src/pazpar2.c

   1 /* $Id: pazpar2.c,v 1.16 2007-01-06 03:02:47 quinn Exp $ */;
   2
   3 #include <stdlib.h>
   4 #include <stdio.h>
   5 #include <string.h>
   6 #include <sys/time.h>
   7 #include <unistd.h>
   8 #include <sys/socket.h>
   9 #include <netdb.h>
  10 #include <signal.h>
  11 #include <ctype.h>
  12 #include <assert.h>
  13
  14 #include <yaz/marcdisp.h>
  15 #include <yaz/comstack.h>
  16 #include <yaz/tcpip.h>
  17 #include <yaz/proto.h>
  18 #include <yaz/readconf.h>
  19 #include <yaz/pquery.h>
  20 #include <yaz/yaz-util.h>
  21 #include <yaz/nmem.h>
  22
  23 #define USE_TIMING 0
  24 #if USE_TIMING
  25 #include <yaz/timing.h>
  26 #endif
  27
  28 #include "pazpar2.h"
  29 #include "eventl.h"
  30 #include "command.h"
  31 #include "http.h"
  32 #include "termlists.h"
  33 #include "reclists.h"
  34 #include "relevance.h"
  35 #include "config.h"
  36
  37 #define PAZPAR2_VERSION "0.1"
  38 #define MAX_CHUNK 15
  39
  40 static void client_fatal(struct client *cl);
  41 static void connection_destroy(struct connection *co);
  42 static int client_prep_connection(struct client *cl);
  43 static void ingest_records(struct client *cl, Z_Records *r);
  44 static struct conf_retrievalprofile *database_retrieval_profile(struct database *db);
  45 void session_alert_watch(struct session *s, int what);
  46
  47 IOCHAN channel_list = 0;  // Master list of connections we're handling events to
  48
  49 static struct connection *connection_freelist = 0;
  50 static struct client *client_freelist = 0;
  51
  52 static struct host *hosts = 0;  // The hosts we know about
  53 static struct database *databases = 0; // The databases we know about
  54
  55 static char *client_states[] = {
  56     "Client_Connecting",
  57     "Client_Connected",
  58     "Client_Idle",
  59     "Client_Initializing",
  60     "Client_Searching",
  61     "Client_Presenting",
  62     "Client_Error",
  63     "Client_Failed",
  64     "Client_Disconnected",
  65     "Client_Stopped"
  66 };
  67
  68 struct parameters global_parameters =
  69 {
  70     0,
  71     30,
  72     "81",
  73     "Index Data PazPar2 (MasterKey)",
  74     PAZPAR2_VERSION,
  75     600, // 10 minutes
  76     60,
  77     100,
  78     MAX_CHUNK,
  79     0,
  80     0,
  81     0,
  82     0
  83 };
  84
  85 static int send_apdu(struct client *c, Z_APDU *a)
  86 {
  87     struct connection *co = c->connection;
  88     char *buf;
  89     int len, r;
  90
  91     if (!z_APDU(global_parameters.odr_out, &a, 0, 0))
  92     {
  93         odr_perror(global_parameters.odr_out, "Encoding APDU");
  94         abort();
  95     }
  96     buf = odr_getbuf(global_parameters.odr_out, &len, 0);
  97     r = cs_put(co->link, buf, len);
  98     if (r < 0)
  99     {
 100         yaz_log(YLOG_WARN, "cs_put: %s", cs_errmsg(cs_errno(co->link)));
 101         return -1;
 102     }
 103     else if (r == 1)
 104     {
 105         fprintf(stderr, "cs_put incomplete (ParaZ does not handle that)\n");
 106         exit(1);
 107     }
 108     odr_reset(global_parameters.odr_out); /* release the APDU structure  */
 109     co->state = Conn_Waiting;
 110     return 0;
 111 }
 112
 113
 114 static void send_init(IOCHAN i)
 115 {
 116     struct connection *co = iochan_getdata(i);
 117     struct client *cl = co->client;
 118     Z_APDU *a = zget_APDU(global_parameters.odr_out, Z_APDU_initRequest);
 119
 120     a->u.initRequest->implementationId = global_parameters.implementationId;
 121     a->u.initRequest->implementationName = global_parameters.implementationName;
 122     a->u.initRequest->implementationVersion =
 123         global_parameters.implementationVersion;
 124     ODR_MASK_SET(a->u.initRequest->options, Z_Options_search);
 125     ODR_MASK_SET(a->u.initRequest->options, Z_Options_present);
 126     ODR_MASK_SET(a->u.initRequest->options, Z_Options_namedResultSets);
 127
 128     ODR_MASK_SET(a->u.initRequest->protocolVersion, Z_ProtocolVersion_1);
 129     ODR_MASK_SET(a->u.initRequest->protocolVersion, Z_ProtocolVersion_2);
 130     ODR_MASK_SET(a->u.initRequest->protocolVersion, Z_ProtocolVersion_3);
 131     if (send_apdu(cl, a) >= 0)
 132     {
 133         iochan_setflags(i, EVENT_INPUT);
 134         cl->state = Client_Initializing;
 135     }
 136     else
 137         cl->state = Client_Error;
 138     odr_reset(global_parameters.odr_out);
 139 }
 140
 141 static void send_search(IOCHAN i)
 142 {
 143     struct connection *co = iochan_getdata(i);
 144     struct client *cl = co->client;
 145     struct session *se = cl->session;
 146     struct database *db = cl->database;
 147     Z_APDU *a = zget_APDU(global_parameters.odr_out, Z_APDU_searchRequest);
 148     int ndb, cerror, cpos;
 149     char **databaselist;
 150     Z_Query *zquery;
 151     struct ccl_rpn_node *cn;
 152     int ssub = 0, lslb = 100000, mspn = 10;
 153
 154     yaz_log(YLOG_DEBUG, "Sending search");
 155
 156     cn = ccl_find_str(global_parameters.ccl_filter, se->query, &cerror, &cpos);
 157     if (!cn)
 158         return;
 159     a->u.searchRequest->query = zquery = odr_malloc(global_parameters.odr_out,
 160             sizeof(Z_Query));
 161     zquery->which = Z_Query_type_1;
 162     zquery->u.type_1 = ccl_rpn_query(global_parameters.odr_out, cn);
 163     ccl_rpn_delete(cn);
 164
 165     for (ndb = 0; db->databases[ndb]; ndb++)
 166         ;
 167     databaselist = odr_malloc(global_parameters.odr_out, sizeof(char*) * ndb);
 168     for (ndb = 0; db->databases[ndb]; ndb++)
 169         databaselist[ndb] = db->databases[ndb];
 170
 171     a->u.presentRequest->preferredRecordSyntax =
 172             yaz_oidval_to_z3950oid(global_parameters.odr_out,
 173             CLASS_RECSYN, VAL_USMARC);
 174     a->u.searchRequest->smallSetUpperBound = &ssub;
 175     a->u.searchRequest->largeSetLowerBound = &lslb;
 176     a->u.searchRequest->mediumSetPresentNumber = &mspn;
 177     a->u.searchRequest->resultSetName = "Default";
 178     a->u.searchRequest->databaseNames = databaselist;
 179     a->u.searchRequest->num_databaseNames = ndb;
 180
 181     if (send_apdu(cl, a) >= 0)
 182     {
 183         iochan_setflags(i, EVENT_INPUT);
 184         cl->state = Client_Searching;
 185         cl->requestid = se->requestid;
 186     }
 187     else
 188         cl->state = Client_Error;
 189
 190     odr_reset(global_parameters.odr_out);
 191 }
 192
 193 static void send_present(IOCHAN i)
 194 {
 195     struct connection *co = iochan_getdata(i);
 196     struct client *cl = co->client;
 197     Z_APDU *a = zget_APDU(global_parameters.odr_out, Z_APDU_presentRequest);
 198     int toget;
 199     int start = cl->records + 1;
 200
 201     toget = global_parameters.chunk;
 202     if (toget > global_parameters.toget - cl->records)
 203         toget = global_parameters.toget - cl->records;
 204     if (toget > cl->hits - cl->records)
 205         toget = cl->hits - cl->records;
 206
 207     yaz_log(YLOG_DEBUG, "Trying to present %d records\n", toget);
 208
 209     a->u.presentRequest->resultSetStartPoint = &start;
 210     a->u.presentRequest->numberOfRecordsRequested = &toget;
 211
 212     a->u.presentRequest->resultSetId = "Default";
 213
 214     a->u.presentRequest->preferredRecordSyntax =
 215             yaz_oidval_to_z3950oid(global_parameters.odr_out,
 216             CLASS_RECSYN, VAL_USMARC);
 217
 218     if (send_apdu(cl, a) >= 0)
 219     {
 220         iochan_setflags(i, EVENT_INPUT);
 221         cl->state = Client_Presenting;
 222     }
 223     else
 224         cl->state = Client_Error;
 225     odr_reset(global_parameters.odr_out);
 226 }
 227
 228 static void do_initResponse(IOCHAN i, Z_APDU *a)
 229 {
 230     struct connection *co = iochan_getdata(i);
 231     struct client *cl = co->client;
 232     Z_InitResponse *r = a->u.initResponse;
 233
 234     yaz_log(YLOG_DEBUG, "Received init response");
 235
 236     if (*r->result)
 237     {
 238         cl->state = Client_Idle;
 239     }
 240     else
 241         cl->state = Client_Failed; // FIXME need to do something to the connection
 242 }
 243
 244 static void do_searchResponse(IOCHAN i, Z_APDU *a)
 245 {
 246     struct connection *co = iochan_getdata(i);
 247     struct client *cl = co->client;
 248     struct session *se = cl->session;
 249     Z_SearchResponse *r = a->u.searchResponse;
 250
 251     yaz_log(YLOG_DEBUG, "Searchresponse (status=%d)", *r->searchStatus);
 252
 253     if (*r->searchStatus)
 254     {
 255         cl->hits = *r->resultCount;
 256         se->total_hits += cl->hits;
 257         if (r->presentStatus && !*r->presentStatus && r->records)
 258         {
 259             yaz_log(YLOG_DEBUG, "Records in search response");
 260             cl->records += *r->numberOfRecordsReturned;
 261             ingest_records(cl, r->records);
 262         }
 263         cl->state = Client_Idle;
 264     }
 265     else
 266     {          /*"FAILED"*/
 267         cl->hits = 0;
 268         cl->state = Client_Error;
 269         if (r->records) {
 270             Z_Records *recs = r->records;
 271             if (recs->which == Z_Records_NSD)
 272             {
 273                 yaz_log(YLOG_WARN, "Non-surrogate diagnostic");
 274                 cl->diagnostic = *recs->u.nonSurrogateDiagnostic->condition;
 275                 cl->state = Client_Error;
 276             }
 277         }
 278     }
 279 }
 280
 281 char *normalize_mergekey(char *buf)
 282 {
 283     char *p = buf, *pout = buf;
 284
 285     while (*p)
 286     {
 287         while (*p && !isalnum(*p))
 288             p++;
 289         while (isalnum(*p))
 290             *(pout++) = tolower(*(p++));
 291         if (*p)
 292             *(pout++) = ' ';
 293         while (*p && !isalnum(*p))
 294             p++;
 295     }
 296     if (buf != pout)
 297         *pout = '\0';
 298
 299     return buf;
 300 }
 301
 302
 303 #ifdef GAGA
 304 // FIXME needs to be generalized. Should flexibly generate X lists per search
 305 static void extract_subject(struct session *s, const char *rec)
 306 {
 307     const char *field, *subfield;
 308
 309     while ((field = find_field(rec, "650")))
 310     {
 311         rec = field;
 312         if ((subfield = find_subfield(field, 'a')))
 313         {
 314             char *e, *ef;
 315             char buf[1024];
 316             int len;
 317
 318             ef = index(subfield, '\n');
 319             if (!ef)
 320                 return;
 321             if ((e = index(subfield, '\t')) && e < ef)
 322                 ef = e;
 323             while (ef > subfield && !isalpha(*(ef - 1)) && *(ef - 1) != ')')
 324                 ef--;
 325             len = ef - subfield;
 326             assert(len < 1023);
 327             memcpy(buf, subfield, len);
 328             buf[len] = '\0';
 329 #ifdef FIXME
 330             if (*buf)
 331                 termlist_insert(s->termlist, buf);
 332 #endif
 333         }
 334     }
 335 }
 336 #endif
 337
 338 static void add_facet(struct session *s, const char *type, const char *value)
 339 {
 340     int i;
 341
 342     for (i = 0; i < s->num_termlists; i++)
 343         if (!strcmp(s->termlists[i].name, type))
 344             break;
 345     if (i == s->num_termlists)
 346     {
 347         if (i == SESSION_MAX_TERMLISTS)
 348         {
 349             yaz_log(YLOG_FATAL, "Too many termlists");
 350             exit(1);
 351         }
 352         s->termlists[i].name = nmem_strdup(s->nmem, type);
 353         s->termlists[i].termlist = termlist_create(s->nmem, s->expected_maxrecs, 15);
 354         s->num_termlists = i + 1;
 355     }
 356     termlist_insert(s->termlists[i].termlist, value);
 357 }
 358
 359 static xmlDoc *normalize_record(struct client *cl, Z_External *rec)
 360 {
 361     struct conf_retrievalprofile *rprofile = cl->database->rprofile;
 362     struct conf_retrievalmap *m;
 363     xmlNode *res;
 364     xmlDoc *rdoc;
 365
 366     // First normalize to XML
 367     if (rprofile->native_syntax == Nativesyn_iso2709)
 368     {
 369         char *buf;
 370         int len;
 371         if (rec->which != Z_External_octet)
 372         {
 373             yaz_log(YLOG_WARN, "Unexpected external branch, probably BER");
 374             return 0;
 375         }
 376         buf = (char*) rec->u.octet_aligned->buf;
 377         len = rec->u.octet_aligned->len;
 378         if (yaz_marc_read_iso2709(rprofile->yaz_marc, buf, len) < 0)
 379         {
 380             yaz_log(YLOG_WARN, "Failed to decode MARC");
 381             return 0;
 382         }
 383         if (yaz_marc_write_xml(rprofile->yaz_marc, &res,
 384                     "http://www.loc.gov/MARC21/slim", 0, 0) < 0)
 385         {
 386             yaz_log(YLOG_WARN, "Failed to encode as XML");
 387             return 0;
 388         }
 389         rdoc = xmlNewDoc("1.0");
 390         xmlDocSetRootElement(rdoc, res);
 391     }
 392     else
 393     {
 394         yaz_log(YLOG_FATAL, "Unknown native_syntax in normalize_record");
 395         exit(1);
 396     }
 397     for (m = rprofile->maplist; m; m = m->next)
 398     {
 399         xmlDoc *new;
 400         if (m->type != Map_xslt)
 401         {
 402             yaz_log(YLOG_WARN, "Unknown map type");
 403             return 0;
 404         }
 405         if (!(new = xsltApplyStylesheet(m->stylesheet, rdoc, 0)))
 406         {
 407             yaz_log(YLOG_WARN, "XSLT transformation failed");
 408             return 0;
 409         }
 410         xmlFreeDoc(rdoc);
 411         rdoc = new;
 412     }
 413     if (global_parameters.dump_records)
 414     {
 415         fprintf(stderr, "Record:\n----------------\n");
 416         xmlDocFormatDump(stderr, rdoc, 1);
 417     }
 418     return rdoc;
 419 }
 420
 421 static struct record *ingest_record(struct client *cl, Z_External *rec)
 422 {
 423     xmlDoc *xdoc = normalize_record(cl, rec);
 424     xmlNode *root, *n;
 425     struct record *res, *head;
 426     struct session *se = cl->session;
 427     xmlChar *mergekey, *mergekey_norm;
 428
 429     if (!xdoc)
 430         return 0;
 431
 432     root = xmlDocGetRootElement(xdoc);
 433     if (!(mergekey = xmlGetProp(root, "mergekey")))
 434     {
 435         yaz_log(YLOG_WARN, "No mergekey found in record");
 436         xmlFreeDoc(xdoc);
 437         return 0;
 438     }
 439
 440     res = nmem_malloc(se->nmem, sizeof(struct record));
 441     res->next_cluster = 0;
 442     res->target_offset = -1;
 443     res->term_frequency_vec = 0;
 444     res->title = "Unknown";
 445     res->relevance = 0;
 446
 447     mergekey_norm = nmem_strdup(se->nmem, (char*) mergekey);
 448     xmlFree(mergekey);
 449     res->merge_key = normalize_mergekey(mergekey_norm);
 450
 451     head = reclist_insert(se->reclist, res);
 452     if (!head)
 453     {
 454         /* no room for record */
 455         xmlFreeDoc(xdoc);
 456         return 0;
 457     }
 458     relevance_newrec(se->relevance, head);
 459
 460     for (n = root->children; n; n = n->next)
 461     {
 462         if (n->type != XML_ELEMENT_NODE)
 463             continue;
 464         if (!strcmp(n->name, "facet"))
 465         {
 466             xmlChar *type = xmlGetProp(n, "type");
 467             xmlChar *value = xmlNodeListGetString(xdoc, n->children, 0);
 468             if (type && value)
 469             {
 470                 add_facet(se, type, value);
 471                 relevance_countwords(se->relevance, head, value, 1);
 472             }
 473             xmlFree(type);
 474             xmlFree(value);
 475         }
 476         else if (!strcmp(n->name, "metadata"))
 477         {
 478             xmlChar *type = xmlGetProp(n, "type");
 479             if (type && !strcmp(type, "title"))
 480             {
 481                 xmlChar *value = xmlNodeListGetString(xdoc, n->children, 0);
 482                 if (value)
 483                 {
 484                     res->title = nmem_strdup(se->nmem, value);
 485                     relevance_countwords(se->relevance, head, value, 4);
 486                     xmlFree(value);
 487                 }
 488             }
 489             xmlFree(type);
 490         }
 491         else
 492             yaz_log(YLOG_WARN, "Unexpected element %s in internal record", n->name);
 493     }
 494
 495     xmlFreeDoc(xdoc);
 496
 497     relevance_donerecord(se->relevance, head);
 498     se->total_records++;
 499
 500     return res;
 501 }
 502
 503 static void ingest_records(struct client *cl, Z_Records *r)
 504 {
 505 #if USE_TIMING
 506     yaz_timing_t t = yaz_timing_create();
 507 #endif
 508     struct record *rec;
 509     struct session *s = cl->session;
 510     Z_NamePlusRecordList *rlist;
 511     int i;
 512
 513     if (r->which != Z_Records_DBOSD)
 514         return;
 515     rlist = r->u.databaseOrSurDiagnostics;
 516     for (i = 0; i < rlist->num_records; i++)
 517     {
 518         Z_NamePlusRecord *npr = rlist->records[i];
 519
 520         if (npr->which != Z_NamePlusRecord_databaseRecord)
 521         {
 522             yaz_log(YLOG_WARN, "Unexpected record type, probably diagnostic");
 523             continue;
 524         }
 525
 526         rec = ingest_record(cl, npr->u.databaseRecord);
 527         if (!rec)
 528             continue;
 529     }
 530     if (s->watchlist[SESSION_WATCH_RECORDS].fun && rlist->num_records)
 531         session_alert_watch(s, SESSION_WATCH_RECORDS);
 532
 533 #if USE_TIMING
 534     yaz_timing_stop(t);
 535     yaz_log(YLOG_LOG, "ingest_records %6.5f %3.2f %3.2f",
 536             yaz_timing_get_real(t), yaz_timing_get_user(t),
 537             yaz_timing_get_sys(t));
 538     yaz_timing_destroy(&t);
 539 #endif
 540 }
 541
 542 static void do_presentResponse(IOCHAN i, Z_APDU *a)
 543 {
 544     struct connection *co = iochan_getdata(i);
 545     struct client *cl = co->client;
 546     Z_PresentResponse *r = a->u.presentResponse;
 547
 548     if (r->records) {
 549         Z_Records *recs = r->records;
 550         if (recs->which == Z_Records_NSD)
 551         {
 552             yaz_log(YLOG_WARN, "Non-surrogate diagnostic");
 553             cl->diagnostic = *recs->u.nonSurrogateDiagnostic->condition;
 554             cl->state = Client_Error;
 555         }
 556     }
 557
 558     if (!*r->presentStatus && cl->state != Client_Error)
 559     {
 560         yaz_log(YLOG_DEBUG, "Good Present response");
 561         cl->records += *r->numberOfRecordsReturned;
 562         ingest_records(cl, r->records);
 563         cl->state = Client_Idle;
 564     }
 565     else if (*r->presentStatus)
 566     {
 567         yaz_log(YLOG_WARN, "Bad Present response");
 568         cl->state = Client_Error;
 569     }
 570 }
 571
 572 static void handler(IOCHAN i, int event)
 573 {
 574     struct connection *co = iochan_getdata(i);
 575     struct client *cl = co->client;
 576     struct session *se = 0;
 577
 578     if (cl)
 579         se = cl->session;
 580     else
 581     {
 582         yaz_log(YLOG_WARN, "Destroying orphan connection");
 583         connection_destroy(co);
 584         return;
 585     }
 586
 587     if (co->state == Conn_Connecting && event & EVENT_OUTPUT)
 588     {
 589         int errcode;
 590         socklen_t errlen = sizeof(errcode);
 591
 592         if (getsockopt(cs_fileno(co->link), SOL_SOCKET, SO_ERROR, &errcode,
 593             &errlen) < 0 || errcode != 0)
 594         {
 595             client_fatal(cl);
 596             return;
 597         }
 598         else
 599         {
 600             yaz_log(YLOG_DEBUG, "Connect OK");
 601             co->state = Conn_Open;
 602             if (cl)
 603                 cl->state = Client_Connected;
 604         }
 605     }
 606
 607     else if (event & EVENT_INPUT)
 608     {
 609         int len = cs_get(co->link, &co->ibuf, &co->ibufsize);
 610
 611         if (len < 0)
 612         {
 613             yaz_log(YLOG_WARN|YLOG_ERRNO, "Error reading from Z server");
 614             connection_destroy(co);
 615             return;
 616         }
 617         else if (len == 0)
 618         {
 619             yaz_log(YLOG_WARN, "EOF reading from Z server");
 620             connection_destroy(co);
 621             return;
 622         }
 623         else if (len > 1) // We discard input if we have no connection
 624         {
 625             co->state = Conn_Open;
 626
 627             if (cl && (cl->requestid == se->requestid || cl->state == Client_Initializing))
 628             {
 629                 Z_APDU *a;
 630
 631                 odr_reset(global_parameters.odr_in);
 632                 odr_setbuf(global_parameters.odr_in, co->ibuf, len, 0);
 633                 if (!z_APDU(global_parameters.odr_in, &a, 0, 0))
 634                 {
 635                     client_fatal(cl);
 636                     return;
 637                 }
 638                 switch (a->which)
 639                 {
 640                     case Z_APDU_initResponse:
 641                         do_initResponse(i, a);
 642                         break;
 643                     case Z_APDU_searchResponse:
 644                         do_searchResponse(i, a);
 645                         break;
 646                     case Z_APDU_presentResponse:
 647                         do_presentResponse(i, a);
 648                         break;
 649                     default:
 650                         yaz_log(YLOG_WARN, "Unexpected result from server");
 651                         client_fatal(cl);
 652                         return;
 653                 }
 654                 // We aren't expecting staggered output from target
 655                 // if (cs_more(t->link))
 656                 //    iochan_setevent(i, EVENT_INPUT);
 657             }
 658             else  // we throw away response and go to idle mode
 659             {
 660                 yaz_log(YLOG_DEBUG, "Ignoring result of expired operation");
 661                 cl->state = Client_Idle;
 662             }
 663         }
 664         /* if len==1 we do nothing but wait for more input */
 665     }
 666
 667     if (cl->state == Client_Connected) {
 668         send_init(i);
 669     }
 670
 671     if (cl->state == Client_Idle)
 672     {
 673         if (cl->requestid != se->requestid && *se->query) {
 674             send_search(i);
 675         }
 676         else if (cl->hits > 0 && cl->records < global_parameters.toget &&
 677             cl->records < cl->hits) {
 678             send_present(i);
 679         }
 680     }
 681 }
 682
 683 // Disassociate connection from client
 684 static void connection_release(struct connection *co)
 685 {
 686     struct client *cl = co->client;
 687
 688     yaz_log(YLOG_DEBUG, "Connection release %s", co->host->hostport);
 689     if (!cl)
 690         return;
 691     cl->connection = 0;
 692     co->client = 0;
 693 }
 694
 695 // Close connection and recycle structure
 696 static void connection_destroy(struct connection *co)
 697 {
 698     struct host *h = co->host;
 699     cs_close(co->link);
 700     iochan_destroy(co->iochan);
 701
 702     yaz_log(YLOG_DEBUG, "Connection destroy %s", co->host->hostport);
 703     if (h->connections == co)
 704         h->connections = co->next;
 705     else
 706     {
 707         struct connection *pco;
 708         for (pco = h->connections; pco && pco->next != co; pco = pco->next)
 709             ;
 710         if (pco)
 711             pco->next = co->next;
 712         else
 713             abort();
 714     }
 715     if (co->client)
 716     {
 717         if (co->client->state != Client_Idle)
 718             co->client->state = Client_Disconnected;
 719         co->client->connection = 0;
 720     }
 721     co->next = connection_freelist;
 722     connection_freelist = co;
 723 }
 724
 725 // Creates a new connection for client, associated with the host of
 726 // client's database
 727 static struct connection *connection_create(struct client *cl)
 728 {
 729     struct connection *new;
 730     COMSTACK link;
 731     int res;
 732     void *addr;
 733
 734     yaz_log(YLOG_DEBUG, "Connection create %s", cl->database->url);
 735     if (!(link = cs_create(tcpip_type, 0, PROTO_Z3950)))
 736     {
 737         yaz_log(YLOG_FATAL|YLOG_ERRNO, "Failed to create comstack");
 738         exit(1);
 739     }
 740
 741     if (!(addr = cs_straddr(link, cl->database->host->ipport)))
 742     {
 743         yaz_log(YLOG_WARN|YLOG_ERRNO, "Lookup of IP address failed?");
 744         return 0;
 745     }
 746
 747     res = cs_connect(link, addr);
 748     if (res < 0)
 749     {
 750         yaz_log(YLOG_WARN|YLOG_ERRNO, "cs_connect %s", cl->database->url);
 751         return 0;
 752     }
 753
 754     if ((new = connection_freelist))
 755         connection_freelist = new->next;
 756     else
 757     {
 758         new = xmalloc(sizeof (struct connection));
 759         new->ibuf = 0;
 760         new->ibufsize = 0;
 761     }
 762     new->state = Conn_Connecting;
 763     new->host = cl->database->host;
 764     new->next = new->host->connections;
 765     new->host->connections = new;
 766     new->client = cl;
 767     cl->connection = new;
 768     new->link = link;
 769
 770     new->iochan = iochan_create(cs_fileno(link), handler, 0);
 771     iochan_setdata(new->iochan, new);
 772     new->iochan->next = channel_list;
 773     channel_list = new->iochan;
 774     return new;
 775 }
 776
 777 // Close connection and set state to error
 778 static void client_fatal(struct client *cl)
 779 {
 780     yaz_log(YLOG_WARN, "Fatal error from %s", cl->database->url);
 781     connection_destroy(cl->connection);
 782     cl->state = Client_Error;
 783 }
 784
 785 // Ensure that client has a connection associated
 786 static int client_prep_connection(struct client *cl)
 787 {
 788     struct connection *co;
 789     struct session *se = cl->session;
 790     struct host *host = cl->database->host;
 791
 792     co = cl->connection;
 793
 794     yaz_log(YLOG_DEBUG, "Client prep %s", cl->database->url);
 795
 796     if (!co)
 797     {
 798         // See if someone else has an idle connection
 799         // We should look at timestamps here to select the longest-idle connection
 800         for (co = host->connections; co; co = co->next)
 801             if (co->state == Conn_Open && (!co->client || co->client->session != se))
 802                 break;
 803         if (co)
 804         {
 805             connection_release(co);
 806             cl->connection = co;
 807             co->client = cl;
 808         }
 809         else
 810             co = connection_create(cl);
 811     }
 812     if (co)
 813     {
 814         if (co->state == Conn_Connecting)
 815         {
 816             cl->state = Client_Connecting;
 817             iochan_setflag(co->iochan, EVENT_OUTPUT);
 818         }
 819         else if (co->state == Conn_Open)
 820         {
 821             if (cl->state == Client_Error || cl->state == Client_Disconnected)
 822                 cl->state = Client_Idle;
 823             iochan_setflag(co->iochan, EVENT_OUTPUT);
 824         }
 825         return 1;
 826     }
 827     else
 828         return 0;
 829 }
 830
 831 // This function will most likely vanish when a proper target profile mechanism is
 832 // introduced.
 833 void load_simpletargets(const char *fn)
 834 {
 835     FILE *f = fopen(fn, "r");
 836     char line[256];
 837
 838     if (!f)
 839     {
 840         yaz_log(YLOG_WARN|YLOG_ERRNO, "open %s", fn);
 841         exit(1);
 842     }
 843
 844     while (fgets(line, 255, f))
 845     {
 846         char *url, *db;
 847         struct host *host;
 848         struct database *database;
 849
 850         if (strncmp(line, "target ", 7))
 851             continue;
 852         url = line + 7;
 853         url[strlen(url) - 1] = '\0';
 854         yaz_log(YLOG_DEBUG, "Target: %s", url);
 855         if ((db = strchr(url, '/')))
 856             *(db++) = '\0';
 857         else
 858             db = "Default";
 859
 860         for (host = hosts; host; host = host->next)
 861             if (!strcmp(url, host->hostport))
 862                 break;
 863         if (!host)
 864         {
 865             struct addrinfo *addrinfo, hints;
 866             char *port;
 867             char ipport[128];
 868             unsigned char addrbuf[4];
 869             int res;
 870
 871             host = xmalloc(sizeof(struct host));
 872             host->hostport = xstrdup(url);
 873             host->connections = 0;
 874
 875             if ((port = strchr(url, ':')))
 876                 *(port++) = '\0';
 877             else
 878                 port = "210";
 879
 880             hints.ai_flags = 0;
 881             hints.ai_family = PF_INET;
 882             hints.ai_socktype = SOCK_STREAM;
 883             hints.ai_protocol = IPPROTO_TCP;
 884             hints.ai_addrlen = 0;
 885             hints.ai_addr = 0;
 886             hints.ai_canonname = 0;
 887             hints.ai_next = 0;
 888             // This is not robust code. It assumes that getaddrinfo returns AF_INET
 889             // address.
 890             if ((res = getaddrinfo(url, port, &hints, &addrinfo)))
 891             {
 892                 yaz_log(YLOG_WARN, "Failed to resolve %s: %s", url, gai_strerror(res));
 893                 xfree(host->hostport);
 894                 xfree(host);
 895                 continue;
 896             }
 897             assert(addrinfo->ai_family == PF_INET);
 898             memcpy(addrbuf, &((struct sockaddr_in*)addrinfo->ai_addr)->sin_addr.s_addr, 4);
 899             sprintf(ipport, "%hhd.%hhd.%hhd.%hhd:%s",
 900                     addrbuf[0], addrbuf[1], addrbuf[2], addrbuf[3], port);
 901             host->ipport = xstrdup(ipport);
 902             freeaddrinfo(addrinfo);
 903             host->next = hosts;
 904             hosts = host;
 905         }
 906         database = xmalloc(sizeof(struct database));
 907         database->host = host;
 908         database->url = xmalloc(strlen(url) + strlen(db) + 2);
 909         strcpy(database->url, url);
 910         strcat(database->url, "/");
 911         strcat(database->url, db);
 912
 913         database->databases = xmalloc(2 * sizeof(char *));
 914         database->databases[0] = xstrdup(db);
 915         database->databases[1] = 0;
 916         database->errors = 0;
 917         database->qprofile = 0;
 918         database->rprofile = database_retrieval_profile(database);
 919         database->next = databases;
 920         databases = database;
 921
 922     }
 923     fclose(f);
 924 }
 925
 926 static void pull_terms(NMEM nmem, struct ccl_rpn_node *n, char **termlist, int *num)
 927 {
 928     switch (n->kind)
 929     {
 930         case CCL_RPN_AND:
 931         case CCL_RPN_OR:
 932         case CCL_RPN_NOT:
 933         case CCL_RPN_PROX:
 934             pull_terms(nmem, n->u.p[0], termlist, num);
 935             pull_terms(nmem, n->u.p[1], termlist, num);
 936             break;
 937         case CCL_RPN_TERM:
 938             termlist[(*num)++] = nmem_strdup(nmem, n->u.t.term);
 939             break;
 940         default: // NOOP
 941             break;
 942     }
 943 }
 944
 945 // Extract terms from query into null-terminated termlist
 946 static int extract_terms(NMEM nmem, char *query, char **termlist)
 947 {
 948     int error, pos;
 949     struct ccl_rpn_node *n;
 950     int num = 0;
 951
 952     n = ccl_find_str(global_parameters.ccl_filter, query, &error, &pos);
 953     if (!n)
 954         return -1;
 955     pull_terms(nmem, n, termlist, &num);
 956     termlist[num] = 0;
 957     ccl_rpn_delete(n);
 958     return 0;
 959 }
 960
 961 static struct client *client_create(void)
 962 {
 963     struct client *r;
 964     if (client_freelist)
 965     {
 966         r = client_freelist;
 967         client_freelist = client_freelist->next;
 968     }
 969     else
 970         r = xmalloc(sizeof(struct client));
 971     r->database = 0;
 972     r->connection = 0;
 973     r->session = 0;
 974     r->hits = 0;
 975     r->records = 0;
 976     r->setno = 0;
 977     r->requestid = -1;
 978     r->diagnostic = 0;
 979     r->state = Client_Disconnected;
 980     r->next = 0;
 981     return r;
 982 }
 983
 984 void client_destroy(struct client *c)
 985 {
 986     struct session *se = c->session;
 987     if (c == se->clients)
 988         se->clients = c->next;
 989     else
 990     {
 991         struct client *cc;
 992         for (cc = se->clients; cc && cc->next != c; cc = cc->next)
 993             ;
 994         if (cc)
 995             cc->next = c->next;
 996     }
 997     if (c->connection)
 998         connection_release(c->connection);
 999     c->next = client_freelist;
1000     client_freelist = c;
1001 }
1002
1003 void session_set_watch(struct session *s, int what, session_watchfun fun, void *data)
1004 {
1005     s->watchlist[what].fun = fun;
1006     s->watchlist[what].data = data;
1007 }
1008
1009 void session_alert_watch(struct session *s, int what)
1010 {
1011     if (!s->watchlist[what].fun)
1012         return;
1013     (*s->watchlist[what].fun)(s->watchlist[what].data);
1014     s->watchlist[what].fun = 0;
1015     s->watchlist[what].data = 0;
1016 }
1017
1018 // This needs to be extended with selection criteria
1019 static struct conf_retrievalprofile *database_retrieval_profile(struct database *db)
1020 {
1021     if (!config)
1022     {
1023         yaz_log(YLOG_FATAL, "Must load configuration (-f)");
1024         exit(1);
1025     }
1026     if (!config->retrievalprofiles)
1027     {
1028         yaz_log(YLOG_FATAL, "No retrieval profiles defined");
1029     }
1030     return config->retrievalprofiles;
1031 }
1032
1033 // This should be extended with parameters to control selection criteria
1034 // Associates a set of clients with a session;
1035 int select_targets(struct session *se)
1036 {
1037     struct database *db;
1038     int c = 0;
1039
1040     while (se->clients)
1041         client_destroy(se->clients);
1042     for (db = databases; db; db = db->next)
1043     {
1044         struct client *cl = client_create();
1045         cl->database = db;
1046         cl->session = se;
1047         cl->next = se->clients;
1048         se->clients = cl;
1049         c++;
1050     }
1051     return c;
1052 }
1053
1054 int session_active_clients(struct session *s)
1055 {
1056     struct client *c;
1057     int res = 0;
1058
1059     for (c = s->clients; c; c = c->next)
1060         if (c->connection && (c->state == Client_Connecting ||
1061                     c->state == Client_Initializing ||
1062                     c->state == Client_Searching ||
1063                     c->state == Client_Presenting))
1064             res++;
1065
1066     return res;
1067 }
1068
1069 char *search(struct session *se, char *query)
1070 {
1071     int live_channels = 0;
1072     struct client *cl;
1073
1074     yaz_log(YLOG_DEBUG, "Search");
1075
1076     strcpy(se->query, query);
1077     se->requestid++;
1078     nmem_reset(se->nmem);
1079     for (cl = se->clients; cl; cl = cl->next)
1080     {
1081         cl->hits = -1;
1082         cl->records = 0;
1083         cl->diagnostic = 0;
1084
1085         if (client_prep_connection(cl))
1086             live_channels++;
1087     }
1088     if (live_channels)
1089     {
1090         char *p[512];
1091         int maxrecs = live_channels * global_parameters.toget;
1092         se->num_termlists = 0;
1093         se->reclist = reclist_create(se->nmem, maxrecs);
1094         extract_terms(se->nmem, query, p);
1095         se->relevance = relevance_create(se->nmem, (const char **) p, maxrecs);
1096         se->total_records = se->total_hits = 0;
1097         se->expected_maxrecs = maxrecs;
1098     }
1099     else
1100         return "NOTARGETS";
1101
1102     return 0;
1103 }
1104
1105 void destroy_session(struct session *s)
1106 {
1107     yaz_log(YLOG_LOG, "Destroying session");
1108     while (s->clients)
1109         client_destroy(s->clients);
1110     nmem_destroy(s->nmem);
1111     wrbuf_free(s->wrbuf, 1);
1112 }
1113
1114 struct session *new_session()
1115 {
1116     int i;
1117     struct session *session = xmalloc(sizeof(*session));
1118
1119     yaz_log(YLOG_DEBUG, "New pazpar2 session");
1120
1121     session->total_hits = 0;
1122     session->total_records = 0;
1123     session->num_termlists = 0;
1124     session->reclist = 0;
1125     session->requestid = -1;
1126     session->clients = 0;
1127     session->expected_maxrecs = 0;
1128     session->query[0] = '\0';
1129     session->nmem = nmem_create();
1130     session->wrbuf = wrbuf_alloc();
1131     for (i = 0; i <= SESSION_WATCH_MAX; i++)
1132     {
1133         session->watchlist[i].data = 0;
1134         session->watchlist[i].fun = 0;
1135     }
1136
1137     select_targets(session);
1138
1139     return session;
1140 }
1141
1142 struct hitsbytarget *hitsbytarget(struct session *se, int *count)
1143 {
1144     static struct hitsbytarget res[1000]; // FIXME MM
1145     struct client *cl;
1146
1147     *count = 0;
1148     for (cl = se->clients; cl; cl = cl->next)
1149     {
1150         strcpy(res[*count].id, cl->database->host->hostport);
1151         res[*count].hits = cl->hits;
1152         res[*count].records = cl->records;
1153         res[*count].diagnostic = cl->diagnostic;
1154         res[*count].state = client_states[cl->state];
1155         res[*count].connected  = cl->connection ? 1 : 0;
1156         (*count)++;
1157     }
1158
1159     return res;
1160 }
1161
1162 struct termlist_score **termlist(struct session *s, const char *name, int *num)
1163 {
1164     int i;
1165
1166     for (i = 0; i < s->num_termlists; i++)
1167         if (!strcmp(s->termlists[i].name, name))
1168             return termlist_highscore(s->termlists[i].termlist, num);
1169     return 0;
1170 }
1171
1172 #ifdef MISSING_HEADERS
1173 void report_nmem_stats(void)
1174 {
1175     size_t in_use, is_free;
1176
1177     nmem_get_memory_in_use(&in_use);
1178     nmem_get_memory_free(&is_free);
1179
1180     yaz_log(YLOG_LOG, "nmem stat: use=%ld free=%ld",
1181             (long) in_use, (long) is_free);
1182 }
1183 #endif
1184
1185 struct record **show(struct session *s, int start, int *num, int *total,
1186                      int *sumhits, NMEM nmem_show)
1187 {
1188     struct record **recs = nmem_malloc(nmem_show, *num
1189                                        * sizeof(struct record *));
1190     int i;
1191 #if USE_TIMING
1192     yaz_timing_t t = yaz_timing_create();
1193 #endif
1194     relevance_prepare_read(s->relevance, s->reclist);
1195
1196     *total = s->reclist->num_records;
1197     *sumhits = s->total_hits;
1198
1199     for (i = 0; i < start; i++)
1200         if (!reclist_read_record(s->reclist))
1201         {
1202             *num = 0;
1203             recs = 0;
1204             break;
1205         }
1206
1207     for (i = 0; i < *num; i++)
1208     {
1209         struct record *r = reclist_read_record(s->reclist);
1210         if (!r)
1211         {
1212             *num = i;
1213             break;
1214         }
1215         recs[i] = r;
1216     }
1217 #if USE_TIMING
1218     yaz_timing_stop(t);
1219     yaz_log(YLOG_LOG, "show %6.5f %3.2f %3.2f",
1220             yaz_timing_get_real(t), yaz_timing_get_user(t),
1221             yaz_timing_get_sys(t));
1222     yaz_timing_destroy(&t);
1223 #endif
1224     return recs;
1225 }
1226
1227 void statistics(struct session *se, struct statistics *stat)
1228 {
1229     struct client *cl;
1230     int count = 0;
1231
1232     bzero(stat, sizeof(*stat));
1233     for (cl = se->clients; cl; cl = cl->next)
1234     {
1235         if (!cl->connection)
1236             stat->num_no_connection++;
1237         switch (cl->state)
1238         {
1239             case Client_Connecting: stat->num_connecting++; break;
1240             case Client_Initializing: stat->num_initializing++; break;
1241             case Client_Searching: stat->num_searching++; break;
1242             case Client_Presenting: stat->num_presenting++; break;
1243             case Client_Idle: stat->num_idle++; break;
1244             case Client_Failed: stat->num_failed++; break;
1245             case Client_Error: stat->num_error++; break;
1246             default: break;
1247         }
1248         count++;
1249     }
1250     stat->num_hits = se->total_hits;
1251     stat->num_records = se->total_records;
1252
1253     stat->num_clients = count;
1254 }
1255
1256 static CCL_bibset load_cclfile(const char *fn)
1257 {
1258     CCL_bibset res = ccl_qual_mk();
1259     if (ccl_qual_fname(res, fn) < 0)
1260     {
1261         yaz_log(YLOG_FATAL|YLOG_ERRNO, "%s", fn);
1262         exit(1);
1263     }
1264     return res;
1265 }
1266
1267 int main(int argc, char **argv)
1268 {
1269     int ret;
1270     char *arg;
1271     int setport = 0;
1272
1273     if (signal(SIGPIPE, SIG_IGN) < 0)
1274         yaz_log(YLOG_WARN|YLOG_ERRNO, "signal");
1275
1276     yaz_log_init(YLOG_DEFAULT_LEVEL, "pazpar2", 0);
1277
1278     while ((ret = options("f:x:c:h:p:C:s:d", argv, argc, &arg)) != -2)
1279     {
1280         switch (ret) {
1281             case 'f':
1282                 if (!read_config(arg))
1283                     exit(1);
1284                 break;
1285             case 'c':
1286                 command_init(atoi(arg));
1287                 setport++;
1288                 break;
1289             case 'h':
1290                 http_init(arg);
1291                 setport++;
1292                 break;
1293             case 'C':
1294                 global_parameters.ccl_filter = load_cclfile(arg);
1295                 break;
1296             case 'p':
1297                 http_set_proxyaddr(arg);
1298                 break;
1299             case 's':
1300                 load_simpletargets(arg);
1301                 break;
1302             case 'd':
1303                 global_parameters.dump_records = 1;
1304                 break;
1305             default:
1306                 fprintf(stderr, "Usage: pazpar2\n"
1307                         "    -f configfile\n"
1308                         "    -h [host:]port          (REST protocol listener)\n"
1309                         "    -c cmdport              (telnet-style)\n"
1310                         "    -C cclconfig\n"
1311                         "    -s simpletargetfile\n"
1312                         "    -p hostname[:portno]    (HTTP proxy)\n");
1313                 exit(1);
1314         }
1315     }
1316
1317     if (!setport)
1318     {
1319         fprintf(stderr, "Set command port with -h or -c\n");
1320         exit(1);
1321     }
1322
1323     global_parameters.ccl_filter = load_cclfile("../etc/default.bib");
1324     global_parameters.yaz_marc = yaz_marc_create();
1325     yaz_marc_subfield_str(global_parameters.yaz_marc, "\t");
1326     global_parameters.odr_in = odr_createmem(ODR_DECODE);
1327     global_parameters.odr_out = odr_createmem(ODR_ENCODE);
1328
1329     event_loop(&channel_list);
1330
1331     return 0;
1332 }
1333
1334 /*
1335  * Local variables:
1336  * c-basic-offset: 4
1337  * indent-tabs-mode: nil
1338  * End:
1339  * vim: shiftwidth=4 tabstop=8 expandtab
1340  */