src/pazpar2.c

   1 /* $Id: pazpar2.c,v 1.19 2007-01-08 18:32:35 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 #if HAVE_CONFIG_H
  24 #include "cconfig.h"
  25 #endif
  26
  27 #define USE_TIMING 0
  28 #if USE_TIMING
  29 #include <yaz/timing.h>
  30 #endif
  31
  32 #include "pazpar2.h"
  33 #include "eventl.h"
  34 #include "http.h"
  35 #include "termlists.h"
  36 #include "reclists.h"
  37 #include "relevance.h"
  38 #include "config.h"
  39
  40 #define MAX_CHUNK 15
  41
  42 static void client_fatal(struct client *cl);
  43 static void connection_destroy(struct connection *co);
  44 static int client_prep_connection(struct client *cl);
  45 static void ingest_records(struct client *cl, Z_Records *r);
  46 static struct conf_retrievalprofile *database_retrieval_profile(struct database *db);
  47 void session_alert_watch(struct session *s, int what);
  48
  49 IOCHAN channel_list = 0;  // Master list of connections we're handling events to
  50
  51 static struct connection *connection_freelist = 0;
  52 static struct client *client_freelist = 0;
  53
  54 static struct host *hosts = 0;  // The hosts we know about
  55 static struct database *databases = 0; // The databases we know about
  56
  57 static char *client_states[] = {
  58     "Client_Connecting",
  59     "Client_Connected",
  60     "Client_Idle",
  61     "Client_Initializing",
  62     "Client_Searching",
  63     "Client_Presenting",
  64     "Client_Error",
  65     "Client_Failed",
  66     "Client_Disconnected",
  67     "Client_Stopped"
  68 };
  69
  70 // Note: Some things in this structure will eventually move to configuration
  71 struct parameters global_parameters =
  72 {
  73     0,
  74     0,
  75     30,
  76     "81",
  77     "Index Data PazPar2 (MasterKey)",
  78     VERSION,
  79     600, // 10 minutes
  80     60,
  81     100,
  82     MAX_CHUNK,
  83     0,
  84     0,
  85     0,
  86     0
  87 };
  88
  89 static int send_apdu(struct client *c, Z_APDU *a)
  90 {
  91     struct connection *co = c->connection;
  92     char *buf;
  93     int len, r;
  94
  95     if (!z_APDU(global_parameters.odr_out, &a, 0, 0))
  96     {
  97         odr_perror(global_parameters.odr_out, "Encoding APDU");
  98         abort();
  99     }
 100     buf = odr_getbuf(global_parameters.odr_out, &len, 0);
 101     r = cs_put(co->link, buf, len);
 102     if (r < 0)
 103     {
 104         yaz_log(YLOG_WARN, "cs_put: %s", cs_errmsg(cs_errno(co->link)));
 105         return -1;
 106     }
 107     else if (r == 1)
 108     {
 109         fprintf(stderr, "cs_put incomplete (ParaZ does not handle that)\n");
 110         exit(1);
 111     }
 112     odr_reset(global_parameters.odr_out); /* release the APDU structure  */
 113     co->state = Conn_Waiting;
 114     return 0;
 115 }
 116
 117
 118 static void send_init(IOCHAN i)
 119 {
 120     struct connection *co = iochan_getdata(i);
 121     struct client *cl = co->client;
 122     Z_APDU *a = zget_APDU(global_parameters.odr_out, Z_APDU_initRequest);
 123
 124     a->u.initRequest->implementationId = global_parameters.implementationId;
 125     a->u.initRequest->implementationName = global_parameters.implementationName;
 126     a->u.initRequest->implementationVersion =
 127         global_parameters.implementationVersion;
 128     ODR_MASK_SET(a->u.initRequest->options, Z_Options_search);
 129     ODR_MASK_SET(a->u.initRequest->options, Z_Options_present);
 130     ODR_MASK_SET(a->u.initRequest->options, Z_Options_namedResultSets);
 131
 132     ODR_MASK_SET(a->u.initRequest->protocolVersion, Z_ProtocolVersion_1);
 133     ODR_MASK_SET(a->u.initRequest->protocolVersion, Z_ProtocolVersion_2);
 134     ODR_MASK_SET(a->u.initRequest->protocolVersion, Z_ProtocolVersion_3);
 135     if (send_apdu(cl, a) >= 0)
 136     {
 137         iochan_setflags(i, EVENT_INPUT);
 138         cl->state = Client_Initializing;
 139     }
 140     else
 141         cl->state = Client_Error;
 142     odr_reset(global_parameters.odr_out);
 143 }
 144
 145 static void send_search(IOCHAN i)
 146 {
 147     struct connection *co = iochan_getdata(i);
 148     struct client *cl = co->client;
 149     struct session *se = cl->session;
 150     struct database *db = cl->database;
 151     Z_APDU *a = zget_APDU(global_parameters.odr_out, Z_APDU_searchRequest);
 152     int ndb, cerror, cpos;
 153     char **databaselist;
 154     Z_Query *zquery;
 155     struct ccl_rpn_node *cn;
 156     int ssub = 0, lslb = 100000, mspn = 10;
 157
 158     yaz_log(YLOG_DEBUG, "Sending search");
 159
 160     cn = ccl_find_str(global_parameters.ccl_filter, se->query, &cerror, &cpos);
 161     if (!cn)
 162         return;
 163     a->u.searchRequest->query = zquery = odr_malloc(global_parameters.odr_out,
 164             sizeof(Z_Query));
 165     zquery->which = Z_Query_type_1;
 166     zquery->u.type_1 = ccl_rpn_query(global_parameters.odr_out, cn);
 167     ccl_rpn_delete(cn);
 168
 169     for (ndb = 0; db->databases[ndb]; ndb++)
 170         ;
 171     databaselist = odr_malloc(global_parameters.odr_out, sizeof(char*) * ndb);
 172     for (ndb = 0; db->databases[ndb]; ndb++)
 173         databaselist[ndb] = db->databases[ndb];
 174
 175     a->u.presentRequest->preferredRecordSyntax =
 176             yaz_oidval_to_z3950oid(global_parameters.odr_out,
 177             CLASS_RECSYN, VAL_USMARC);
 178     a->u.searchRequest->smallSetUpperBound = &ssub;
 179     a->u.searchRequest->largeSetLowerBound = &lslb;
 180     a->u.searchRequest->mediumSetPresentNumber = &mspn;
 181     a->u.searchRequest->resultSetName = "Default";
 182     a->u.searchRequest->databaseNames = databaselist;
 183     a->u.searchRequest->num_databaseNames = ndb;
 184
 185     if (send_apdu(cl, a) >= 0)
 186     {
 187         iochan_setflags(i, EVENT_INPUT);
 188         cl->state = Client_Searching;
 189         cl->requestid = se->requestid;
 190     }
 191     else
 192         cl->state = Client_Error;
 193
 194     odr_reset(global_parameters.odr_out);
 195 }
 196
 197 static void send_present(IOCHAN i)
 198 {
 199     struct connection *co = iochan_getdata(i);
 200     struct client *cl = co->client;
 201     Z_APDU *a = zget_APDU(global_parameters.odr_out, Z_APDU_presentRequest);
 202     int toget;
 203     int start = cl->records + 1;
 204
 205     toget = global_parameters.chunk;
 206     if (toget > global_parameters.toget - cl->records)
 207         toget = global_parameters.toget - cl->records;
 208     if (toget > cl->hits - cl->records)
 209         toget = cl->hits - cl->records;
 210
 211     yaz_log(YLOG_DEBUG, "Trying to present %d records\n", toget);
 212
 213     a->u.presentRequest->resultSetStartPoint = &start;
 214     a->u.presentRequest->numberOfRecordsRequested = &toget;
 215
 216     a->u.presentRequest->resultSetId = "Default";
 217
 218     a->u.presentRequest->preferredRecordSyntax =
 219             yaz_oidval_to_z3950oid(global_parameters.odr_out,
 220             CLASS_RECSYN, VAL_USMARC);
 221
 222     if (send_apdu(cl, a) >= 0)
 223     {
 224         iochan_setflags(i, EVENT_INPUT);
 225         cl->state = Client_Presenting;
 226     }
 227     else
 228         cl->state = Client_Error;
 229     odr_reset(global_parameters.odr_out);
 230 }
 231
 232 static void do_initResponse(IOCHAN i, Z_APDU *a)
 233 {
 234     struct connection *co = iochan_getdata(i);
 235     struct client *cl = co->client;
 236     Z_InitResponse *r = a->u.initResponse;
 237
 238     yaz_log(YLOG_DEBUG, "Received init response");
 239
 240     if (*r->result)
 241     {
 242         cl->state = Client_Idle;
 243     }
 244     else
 245         cl->state = Client_Failed; // FIXME need to do something to the connection
 246 }
 247
 248 static void do_searchResponse(IOCHAN i, Z_APDU *a)
 249 {
 250     struct connection *co = iochan_getdata(i);
 251     struct client *cl = co->client;
 252     struct session *se = cl->session;
 253     Z_SearchResponse *r = a->u.searchResponse;
 254
 255     yaz_log(YLOG_DEBUG, "Searchresponse (status=%d)", *r->searchStatus);
 256
 257     if (*r->searchStatus)
 258     {
 259         cl->hits = *r->resultCount;
 260         se->total_hits += cl->hits;
 261         if (r->presentStatus && !*r->presentStatus && r->records)
 262         {
 263             yaz_log(YLOG_DEBUG, "Records in search response");
 264             cl->records += *r->numberOfRecordsReturned;
 265             ingest_records(cl, r->records);
 266         }
 267         cl->state = Client_Idle;
 268     }
 269     else
 270     {          /*"FAILED"*/
 271         cl->hits = 0;
 272         cl->state = Client_Error;
 273         if (r->records) {
 274             Z_Records *recs = r->records;
 275             if (recs->which == Z_Records_NSD)
 276             {
 277                 yaz_log(YLOG_WARN, "Non-surrogate diagnostic");
 278                 cl->diagnostic = *recs->u.nonSurrogateDiagnostic->condition;
 279                 cl->state = Client_Error;
 280             }
 281         }
 282     }
 283 }
 284
 285 char *normalize_mergekey(char *buf)
 286 {
 287     char *p = buf, *pout = buf;
 288
 289     while (*p)
 290     {
 291         while (*p && !isalnum(*p))
 292             p++;
 293         while (isalnum(*p))
 294             *(pout++) = tolower(*(p++));
 295         if (*p)
 296             *(pout++) = ' ';
 297         while (*p && !isalnum(*p))
 298             p++;
 299     }
 300     if (buf != pout)
 301         *pout = '\0';
 302
 303     return buf;
 304 }
 305
 306
 307 #ifdef GAGA
 308 // FIXME needs to be generalized. Should flexibly generate X lists per search
 309 static void extract_subject(struct session *s, const char *rec)
 310 {
 311     const char *field, *subfield;
 312
 313     while ((field = find_field(rec, "650")))
 314     {
 315         rec = field;
 316         if ((subfield = find_subfield(field, 'a')))
 317         {
 318             char *e, *ef;
 319             char buf[1024];
 320             int len;
 321
 322             ef = index(subfield, '\n');
 323             if (!ef)
 324                 return;
 325             if ((e = index(subfield, '\t')) && e < ef)
 326                 ef = e;
 327             while (ef > subfield && !isalpha(*(ef - 1)) && *(ef - 1) != ')')
 328                 ef--;
 329             len = ef - subfield;
 330             assert(len < 1023);
 331             memcpy(buf, subfield, len);
 332             buf[len] = '\0';
 333 #ifdef FIXME
 334             if (*buf)
 335                 termlist_insert(s->termlist, buf);
 336 #endif
 337         }
 338     }
 339 }
 340 #endif
 341
 342 static void add_facet(struct session *s, const char *type, const char *value)
 343 {
 344     int i;
 345
 346     for (i = 0; i < s->num_termlists; i++)
 347         if (!strcmp(s->termlists[i].name, type))
 348             break;
 349     if (i == s->num_termlists)
 350     {
 351         if (i == SESSION_MAX_TERMLISTS)
 352         {
 353             yaz_log(YLOG_FATAL, "Too many termlists");
 354             exit(1);
 355         }
 356         s->termlists[i].name = nmem_strdup(s->nmem, type);
 357         s->termlists[i].termlist = termlist_create(s->nmem, s->expected_maxrecs, 15);
 358         s->num_termlists = i + 1;
 359     }
 360     termlist_insert(s->termlists[i].termlist, value);
 361 }
 362
 363 int yaz_marc_write_xml();
 364
 365 static xmlDoc *normalize_record(struct client *cl, Z_External *rec)
 366 {
 367     struct conf_retrievalprofile *rprofile = cl->database->rprofile;
 368     struct conf_retrievalmap *m;
 369     xmlNode *res;
 370     xmlDoc *rdoc;
 371
 372     // First normalize to XML
 373     if (rprofile->native_syntax == Nativesyn_iso2709)
 374     {
 375         char *buf;
 376         int len;
 377         if (rec->which != Z_External_octet)
 378         {
 379             yaz_log(YLOG_WARN, "Unexpected external branch, probably BER");
 380             return 0;
 381         }
 382         buf = (char*) rec->u.octet_aligned->buf;
 383         len = rec->u.octet_aligned->len;
 384         if (yaz_marc_read_iso2709(rprofile->yaz_marc, buf, len) < 0)
 385         {
 386             yaz_log(YLOG_WARN, "Failed to decode MARC");
 387             return 0;
 388         }
 389         if (yaz_marc_write_xml(rprofile->yaz_marc, &res,
 390                     "http://www.loc.gov/MARC21/slim", 0, 0) < 0)
 391         {
 392             yaz_log(YLOG_WARN, "Failed to encode as XML");
 393             return 0;
 394         }
 395         rdoc = xmlNewDoc("1.0");
 396         xmlDocSetRootElement(rdoc, res);
 397     }
 398     else
 399     {
 400         yaz_log(YLOG_FATAL, "Unknown native_syntax in normalize_record");
 401         exit(1);
 402     }
 403     for (m = rprofile->maplist; m; m = m->next)
 404     {
 405         xmlDoc *new;
 406         if (m->type != Map_xslt)
 407         {
 408             yaz_log(YLOG_WARN, "Unknown map type");
 409             return 0;
 410         }
 411         if (!(new = xsltApplyStylesheet(m->stylesheet, rdoc, 0)))
 412         {
 413             yaz_log(YLOG_WARN, "XSLT transformation failed");
 414             return 0;
 415         }
 416         xmlFreeDoc(rdoc);
 417         rdoc = new;
 418     }
 419     if (global_parameters.dump_records)
 420     {
 421         fprintf(stderr, "Record:\n----------------\n");
 422         xmlDocFormatDump(stderr, rdoc, 1);
 423     }
 424     return rdoc;
 425 }
 426
 427 static struct record *ingest_record(struct client *cl, Z_External *rec)
 428 {
 429     xmlDoc *xdoc = normalize_record(cl, rec);
 430     xmlNode *root, *n;
 431     struct record *res;
 432     struct record_cluster *cluster;
 433     struct session *se = cl->session;
 434     xmlChar *mergekey, *mergekey_norm;
 435     xmlChar *type;
 436     xmlChar *value;
 437     struct conf_service *service = global_parameters.server->service;
 438
 439     if (!xdoc)
 440         return 0;
 441
 442     root = xmlDocGetRootElement(xdoc);
 443     if (!(mergekey = xmlGetProp(root, "mergekey")))
 444     {
 445         yaz_log(YLOG_WARN, "No mergekey found in record");
 446         xmlFreeDoc(xdoc);
 447         return 0;
 448     }
 449
 450     res = nmem_malloc(se->nmem, sizeof(struct record));
 451     res->next = 0;
 452     res->target_offset = -1;
 453     res->term_frequency_vec = 0;
 454     res->metadata = nmem_malloc(se->nmem,
 455             sizeof(struct record_metadata*) * service->num_metadata);
 456     bzero(res->metadata, sizeof(struct record_metadata*) * service->num_metadata);
 457     res->relevance = 0;
 458
 459     mergekey_norm = nmem_strdup(se->nmem, (char*) mergekey);
 460     xmlFree(mergekey);
 461     normalize_mergekey(mergekey_norm);
 462
 463     cluster = reclist_insert(se->reclist, res, mergekey_norm);
 464     if (!cluster)
 465     {
 466         /* no room for record */
 467         xmlFreeDoc(xdoc);
 468         return 0;
 469     }
 470     relevance_newrec(se->relevance, cluster);
 471
 472     type = value = 0;
 473     for (n = root->children; n; n = n->next)
 474     {
 475         if (type)
 476             xmlFree(type);
 477         if (value)
 478             xmlFree(value);
 479         type = value = 0;
 480
 481         if (n->type != XML_ELEMENT_NODE)
 482             continue;
 483         if (!strcmp(n->name, "metadata"))
 484         {
 485             type = xmlGetProp(n, "type");
 486             value = xmlNodeListGetString(xdoc, n->children, 0);
 487             struct conf_metadata *md = 0;
 488             struct record_metadata **wheretoput, *newm;
 489             int imeta;
 490
 491             // First, find out what field we're looking at
 492             for (imeta = 0; imeta < service->num_metadata; imeta++)
 493                 if (!strcmp(type, service->metadata[imeta].name))
 494                 {
 495                     md = &service->metadata[imeta];
 496                     break;
 497                 }
 498             if (!md)
 499             {
 500                 yaz_log(YLOG_WARN, "Ignoring unknown metadata element: %s", type);
 501                 continue;
 502             }
 503
 504             // Find out where we are putting it
 505             if (md->merge == Metadata_merge_no)
 506                 wheretoput = &res->metadata[imeta];
 507             else
 508                 wheretoput = &cluster->metadata[imeta];
 509
 510             // Put it there
 511             newm = nmem_malloc(se->nmem, sizeof(struct record_metadata));
 512             newm->next = 0;
 513             if (md->type == Metadata_type_generic)
 514             {
 515                 newm->data.text = nmem_strdup(se->nmem, value);
 516             }
 517             else
 518             {
 519                 yaz_log(YLOG_WARN, "Unknown type in metadata element %s", type);
 520                 continue;
 521             }
 522             if (md->merge == Metadata_merge_unique)
 523             {
 524                 struct record_metadata *mnode;
 525                 for (mnode = *wheretoput; mnode; mnode = mnode->next)
 526                     if (!strcmp(mnode->data.text, mnode->data.text))
 527                         break;
 528                 if (!mnode)
 529                 {
 530                     newm->next = *wheretoput;
 531                     *wheretoput = newm;
 532                 }
 533             }
 534             else if (md->merge == Metadata_merge_longest)
 535             {
 536                 if (!*wheretoput ||
 537                         strlen(newm->data.text) > strlen((*wheretoput)->data.text))
 538                 *wheretoput = newm;
 539             }
 540             else if (md->merge == Metadata_merge_all || md->merge == Metadata_merge_no)
 541             {
 542                 newm->next = *wheretoput;
 543                 *wheretoput = newm;
 544             }
 545             else
 546                 yaz_log(YLOG_WARN, "Don't know how to merge on element name %s", md->name);
 547
 548             relevance_countwords(se->relevance, cluster, value, 4);
 549             if (md->termlist)
 550                 add_facet(se, type, value);
 551             xmlFree(type);
 552             xmlFree(value);
 553             type = value = 0;
 554         }
 555         else
 556             yaz_log(YLOG_WARN, "Unexpected element %s in internal record", n->name);
 557     }
 558
 559     xmlFreeDoc(xdoc);
 560
 561     relevance_donerecord(se->relevance, cluster);
 562     se->total_records++;
 563
 564     return res;
 565 }
 566
 567 static void ingest_records(struct client *cl, Z_Records *r)
 568 {
 569 #if USE_TIMING
 570     yaz_timing_t t = yaz_timing_create();
 571 #endif
 572     struct record *rec;
 573     struct session *s = cl->session;
 574     Z_NamePlusRecordList *rlist;
 575     int i;
 576
 577     if (r->which != Z_Records_DBOSD)
 578         return;
 579     rlist = r->u.databaseOrSurDiagnostics;
 580     for (i = 0; i < rlist->num_records; i++)
 581     {
 582         Z_NamePlusRecord *npr = rlist->records[i];
 583
 584         if (npr->which != Z_NamePlusRecord_databaseRecord)
 585         {
 586             yaz_log(YLOG_WARN, "Unexpected record type, probably diagnostic");
 587             continue;
 588         }
 589
 590         rec = ingest_record(cl, npr->u.databaseRecord);
 591         if (!rec)
 592             continue;
 593     }
 594     if (s->watchlist[SESSION_WATCH_RECORDS].fun && rlist->num_records)
 595         session_alert_watch(s, SESSION_WATCH_RECORDS);
 596
 597 #if USE_TIMING
 598     yaz_timing_stop(t);
 599     yaz_log(YLOG_LOG, "ingest_records %6.5f %3.2f %3.2f",
 600             yaz_timing_get_real(t), yaz_timing_get_user(t),
 601             yaz_timing_get_sys(t));
 602     yaz_timing_destroy(&t);
 603 #endif
 604 }
 605
 606 static void do_presentResponse(IOCHAN i, Z_APDU *a)
 607 {
 608     struct connection *co = iochan_getdata(i);
 609     struct client *cl = co->client;
 610     Z_PresentResponse *r = a->u.presentResponse;
 611
 612     if (r->records) {
 613         Z_Records *recs = r->records;
 614         if (recs->which == Z_Records_NSD)
 615         {
 616             yaz_log(YLOG_WARN, "Non-surrogate diagnostic");
 617             cl->diagnostic = *recs->u.nonSurrogateDiagnostic->condition;
 618             cl->state = Client_Error;
 619         }
 620     }
 621
 622     if (!*r->presentStatus && cl->state != Client_Error)
 623     {
 624         yaz_log(YLOG_DEBUG, "Good Present response");
 625         cl->records += *r->numberOfRecordsReturned;
 626         ingest_records(cl, r->records);
 627         cl->state = Client_Idle;
 628     }
 629     else if (*r->presentStatus)
 630     {
 631         yaz_log(YLOG_WARN, "Bad Present response");
 632         cl->state = Client_Error;
 633     }
 634 }
 635
 636 static void handler(IOCHAN i, int event)
 637 {
 638     struct connection *co = iochan_getdata(i);
 639     struct client *cl = co->client;
 640     struct session *se = 0;
 641
 642     if (cl)
 643         se = cl->session;
 644     else
 645     {
 646         yaz_log(YLOG_WARN, "Destroying orphan connection");
 647         connection_destroy(co);
 648         return;
 649     }
 650
 651     if (co->state == Conn_Connecting && event & EVENT_OUTPUT)
 652     {
 653         int errcode;
 654         socklen_t errlen = sizeof(errcode);
 655
 656         if (getsockopt(cs_fileno(co->link), SOL_SOCKET, SO_ERROR, &errcode,
 657             &errlen) < 0 || errcode != 0)
 658         {
 659             client_fatal(cl);
 660             return;
 661         }
 662         else
 663         {
 664             yaz_log(YLOG_DEBUG, "Connect OK");
 665             co->state = Conn_Open;
 666             if (cl)
 667                 cl->state = Client_Connected;
 668         }
 669     }
 670
 671     else if (event & EVENT_INPUT)
 672     {
 673         int len = cs_get(co->link, &co->ibuf, &co->ibufsize);
 674
 675         if (len < 0)
 676         {
 677             yaz_log(YLOG_WARN|YLOG_ERRNO, "Error reading from Z server");
 678             connection_destroy(co);
 679             return;
 680         }
 681         else if (len == 0)
 682         {
 683             yaz_log(YLOG_WARN, "EOF reading from Z server");
 684             connection_destroy(co);
 685             return;
 686         }
 687         else if (len > 1) // We discard input if we have no connection
 688         {
 689             co->state = Conn_Open;
 690
 691             if (cl && (cl->requestid == se->requestid || cl->state == Client_Initializing))
 692             {
 693                 Z_APDU *a;
 694
 695                 odr_reset(global_parameters.odr_in);
 696                 odr_setbuf(global_parameters.odr_in, co->ibuf, len, 0);
 697                 if (!z_APDU(global_parameters.odr_in, &a, 0, 0))
 698                 {
 699                     client_fatal(cl);
 700                     return;
 701                 }
 702                 switch (a->which)
 703                 {
 704                     case Z_APDU_initResponse:
 705                         do_initResponse(i, a);
 706                         break;
 707                     case Z_APDU_searchResponse:
 708                         do_searchResponse(i, a);
 709                         break;
 710                     case Z_APDU_presentResponse:
 711                         do_presentResponse(i, a);
 712                         break;
 713                     default:
 714                         yaz_log(YLOG_WARN, "Unexpected result from server");
 715                         client_fatal(cl);
 716                         return;
 717                 }
 718                 // We aren't expecting staggered output from target
 719                 // if (cs_more(t->link))
 720                 //    iochan_setevent(i, EVENT_INPUT);
 721             }
 722             else  // we throw away response and go to idle mode
 723             {
 724                 yaz_log(YLOG_DEBUG, "Ignoring result of expired operation");
 725                 cl->state = Client_Idle;
 726             }
 727         }
 728         /* if len==1 we do nothing but wait for more input */
 729     }
 730
 731     if (cl->state == Client_Connected) {
 732         send_init(i);
 733     }
 734
 735     if (cl->state == Client_Idle)
 736     {
 737         if (cl->requestid != se->requestid && *se->query) {
 738             send_search(i);
 739         }
 740         else if (cl->hits > 0 && cl->records < global_parameters.toget &&
 741             cl->records < cl->hits) {
 742             send_present(i);
 743         }
 744     }
 745 }
 746
 747 // Disassociate connection from client
 748 static void connection_release(struct connection *co)
 749 {
 750     struct client *cl = co->client;
 751
 752     yaz_log(YLOG_DEBUG, "Connection release %s", co->host->hostport);
 753     if (!cl)
 754         return;
 755     cl->connection = 0;
 756     co->client = 0;
 757 }
 758
 759 // Close connection and recycle structure
 760 static void connection_destroy(struct connection *co)
 761 {
 762     struct host *h = co->host;
 763     cs_close(co->link);
 764     iochan_destroy(co->iochan);
 765
 766     yaz_log(YLOG_DEBUG, "Connection destroy %s", co->host->hostport);
 767     if (h->connections == co)
 768         h->connections = co->next;
 769     else
 770     {
 771         struct connection *pco;
 772         for (pco = h->connections; pco && pco->next != co; pco = pco->next)
 773             ;
 774         if (pco)
 775             pco->next = co->next;
 776         else
 777             abort();
 778     }
 779     if (co->client)
 780     {
 781         if (co->client->state != Client_Idle)
 782             co->client->state = Client_Disconnected;
 783         co->client->connection = 0;
 784     }
 785     co->next = connection_freelist;
 786     connection_freelist = co;
 787 }
 788
 789 // Creates a new connection for client, associated with the host of
 790 // client's database
 791 static struct connection *connection_create(struct client *cl)
 792 {
 793     struct connection *new;
 794     COMSTACK link;
 795     int res;
 796     void *addr;
 797
 798     yaz_log(YLOG_DEBUG, "Connection create %s", cl->database->url);
 799     if (!(link = cs_create(tcpip_type, 0, PROTO_Z3950)))
 800     {
 801         yaz_log(YLOG_FATAL|YLOG_ERRNO, "Failed to create comstack");
 802         exit(1);
 803     }
 804
 805     if (!(addr = cs_straddr(link, cl->database->host->ipport)))
 806     {
 807         yaz_log(YLOG_WARN|YLOG_ERRNO, "Lookup of IP address failed?");
 808         return 0;
 809     }
 810
 811     res = cs_connect(link, addr);
 812     if (res < 0)
 813     {
 814         yaz_log(YLOG_WARN|YLOG_ERRNO, "cs_connect %s", cl->database->url);
 815         return 0;
 816     }
 817
 818     if ((new = connection_freelist))
 819         connection_freelist = new->next;
 820     else
 821     {
 822         new = xmalloc(sizeof (struct connection));
 823         new->ibuf = 0;
 824         new->ibufsize = 0;
 825     }
 826     new->state = Conn_Connecting;
 827     new->host = cl->database->host;
 828     new->next = new->host->connections;
 829     new->host->connections = new;
 830     new->client = cl;
 831     cl->connection = new;
 832     new->link = link;
 833
 834     new->iochan = iochan_create(cs_fileno(link), handler, 0);
 835     iochan_setdata(new->iochan, new);
 836     new->iochan->next = channel_list;
 837     channel_list = new->iochan;
 838     return new;
 839 }
 840
 841 // Close connection and set state to error
 842 static void client_fatal(struct client *cl)
 843 {
 844     yaz_log(YLOG_WARN, "Fatal error from %s", cl->database->url);
 845     connection_destroy(cl->connection);
 846     cl->state = Client_Error;
 847 }
 848
 849 // Ensure that client has a connection associated
 850 static int client_prep_connection(struct client *cl)
 851 {
 852     struct connection *co;
 853     struct session *se = cl->session;
 854     struct host *host = cl->database->host;
 855
 856     co = cl->connection;
 857
 858     yaz_log(YLOG_DEBUG, "Client prep %s", cl->database->url);
 859
 860     if (!co)
 861     {
 862         // See if someone else has an idle connection
 863         // We should look at timestamps here to select the longest-idle connection
 864         for (co = host->connections; co; co = co->next)
 865             if (co->state == Conn_Open && (!co->client || co->client->session != se))
 866                 break;
 867         if (co)
 868         {
 869             connection_release(co);
 870             cl->connection = co;
 871             co->client = cl;
 872         }
 873         else
 874             co = connection_create(cl);
 875     }
 876     if (co)
 877     {
 878         if (co->state == Conn_Connecting)
 879         {
 880             cl->state = Client_Connecting;
 881             iochan_setflag(co->iochan, EVENT_OUTPUT);
 882         }
 883         else if (co->state == Conn_Open)
 884         {
 885             if (cl->state == Client_Error || cl->state == Client_Disconnected)
 886                 cl->state = Client_Idle;
 887             iochan_setflag(co->iochan, EVENT_OUTPUT);
 888         }
 889         return 1;
 890     }
 891     else
 892         return 0;
 893 }
 894
 895 // This function will most likely vanish when a proper target profile mechanism is
 896 // introduced.
 897 void load_simpletargets(const char *fn)
 898 {
 899     FILE *f = fopen(fn, "r");
 900     char line[256];
 901
 902     if (!f)
 903     {
 904         yaz_log(YLOG_WARN|YLOG_ERRNO, "open %s", fn);
 905         exit(1);
 906     }
 907
 908     while (fgets(line, 255, f))
 909     {
 910         char *url, *db;
 911         struct host *host;
 912         struct database *database;
 913
 914         if (strncmp(line, "target ", 7))
 915             continue;
 916         url = line + 7;
 917         url[strlen(url) - 1] = '\0';
 918         yaz_log(YLOG_DEBUG, "Target: %s", url);
 919         if ((db = strchr(url, '/')))
 920             *(db++) = '\0';
 921         else
 922             db = "Default";
 923
 924         for (host = hosts; host; host = host->next)
 925             if (!strcmp(url, host->hostport))
 926                 break;
 927         if (!host)
 928         {
 929             struct addrinfo *addrinfo, hints;
 930             char *port;
 931             char ipport[128];
 932             unsigned char addrbuf[4];
 933             int res;
 934
 935             host = xmalloc(sizeof(struct host));
 936             host->hostport = xstrdup(url);
 937             host->connections = 0;
 938
 939             if ((port = strchr(url, ':')))
 940                 *(port++) = '\0';
 941             else
 942                 port = "210";
 943
 944             hints.ai_flags = 0;
 945             hints.ai_family = PF_INET;
 946             hints.ai_socktype = SOCK_STREAM;
 947             hints.ai_protocol = IPPROTO_TCP;
 948             hints.ai_addrlen = 0;
 949             hints.ai_addr = 0;
 950             hints.ai_canonname = 0;
 951             hints.ai_next = 0;
 952             // This is not robust code. It assumes that getaddrinfo returns AF_INET
 953             // address.
 954             if ((res = getaddrinfo(url, port, &hints, &addrinfo)))
 955             {
 956                 yaz_log(YLOG_WARN, "Failed to resolve %s: %s", url, gai_strerror(res));
 957                 xfree(host->hostport);
 958                 xfree(host);
 959                 continue;
 960             }
 961             assert(addrinfo->ai_family == PF_INET);
 962             memcpy(addrbuf, &((struct sockaddr_in*)addrinfo->ai_addr)->sin_addr.s_addr, 4);
 963             sprintf(ipport, "%hhd.%hhd.%hhd.%hhd:%s",
 964                     addrbuf[0], addrbuf[1], addrbuf[2], addrbuf[3], port);
 965             host->ipport = xstrdup(ipport);
 966             freeaddrinfo(addrinfo);
 967             host->next = hosts;
 968             hosts = host;
 969         }
 970         database = xmalloc(sizeof(struct database));
 971         database->host = host;
 972         database->url = xmalloc(strlen(url) + strlen(db) + 2);
 973         strcpy(database->url, url);
 974         strcat(database->url, "/");
 975         strcat(database->url, db);
 976
 977         database->databases = xmalloc(2 * sizeof(char *));
 978         database->databases[0] = xstrdup(db);
 979         database->databases[1] = 0;
 980         database->errors = 0;
 981         database->qprofile = 0;
 982         database->rprofile = database_retrieval_profile(database);
 983         database->next = databases;
 984         databases = database;
 985
 986     }
 987     fclose(f);
 988 }
 989
 990 static void pull_terms(NMEM nmem, struct ccl_rpn_node *n, char **termlist, int *num)
 991 {
 992     switch (n->kind)
 993     {
 994         case CCL_RPN_AND:
 995         case CCL_RPN_OR:
 996         case CCL_RPN_NOT:
 997         case CCL_RPN_PROX:
 998             pull_terms(nmem, n->u.p[0], termlist, num);
 999             pull_terms(nmem, n->u.p[1], termlist, num);
1000             break;
1001         case CCL_RPN_TERM:
1002             termlist[(*num)++] = nmem_strdup(nmem, n->u.t.term);
1003             break;
1004         default: // NOOP
1005             break;
1006     }
1007 }
1008
1009 // Extract terms from query into null-terminated termlist
1010 static int extract_terms(NMEM nmem, char *query, char **termlist)
1011 {
1012     int error, pos;
1013     struct ccl_rpn_node *n;
1014     int num = 0;
1015
1016     n = ccl_find_str(global_parameters.ccl_filter, query, &error, &pos);
1017     if (!n)
1018         return -1;
1019     pull_terms(nmem, n, termlist, &num);
1020     termlist[num] = 0;
1021     ccl_rpn_delete(n);
1022     return 0;
1023 }
1024
1025 static struct client *client_create(void)
1026 {
1027     struct client *r;
1028     if (client_freelist)
1029     {
1030         r = client_freelist;
1031         client_freelist = client_freelist->next;
1032     }
1033     else
1034         r = xmalloc(sizeof(struct client));
1035     r->database = 0;
1036     r->connection = 0;
1037     r->session = 0;
1038     r->hits = 0;
1039     r->records = 0;
1040     r->setno = 0;
1041     r->requestid = -1;
1042     r->diagnostic = 0;
1043     r->state = Client_Disconnected;
1044     r->next = 0;
1045     return r;
1046 }
1047
1048 void client_destroy(struct client *c)
1049 {
1050     struct session *se = c->session;
1051     if (c == se->clients)
1052         se->clients = c->next;
1053     else
1054     {
1055         struct client *cc;
1056         for (cc = se->clients; cc && cc->next != c; cc = cc->next)
1057             ;
1058         if (cc)
1059             cc->next = c->next;
1060     }
1061     if (c->connection)
1062         connection_release(c->connection);
1063     c->next = client_freelist;
1064     client_freelist = c;
1065 }
1066
1067 void session_set_watch(struct session *s, int what, session_watchfun fun, void *data)
1068 {
1069     s->watchlist[what].fun = fun;
1070     s->watchlist[what].data = data;
1071 }
1072
1073 void session_alert_watch(struct session *s, int what)
1074 {
1075     if (!s->watchlist[what].fun)
1076         return;
1077     (*s->watchlist[what].fun)(s->watchlist[what].data);
1078     s->watchlist[what].fun = 0;
1079     s->watchlist[what].data = 0;
1080 }
1081
1082 // This needs to be extended with selection criteria
1083 static struct conf_retrievalprofile *database_retrieval_profile(struct database *db)
1084 {
1085     if (!config)
1086     {
1087         yaz_log(YLOG_FATAL, "Must load configuration (-f)");
1088         exit(1);
1089     }
1090     if (!config->retrievalprofiles)
1091     {
1092         yaz_log(YLOG_FATAL, "No retrieval profiles defined");
1093     }
1094     return config->retrievalprofiles;
1095 }
1096
1097 // This should be extended with parameters to control selection criteria
1098 // Associates a set of clients with a session;
1099 int select_targets(struct session *se)
1100 {
1101     struct database *db;
1102     int c = 0;
1103
1104     while (se->clients)
1105         client_destroy(se->clients);
1106     for (db = databases; db; db = db->next)
1107     {
1108         struct client *cl = client_create();
1109         cl->database = db;
1110         cl->session = se;
1111         cl->next = se->clients;
1112         se->clients = cl;
1113         c++;
1114     }
1115     return c;
1116 }
1117
1118 int session_active_clients(struct session *s)
1119 {
1120     struct client *c;
1121     int res = 0;
1122
1123     for (c = s->clients; c; c = c->next)
1124         if (c->connection && (c->state == Client_Connecting ||
1125                     c->state == Client_Initializing ||
1126                     c->state == Client_Searching ||
1127                     c->state == Client_Presenting))
1128             res++;
1129
1130     return res;
1131 }
1132
1133 char *search(struct session *se, char *query)
1134 {
1135     int live_channels = 0;
1136     struct client *cl;
1137
1138     yaz_log(YLOG_DEBUG, "Search");
1139
1140     strcpy(se->query, query);
1141     se->requestid++;
1142     nmem_reset(se->nmem);
1143     for (cl = se->clients; cl; cl = cl->next)
1144     {
1145         cl->hits = -1;
1146         cl->records = 0;
1147         cl->diagnostic = 0;
1148
1149         if (client_prep_connection(cl))
1150             live_channels++;
1151     }
1152     if (live_channels)
1153     {
1154         char *p[512];
1155         int maxrecs = live_channels * global_parameters.toget;
1156         se->num_termlists = 0;
1157         se->reclist = reclist_create(se->nmem, maxrecs);
1158         extract_terms(se->nmem, query, p);
1159         se->relevance = relevance_create(se->nmem, (const char **) p, maxrecs);
1160         se->total_records = se->total_hits = 0;
1161         se->expected_maxrecs = maxrecs;
1162     }
1163     else
1164         return "NOTARGETS";
1165
1166     return 0;
1167 }
1168
1169 void destroy_session(struct session *s)
1170 {
1171     yaz_log(YLOG_LOG, "Destroying session");
1172     while (s->clients)
1173         client_destroy(s->clients);
1174     nmem_destroy(s->nmem);
1175     wrbuf_free(s->wrbuf, 1);
1176 }
1177
1178 struct session *new_session()
1179 {
1180     int i;
1181     struct session *session = xmalloc(sizeof(*session));
1182
1183     yaz_log(YLOG_DEBUG, "New pazpar2 session");
1184
1185     session->total_hits = 0;
1186     session->total_records = 0;
1187     session->num_termlists = 0;
1188     session->reclist = 0;
1189     session->requestid = -1;
1190     session->clients = 0;
1191     session->expected_maxrecs = 0;
1192     session->query[0] = '\0';
1193     session->nmem = nmem_create();
1194     session->wrbuf = wrbuf_alloc();
1195     for (i = 0; i <= SESSION_WATCH_MAX; i++)
1196     {
1197         session->watchlist[i].data = 0;
1198         session->watchlist[i].fun = 0;
1199     }
1200
1201     select_targets(session);
1202
1203     return session;
1204 }
1205
1206 struct hitsbytarget *hitsbytarget(struct session *se, int *count)
1207 {
1208     static struct hitsbytarget res[1000]; // FIXME MM
1209     struct client *cl;
1210
1211     *count = 0;
1212     for (cl = se->clients; cl; cl = cl->next)
1213     {
1214         strcpy(res[*count].id, cl->database->host->hostport);
1215         res[*count].hits = cl->hits;
1216         res[*count].records = cl->records;
1217         res[*count].diagnostic = cl->diagnostic;
1218         res[*count].state = client_states[cl->state];
1219         res[*count].connected  = cl->connection ? 1 : 0;
1220         (*count)++;
1221     }
1222
1223     return res;
1224 }
1225
1226 struct termlist_score **termlist(struct session *s, const char *name, int *num)
1227 {
1228     int i;
1229
1230     for (i = 0; i < s->num_termlists; i++)
1231         if (!strcmp(s->termlists[i].name, name))
1232             return termlist_highscore(s->termlists[i].termlist, num);
1233     return 0;
1234 }
1235
1236 #ifdef MISSING_HEADERS
1237 void report_nmem_stats(void)
1238 {
1239     size_t in_use, is_free;
1240
1241     nmem_get_memory_in_use(&in_use);
1242     nmem_get_memory_free(&is_free);
1243
1244     yaz_log(YLOG_LOG, "nmem stat: use=%ld free=%ld",
1245             (long) in_use, (long) is_free);
1246 }
1247 #endif
1248
1249 struct record_cluster **show(struct session *s, int start, int *num, int *total,
1250                      int *sumhits, NMEM nmem_show)
1251 {
1252     struct record_cluster **recs = nmem_malloc(nmem_show, *num
1253                                        * sizeof(struct record_cluster *));
1254     int i;
1255 #if USE_TIMING
1256     yaz_timing_t t = yaz_timing_create();
1257 #endif
1258     relevance_prepare_read(s->relevance, s->reclist);
1259
1260     *total = s->reclist->num_records;
1261     *sumhits = s->total_hits;
1262
1263     for (i = 0; i < start; i++)
1264         if (!reclist_read_record(s->reclist))
1265         {
1266             *num = 0;
1267             recs = 0;
1268             break;
1269         }
1270
1271     for (i = 0; i < *num; i++)
1272     {
1273         struct record_cluster *r = reclist_read_record(s->reclist);
1274         if (!r)
1275         {
1276             *num = i;
1277             break;
1278         }
1279         recs[i] = r;
1280     }
1281 #if USE_TIMING
1282     yaz_timing_stop(t);
1283     yaz_log(YLOG_LOG, "show %6.5f %3.2f %3.2f",
1284             yaz_timing_get_real(t), yaz_timing_get_user(t),
1285             yaz_timing_get_sys(t));
1286     yaz_timing_destroy(&t);
1287 #endif
1288     return recs;
1289 }
1290
1291 void statistics(struct session *se, struct statistics *stat)
1292 {
1293     struct client *cl;
1294     int count = 0;
1295
1296     bzero(stat, sizeof(*stat));
1297     for (cl = se->clients; cl; cl = cl->next)
1298     {
1299         if (!cl->connection)
1300             stat->num_no_connection++;
1301         switch (cl->state)
1302         {
1303             case Client_Connecting: stat->num_connecting++; break;
1304             case Client_Initializing: stat->num_initializing++; break;
1305             case Client_Searching: stat->num_searching++; break;
1306             case Client_Presenting: stat->num_presenting++; break;
1307             case Client_Idle: stat->num_idle++; break;
1308             case Client_Failed: stat->num_failed++; break;
1309             case Client_Error: stat->num_error++; break;
1310             default: break;
1311         }
1312         count++;
1313     }
1314     stat->num_hits = se->total_hits;
1315     stat->num_records = se->total_records;
1316
1317     stat->num_clients = count;
1318 }
1319
1320 static CCL_bibset load_cclfile(const char *fn)
1321 {
1322     CCL_bibset res = ccl_qual_mk();
1323     if (ccl_qual_fname(res, fn) < 0)
1324     {
1325         yaz_log(YLOG_FATAL|YLOG_ERRNO, "%s", fn);
1326         exit(1);
1327     }
1328     return res;
1329 }
1330
1331 int main(int argc, char **argv)
1332 {
1333     int ret;
1334     char *arg;
1335     int setport = 0;
1336
1337     if (signal(SIGPIPE, SIG_IGN) < 0)
1338         yaz_log(YLOG_WARN|YLOG_ERRNO, "signal");
1339
1340     yaz_log_init(YLOG_DEFAULT_LEVEL, "pazpar2", 0);
1341
1342     while ((ret = options("f:x:h:p:C:s:d", argv, argc, &arg)) != -2)
1343     {
1344         switch (ret) {
1345             case 'f':
1346                 if (!read_config(arg))
1347                     exit(1);
1348                 break;
1349             case 'h':
1350                 http_init(arg);
1351                 setport++;
1352                 break;
1353             case 'C':
1354                 global_parameters.ccl_filter = load_cclfile(arg);
1355                 break;
1356             case 'p':
1357                 http_set_proxyaddr(arg);
1358                 break;
1359             case 's':
1360                 load_simpletargets(arg);
1361                 break;
1362             case 'd':
1363                 global_parameters.dump_records = 1;
1364                 break;
1365             default:
1366                 fprintf(stderr, "Usage: pazpar2\n"
1367                         "    -f configfile\n"
1368                         "    -h [host:]port          (REST protocol listener)\n"
1369                         "    -C cclconfig\n"
1370                         "    -s simpletargetfile\n"
1371                         "    -p hostname[:portno]    (HTTP proxy)\n");
1372                 exit(1);
1373         }
1374     }
1375
1376     if (!config)
1377     {
1378         yaz_log(YLOG_FATAL, "Load config with -f");
1379         exit(1);
1380     }
1381     global_parameters.server = config->servers;
1382
1383     if (!setport)
1384     {
1385         fprintf(stderr, "Set command port with -h\n");
1386         exit(1);
1387     }
1388
1389     global_parameters.ccl_filter = load_cclfile("../etc/default.bib");
1390     global_parameters.yaz_marc = yaz_marc_create();
1391     yaz_marc_subfield_str(global_parameters.yaz_marc, "\t");
1392     global_parameters.odr_in = odr_createmem(ODR_DECODE);
1393     global_parameters.odr_out = odr_createmem(ODR_ENCODE);
1394
1395     event_loop(&channel_list);
1396
1397     return 0;
1398 }
1399
1400 /*
1401  * Local variables:
1402  * c-basic-offset: 4
1403  * indent-tabs-mode: nil
1404  * End:
1405  * vim: shiftwidth=4 tabstop=8 expandtab
1406  */