsbytarget *hitsbytarget(struct session *se, int *count)
1126 {
1127     static struct hitsbytarget res[1000]; // FIXME MM
1128     struct client *cl;
1129
1130     *count = 0;
1131     for (cl = se->clients; cl; cl = cl->next)
1132     {
1133         strcpy(res[*count].id, cl->database->host->hostport);
1134         res[*count].hits = cl->hits;
1135         res[*count].records = cl->records;
1136         res[*count].diagnostic = cl->diagnostic;
1137         res[*count].state = client_states[cl->state];
1138         res[*count].connected  = cl->connection ? 1 : 0;
1139         (*count)++;
1140     }
1141
1142     return res;
1143 }
1144
1145 struct termlist_score **termlist(struct session *s, const char *name, int *num)
1146 {
1147     int i;
1148
1149     for (i = 0; i < s->num_termlists; i++)
1150         if (!strcmp(s->termlists[i].name, name))
1151             return termlist_highscore(s->termlists[i].termlist, num);
1152     return 0;
1153 }
1154
1155 #ifdef REPORT_NMEM
1156 // conditional compilation by SH: This lead to a warning with currently installed
1157 // YAZ header files on us1
1158 void report_nmem_stats(void)
1159 {
1160     size_t in_use, is_free;
1161
1162     nmem_get_memory_in_use(&in_use);
1163     nmem_get_memory_free(&is_free);
1164
1165     yaz_log(YLOG_LOG, "nmem stat: use=%ld free=%ld",
1166             (long) in_use, (long) is_free);
1167 }
1168 #endif
1169
1170 struct record **show(struct session *s, int start, int *num, int *total,
1171                      int *sumhits, NMEM nmem_show)
1172 {
1173     struct record **recs = nmem_malloc(nmem_show, *num
1174                                        * sizeof(struct record *));
1175     int i;
1176 #if USE_TIMING
1177     yaz_timing_t t = yaz_timing_create();
1178 #endif
1179     relevance_prepare_read(s->relevance, s->reclist);
1180
1181     *total = s->reclist->num_records;
1182     *sumhits = s->total_hits;
1183
1184     for (i = 0; i < start; i++)
1185         if (!reclist_read_record(s->reclist))
1186         {
1187             *num = 0;
1188             recs = 0;
1189             break;
1190         }
1191
1192     for (i = 0; i < *num; i++)
1193     {
1194         struct record *r = reclist_read_record(s->reclist);
1195         if (!r)
1196         {
1197             *num = i;
1198             break;
1199         }
1200         recs[i] = r;
1201     }
1202 #if USE_TIMING
1203     yaz_timing_stop(t);
1204     yaz_log(YLOG_LOG, "show %6.5f %3.2f %3.2f",
1205             yaz_timing_get_real(t), yaz_timing_get_user(t),
1206             yaz_timing_get_sys(t));
1207     yaz_timing_destroy(&t);
1208 #endif
1209     return recs;
1210 }
1211
1212 void statistics(struct session *se, struct statistics *stat)
1213 {
1214     struct client *cl;
1215     int count = 0;
1216
1217     bzero(stat, sizeof(*stat));
1218     for (cl = se->clients; cl; cl = cl->next)
1219     {
1220         if (!cl->connection)
1221             stat->num_no_connection++;
1222         switch (cl->state)
1223         {
1224             case Client_Connecting: stat->num_connecting++; break;
1225             case Client_Initializing: stat->num_initializing++; break;
1226             case Client_Searching: stat->num_searching++; break;
1227             case Client_Presenting: stat->num_presenting++; break;
1228             case Client_Idle: stat->num_idle++; break;
1229             case Client_Failed: stat->num_failed++; break;
1230             case Client_Error: stat->num_error++; break;
1231             default: break;
1232         }
1233         count++;
1234     }
1235     stat->num_hits = se->total_hits;
1236     stat->num_records = se->total_records;
1237
1238     stat->num_clients = count;
1239 }
1240
1241 static CCL_bibset load_cclfile(const char *fn)
1242 {
1243     CCL_bibset res = ccl_qual_mk();
1244     if (ccl_qual_fname(res, fn) < 0)
1245     {
1246         yaz_log(YLOG_FATAL|YLOG_ERRNO, "%s", fn);
1247         exit(1);
1248     }
1249     return res;
1250 }
1251
1252 int main(int argc, char **argv)
1253 {
1254     int ret;
1255     char *arg;
1256     int setport = 0;
1257
1258     if (signal(SIGPIPE, SIG_IGN) < 0)
1259         yaz_log(YLOG_WARN|YLOG_ERRNO, "signal");
1260
1261     yaz_log_init(YLOG_DEFAULT_LEVEL, "pazpar2", 0);
1262
1263     while ((ret = options("f:x:c:h:p:C:s:d", argv, argc, &arg)) != -2)
1264     {
1265         switch (ret) {
1266             case 'f':
1267                 if (!read_config(arg))
1268                     exit(1);
1269                 break;
1270             case 'c':
1271                 command_init(atoi(arg));
1272                 setport++;
1273                 break;
1274             case 'h':
1275                 http_init(arg);
1276                 setport++;
1277                 break;
1278             case 'C':
1279                 global_parameters.ccl_filter = load_cclfile(arg);
1280                 break;
1281             case 'p':
1282                 http_set_proxyaddr(arg);
1283                 break;
1284             case 's':
1285                 load_simpletargets(arg);
1286                 break;
1287             case 'd':
1288                 global_parameters.dump_records = 1;
1289                 break;
1290             default:
1291                 fprintf(stderr, "Usage: pazpar2\n"
1292                         "    -f configfile\n"
1293                         "    -h [host:]port          (REST protocol listener)\n"
1294                         "    -c cmdport              (telnet-style)\n"
1295                         "    -C cclconfig\n"
1296                         "    -s simpletargetfile\n"
1297                         "    -p hostname[:portno]    (HTTP proxy)\n");
1298                 exit(1);
1299         }
1300     }
1301
1302     if (!setport)
1303     {
1304         fprintf(stderr, "Set command port with -h or -c\n");
1305         exit(1);
1306     }
1307
1308     global_parameters.ccl_filter = load_cclfile("../etc/default.bib");
1309     global_parameters.yaz_marc = yaz_marc_create();
1310     yaz_marc_subfield_str(global_parameters.yaz_marc, "\t");
1311     global_parameters.odr_in = odr_createmem(ODR_DECODE);
1312     global_parameters.odr_out = odr_createmem(ODR_ENCODE);
1313
1314     event_loop(&channel_list);
1315
1316     return 0;
1317 }
1318
1319 /*
1320  * Local variables:
1321  * c-basic-offset: 4
1322  * indent-tabs-mode: nil
1323  * End:
1324  * vim: shiftwidth=4 tabstop=8 expandtab
1325  */