src/pazpar2.c

   1 /* $Id: pazpar2.c,v 1.17 2007-01-06 04:54:58 quinn Exp $ */;
   2
   3 #include <stdlib.h>
   4 #include <stdio.h>
   5 #include <string.h>
   6 #include <sys/time.h>
   7 #include <unistd.h>
   8 #include <sys/socket.h>
   9 #include <netdb.h>
  10 #include <signal.h>
  11 #include <ctype.h>
  12 #include <assert.h>
  13
  14 #include <yaz/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 "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 > global_parameters.toget - cl->records)
 202         toget = global_parameters.toget - cl->records;
 203     if (toget > cl->hits - cl->records)
 204         toget = cl->hits - cl->records;
 205
 206     yaz_log(YLOG_DEBUG, "Trying to present %d records\n", toget);
 207
 208     a->u.presentRequest->resultSetStartPoint = &start;
 209     a->u.presentRequest->numberOfRecordsRequested = &toget;
 210
 211     a->u.presentRequest->resultSetId = "Default";
 212
 213     a->u.presentRequest->preferredRecordSyntax =
 214             yaz_oidval_to_z3950oid(global_parameters.odr_out,
 215             CLASS_RECSYN, VAL_USMARC);
 216
 217     if (send_apdu(cl, a) >= 0)
 218     {
 219         iochan_setflags(i, EVENT_INPUT);
 220         cl->state = Client_Presenting;
 221     }
 222     else
 223         cl->state = Client_Error;
 224     odr_reset(global_parameters.odr_out);
 225 }
 226
 227 static void do_initResponse(IOCHAN i, Z_APDU *a)
 228 {
 229     struct connection *co = iochan_getdata(i);
 230     struct client *cl = co->client;
 231     Z_InitResponse *r = a->u.initResponse;
 232
 233     yaz_log(YLOG_DEBUG, "Received init response");
 234
 235     if (*r->result)
 236     {
 237         cl->state = Client_Idle;
 238     }
 239     else
 240         cl->state = Client_Failed; // FIXME need to do something to the connection
 241 }
 242
 243 static void do_searchResponse(IOCHAN i, Z_APDU *a)
 244 {
 245     struct connection *co = iochan_getdata(i);
 246     struct client *cl = co->client;
 247     struct session *se = cl->session;
 248     Z_SearchResponse *r = a->u.searchResponse;
 249
 250     yaz_log(YLOG_DEBUG, "Searchresponse (status=%d)", *r->searchStatus);
 251
 252     if (*r->searchStatus)
 253     {
 254         cl->hits = *r->resultCount;
 255         se->total_hits += cl->hits;
 256         if (r->presentStatus && !*r->presentStatus && r->records)
 257         {
 258             yaz_log(YLOG_DEBUG, "Records in search response");
 259             cl->records += *r->numberOfRecordsReturned;
 260             ingest_records(cl, r->records);
 261         }
 262         cl->state = Client_Idle;
 263     }
 264     else
 265     {          /*"FAILED"*/
 266         cl->hits = 0;
 267         cl->state = Client_Error;
 268         if (r->records) {
 269             Z_Records *recs = r->records;
 270             if (recs->which == Z_Records_NSD)
 271             {
 272                 yaz_log(YLOG_WARN, "Non-surrogate diagnostic");
 273                 cl->diagnostic = *recs->u.nonSurrogateDiagnostic->condition;
 274                 cl->state = Client_Error;
 275             }
 276         }
 277     }
 278 }
 279
 280 char *normalize_mergekey(char *buf)
 281 {
 282     char *p = buf, *pout = buf;
 283
 284     while (*p)
 285     {
 286         while (*p && !isalnum(*p))
 287             p++;
 288         while (isalnum(*p))
 289             *(pout++) = tolower(*(p++));
 290         if (*p)
 291             *(pout++) = ' ';
 292         while (*p && !isalnum(*p))
 293             p++;
 294     }
 295     if (buf != pout)
 296         *pout = '\0';
 297
 298     return buf;
 299 }
 300
 301
 302 #ifdef GAGA
 303 // FIXME needs to be generalized. Should flexibly generate X lists per search
 304 static void extract_subject(struct session *s, const char *rec)
 305 {
 306     const char *field, *subfield;
 307
 308     while ((field = find_field(rec, "650")))
 309     {
 310         rec = field;
 311         if ((subfield = find_subfield(field, 'a')))
 312         {
 313             char *e, *ef;
 314             char buf[1024];
 315             int len;
 316
 317             ef = index(subfield, '\n');
 318             if (!ef)
 319                 return;
 320             if ((e = index(subfield, '\t')) && e < ef)
 321                 ef = e;
 322             while (ef > subfield && !isalpha(*(ef - 1)) && *(ef - 1) != ')')
 323                 ef--;
 324             len = ef - subfield;
 325             assert(len < 1023);
 326             memcpy(buf, subfield, len);
 327             buf[len] = '\0';
 328 #ifdef FIXME
 329             if (*buf)
 330                 termlist_insert(s->termlist, buf);
 331 #endif
 332         }
 333     }
 334 }
 335 #endif
 336
 337 static void add_facet(struct session *s, const char *type, const char *value)
 338 {
 339     int i;
 340
 341     for (i = 0; i < s->num_termlists; i++)
 342         if (!strcmp(s->termlists[i].name, type))
 343             break;
 344     if (i == s->num_termlists)
 345     {
 346         if (i == SESSION_MAX_TERMLISTS)
 347         {
 348             yaz_log(YLOG_FATAL, "Too many termlists");
 349             exit(1);
 350         }
 351         s->termlists[i].name = nmem_strdup(s->nmem, type);
 352         s->termlists[i].termlist = termlist_create(s->nmem, s->expected_maxrecs, 15);
 353         s->num_termlists = i + 1;
 354     }
 355     termlist_insert(s->termlists[i].termlist, value);
 356 }
 357
 358 static xmlDoc *normalize_record(struct client *cl, Z_External *rec)
 359 {
 360     struct conf_retrievalprofile *rprofile = cl->database->rprofile;
 361     struct conf_retrievalmap *m;
 362     xmlNode *res;
 363     xmlDoc *rdoc;
 364
 365     // First normalize to XML
 366     if (rprofile->native_syntax == Nativesyn_iso2709)
 367     {
 368         char *buf;
 369         int len;
 370         if (rec->which != Z_External_octet)
 371         {
 372             yaz_log(YLOG_WARN, "Unexpected external branch, probably BER");
 373             return 0;
 374         }
 375         buf = (char*) rec->u.octet_aligned->buf;
 376         len = rec->u.octet_aligned->len;
 377         if (yaz_marc_read_iso2709(rprofile->yaz_marc, buf, len) < 0)
 378         {
 379             yaz_log(YLOG_WARN, "Failed to decode MARC");
 380             return 0;
 381         }
 382         if (yaz_marc_write_xml(rprofile->yaz_marc, &res,
 383                     "http://www.loc.gov/MARC21/slim", 0, 0) < 0)
 384         {
 385             yaz_log(YLOG_WARN, "Failed to encode as XML");
 386             return 0;
 387         }
 388         rdoc = xmlNewDoc("1.0");
 389         xmlDocSetRootElement(rdoc, res);
 390     }
 391     else
 392     {
 393         yaz_log(YLOG_FATAL, "Unknown native_syntax in normalize_record");
 394         exit(1);
 395     }
 396     for (m = rprofile->maplist; m; m = m->next)
 397     {
 398         xmlDoc *new;
 399         if (m->type != Map_xslt)
 400         {
 401             yaz_log(YLOG_WARN, "Unknown map type");
 402             return 0;
 403         }
 404         if (!(new = xsltApplyStylesheet(m->stylesheet, rdoc, 0)))
 405         {
 406             yaz_log(YLOG_WARN, "XSLT transformation failed");
 407             return 0;
 408         }
 409         xmlFreeDoc(rdoc);
 410         rdoc = new;
 411     }
 412     if (global_parameters.dump_records)
 413     {
 414         fprintf(stderr, "Record:\n----------------\n");
 415         xmlDocFormatDump(stderr, rdoc, 1);
 416     }
 417     return rdoc;
 418 }
 419
 420 static struct record *ingest_record(struct client *cl, Z_External *rec)
 421 {
 422     xmlDoc *xdoc = normalize_record(cl, rec);
 423     xmlNode *root, *n;
 424     struct record *res, *head;
 425     struct session *se = cl->session;
 426     xmlChar *mergekey, *mergekey_norm;
 427
 428     if (!xdoc)
 429         return 0;
 430
 431     root = xmlDocGetRootElement(xdoc);
 432     if (!(mergekey = xmlGetProp(root, "mergekey")))
 433     {
 434         yaz_log(YLOG_WARN, "No mergekey found in record");
 435         xmlFreeDoc(xdoc);
 436         return 0;
 437     }
 438
 439     res = nmem_malloc(se->nmem, sizeof(struct record));
 440     res->next_cluster = 0;
 441     res->target_offset = -1;
 442     res->term_frequency_vec = 0;
 443     res->title = "Unknown";
 444     res->relevance = 0;
 445
 446     mergekey_norm = nmem_strdup(se->nmem, (char*) mergekey);
 447     xmlFree(mergekey);
 448     res->merge_key = normalize_mergekey(mergekey_norm);
 449
 450     head = reclist_insert(se->reclist, res);
 451     if (!head)
 452     {
 453         /* no room for record */
 454         xmlFreeDoc(xdoc);
 455         return 0;
 456     }
 457     relevance_newrec(se->relevance, head);
 458
 459     for (n = root->children; n; n = n->next)
 460     {
 461         if (n->type != XML_ELEMENT_NODE)
 462             continue;
 463         if (!strcmp(n->name, "facet"))
 464         {
 465             xmlChar *type = xmlGetProp(n, "type");
 466             xmlChar *value = xmlNodeListGetString(xdoc, n->children, 0);
 467             if (type && value)
 468             {
 469                 add_facet(se, type, value);
 470                 relevance_countwords(se->relevance, head, value, 1);
 471             }
 472             xmlFree(type);
 473             xmlFree(value);
 474         }
 475         else if (!strcmp(n->name, "metadata"))
 476         {
 477             xmlChar *type = xmlGetProp(n, "type");
 478             if (type && !strcmp(type, "title"))
 479             {
 480                 xmlChar *value = xmlNodeListGetString(xdoc, n->children, 0);
 481                 if (value)
 482                 {
 483                     res->title = nmem_strdup(se->nmem, value);
 484                     relevance_countwords(se->relevance, head, value, 4);
 485                     xmlFree(value);
 486                 }
 487             }
 488             xmlFree(type);
 489         }
 490         else
 491             yaz_log(YLOG_WARN, "Unexpected element %s in internal record", n->name);
 492     }
 493
 494     xmlFreeDoc(xdoc);
 495
 496     relevance_donerecord(se->relevance, head);
 497     se->total_records++;
 498
 499     return res;
 500 }
 501
 502 static void ingest_records(struct client *cl, Z_Records *r)
 503 {
 504 #if USE_TIMING
 505     yaz_timing_t t = yaz_timing_create();
 506 #endif
 507     struct record *rec;
 508     struct session *s = cl->session;
 509     Z_NamePlusRecordList *rlist;
 510     int i;
 511
 512     if (r->which != Z_Records_DBOSD)
 513         return;
 514     rlist = r->u.databaseOrSurDiagnostics;
 515     for (i = 0; i < rlist->num_records; i++)
 516     {
 517         Z_NamePlusRecord *npr = rlist->records[i];
 518
 519         if (npr->which != Z_NamePlusRecord_databaseRecord)
 520         {
 521             yaz_log(YLOG_WARN, "Unexpected record type, probably diagnostic");
 522             continue;
 523         }
 524
 525         rec = ingest_record(cl, npr->u.databaseRecord);
 526         if (!rec)
 527             continue;
 528     }
 529     if (s->watchlist[SESSION_WATCH_RECORDS].fun && rlist->num_records)
 530         session_alert_watch(s, SESSION_WATCH_RECORDS);
 531
 532 #if USE_TIMING
 533     yaz_timing_stop(t);
 534     yaz_log(YLOG_LOG, "ingest_records %6.5f %3.2f %3.2f",
 535             yaz_timing_get_real(t), yaz_timing_get_user(t),
 536             yaz_timing_get_sys(t));
 537     yaz_timing_destroy(&t);
 538 #endif
 539 }
 540
 541 static void do_presentResponse(IOCHAN i, Z_APDU *a)
 542 {
 543     struct connection *co = iochan_getdata(i);
 544     struct client *cl = co->client;
 545     Z_PresentResponse *r = a->u.presentResponse;
 546
 547     if (r->records) {
 548         Z_Records *recs = r->records;
 549         if (recs->which == Z_Records_NSD)
 550         {
 551             yaz_log(YLOG_WARN, "Non-surrogate diagnostic");
 552             cl->diagnostic = *recs->u.nonSurrogateDiagnostic->condition;
 553             cl->state = Client_Error;
 554         }
 555     }
 556
 557     if (!*r->presentStatus && cl->state != Client_Error)
 558     {
 559         yaz_log(YLOG_DEBUG, "Good Present response");
 560         cl->records += *r->numberOfRecordsReturned;
 561         ingest_records(cl, r->records);
 562         cl->state = Client_Idle;
 563     }
 564     else if (*r->presentStatus)
 565     {
 566         yaz_log(YLOG_WARN, "Bad Present response");
 567         cl->state = Client_Error;
 568     }
 569 }
 570
 571 static void handler(IOCHAN i, int event)
 572 {
 573     struct connection *co = iochan_getdata(i);
 574     struct client *cl = co->client;
 575     struct session *se = 0;
 576
 577     if (cl)
 578         se = cl->session;
 579     else
 580     {
 581         yaz_log(YLOG_WARN, "Destroying orphan connection");
 582         connection_destroy(co);
 583         return;
 584     }
 585
 586     if (co->state == Conn_Connecting && event & EVENT_OUTPUT)
 587     {
 588         int errcode;
 589         socklen_t errlen = sizeof(errcode);
 590
 591         if (getsockopt(cs_fileno(co->link), SOL_SOCKET, SO_ERROR, &errcode,
 592             &errlen) < 0 || errcode != 0)
 593         {
 594             client_fatal(cl);
 595             return;
 596         }
 597         else
 598         {
 599             yaz_log(YLOG_DEBUG, "Connect OK");
 600             co->state = Conn_Open;
 601             if (cl)
 602                 cl->state = Client_Connected;
 603         }
 604     }
 605
 606     else if (event & EVENT_INPUT)
 607     {
 608         int len = cs_get(co->link, &co->ibuf, &co->ibufsize);
 609
 610         if (len < 0)
 611         {
 612             yaz_log(YLOG_WARN|YLOG_ERRNO, "Error reading from Z server");
 613             connection_destroy(co);
 614             return;
 615         }
 616         else if (len == 0)
 617         {
 618             yaz_log(YLOG_WARN, "EOF reading from Z server");
 619             connection_destroy(co);
 620             return;
 621         }
 622         else if (len > 1) // We discard input if we have no connection
 623         {
 624             co->state = Conn_Open;
 625
 626             if (cl && (cl->requestid == se->requestid || cl->state == Client_Initializing))
 627             {
 628                 Z_APDU *a;
 629
 630                 odr_reset(global_parameters.odr_in);
 631                 odr_setbuf(global_parameters.odr_in, co->ibuf, len, 0);
 632                 if (!z_APDU(global_parameters.odr_in, &a, 0, 0))
 633                 {
 634                     client_fatal(cl);
 635                     return;
 636                 }
 637                 switch (a->which)
 638                 {
 639                     case Z_APDU_initResponse:
 640                         do_initResponse(i, a);
 641                         break;
 642                     case Z_APDU_searchResponse:
 643                         do_searchResponse(i, a);
 644                         break;
 645                     case Z_APDU_presentResponse:
 646                         do_presentResponse(i, a);
 647                         break;
 648                     default:
 649                         yaz_log(YLOG_WARN, "Unexpected result from server");
 650                         client_fatal(cl);
 651                         return;
 652                 }
 653                 // We aren't expecting staggered output from target
 654                 // if (cs_more(t->link))
 655                 //    iochan_setevent(i, EVENT_INPUT);
 656             }
 657             else  // we throw away response and go to idle mode
 658             {
 659                 yaz_log(YLOG_DEBUG, "Ignoring result of expired operation");
 660                 cl->state = Client_Idle;
 661             }
 662         }
 663         /* if len==1 we do nothing but wait for more input */
 664     }
 665
 666     if (cl->state == Client_Connected) {
 667         send_init(i);
 668     }
 669
 670     if (cl->state == Client_Idle)
 671     {
 672         if (cl->requestid != se->requestid && *se->query) {
 673             send_search(i);
 674         }
 675         else if (cl->hits > 0 && cl->records < global_parameters.toget &&
 676             cl->records < cl->hits) {
 677             send_present(i);
 678         }
 679     }
 680 }
 681
 682 // Disassociate connection from client
 683 static void connection_release(struct connection *co)
 684 {
 685     struct client *cl = co->client;
 686
 687     yaz_log(YLOG_DEBUG, "Connection release %s", co->host->hostport);
 688     if (!cl)
 689         return;
 690     cl->connection = 0;
 691     co->client = 0;
 692 }
 693
 694 // Close connection and recycle structure
 695 static void connection_destroy(struct connection *co)
 696 {
 697     struct host *h = co->host;
 698     cs_close(co->link);
 699     iochan_destroy(co->iochan);
 700
 701     yaz_log(YLOG_DEBUG, "Connection destroy %s", co->host->hostport);
 702     if (h->connections == co)
 703         h->connections = co->next;
 704     else
 705     {
 706         struct connection *pco;
 707         for (pco = h->connections; pco && pco->next != co; pco = pco->next)
 708             ;
 709         if (pco)
 710             pco->next = co->next;
 711         else
 712             abort();
 713     }
 714     if (co->client)
 715     {
 716         if (co->client->state != Client_Idle)
 717             co->client->state = Client_Disconnected;
 718         co->client->connection = 0;
 719     }
 720     co->next = connection_freelist;
 721     connection_freelist = co;
 722 }
 723
 724 // Creates a new connection for client, associated with the host of
 725 // client's database
 726 static struct connection *connection_create(struct client *cl)
 727 {
 728     struct connection *new;
 729     COMSTACK link;
 730     int res;
 731     void *addr;
 732
 733     yaz_log(YLOG_DEBUG, "Connection create %s", cl->database->url);
 734     if (!(link = cs_create(tcpip_type, 0, PROTO_Z3950)))
 735     {
 736         yaz_log(YLOG_FATAL|YLOG_ERRNO, "Failed to create comstack");
 737         exit(1);
 738     }
 739
 740     if (!(addr = cs_straddr(link, cl->database->host->ipport)))
 741     {
 742         yaz_log(YLOG_WARN|YLOG_ERRNO, "Lookup of IP address failed?");
 743         return 0;
 744     }
 745
 746     res = cs_connect(link, addr);
 747     if (res < 0)
 748     {
 749         yaz_log(YLOG_WARN|YLOG_ERRNO, "cs_connect %s", cl->database->url);
 750         return 0;
 751     }
 752
 753     if ((new = connection_freelist))
 754         connection_freelist = new->next;
 755     else
 756     {
 757         new = xmalloc(sizeof (struct connection));
 758         new->ibuf = 0;
 759         new->ibufsize = 0;
 760     }
 761     new->state = Conn_Connecting;
 762     new->host = cl->database->host;
 763     new->next = new->host->connections;
 764     new->host->connections = new;
 765     new->client = cl;
 766     cl->connection = new;
 767     new->link = link;
 768
 769     new->iochan = iochan_create(cs_fileno(link), handler, 0);
 770     iochan_setdata(new->iochan, new);
 771     new->iochan->next = channel_list;
 772     channel_list = new->iochan;
 773     return new;
 774 }
 775
 776 // Close connection and set state to error
 777 static void client_fatal(struct client *cl)
 778 {
 779     yaz_log(YLOG_WARN, "Fatal error from %s", cl->database->url);
 780     connection_destroy(cl->connection);
 781     cl->state = Client_Error;
 782 }
 783
 784 // Ensure that client has a connection associated
 785 static int client_prep_connection(struct client *cl)
 786 {
 787     struct connection *co;
 788     struct session *se = cl->session;
 789     struct host *host = cl->database->host;
 790
 791     co = cl->connection;
 792
 793     yaz_log(YLOG_DEBUG, "Client prep %s", cl->database->url);
 794
 795     if (!co)
 796     {
 797         // See if someone else has an idle connection
 798         // We should look at timestamps here to select the longest-idle connection
 799         for (co = host->connections; co; co = co->next)
 800             if (co->state == Conn_Open && (!co->client || co->client->session != se))
 801                 break;
 802         if (co)
 803         {
 804             connection_release(co);
 805             cl->connection = co;
 806             co->client = cl;
 807         }
 808         else
 809             co = connection_create(cl);
 810     }
 811     if (co)
 812     {
 813         if (co->state == Conn_Connecting)
 814         {
 815             cl->state = Client_Connecting;
 816             iochan_setflag(co->iochan, EVENT_OUTPUT);
 817         }
 818         else if (co->state == Conn_Open)
 819         {
 820             if (cl->state == Client_Error || cl->state == Client_Disconnected)
 821                 cl->state = Client_Idle;
 822             iochan_setflag(co->iochan, EVENT_OUTPUT);
 823         }
 824         return 1;
 825     }
 826     else
 827         return 0;
 828 }
 829
 830 // This function will most likely vanish when a proper target profile mechanism is
 831 // introduced.
 832 void load_simpletargets(const char *fn)
 833 {
 834     FILE *f = fopen(fn, "r");
 835     char line[256];
 836
 837     if (!f)
 838     {
 839         yaz_log(YLOG_WARN|YLOG_ERRNO, "open %s", fn);
 840         exit(1);
 841     }
 842
 843     while (fgets(line, 255, f))
 844     {
 845         char *url, *db;
 846         struct host *host;
 847         struct database *database;
 848
 849         if (strncmp(line, "target ", 7))
 850             continue;
 851         url = line + 7;
 852         url[strlen(url) - 1] = '\0';
 853         yaz_log(YLOG_DEBUG, "Target: %s", url);
 854         if ((db = strchr(url, '/')))
 855             *(db++) = '\0';
 856         else
 857             db = "Default";
 858
 859         for (host = hosts; host; host = host->next)
 860             if (!strcmp(url, host->hostport))
 861                 break;
 862         if (!host)
 863         {
 864             struct addrinfo *addrinfo, hints;
 865             char *port;
 866             char ipport[128];
 867             unsigned char addrbuf[4];
 868             int res;
 869
 870             host = xmalloc(sizeof(struct host));
 871             host->hostport = xstrdup(url);
 872             host->connections = 0;
 873
 874             if ((port = strchr(url, ':')))
 875                 *(port++) = '\0';
 876             else
 877                 port = "210";
 878
 879             hints.ai_flags = 0;
 880             hints.ai_family = PF_INET;
 881             hints.ai_socktype = SOCK_STREAM;
 882             hints.ai_protocol = IPPROTO_TCP;
 883             hints.ai_addrlen = 0;
 884             hints.ai_addr = 0;
 885             hints.ai_canonname = 0;
 886             hints.ai_next = 0;
 887             // This is not robust code. It assumes that getaddrinfo returns AF_INET
 888             // address.
 889             if ((res = getaddrinfo(url, port, &hints, &addrinfo)))
 890             {
 891                 yaz_log(YLOG_WARN, "Failed to resolve %s: %s", url, gai_strerror(res));
 892                 xfree(host->hostport);
 893                 xfree(host);
 894                 continue;
 895             }
 896             assert(addrinfo->ai_family == PF_INET);
 897             memcpy(addrbuf, &((struct sockaddr_in*)addrinfo->ai_addr)->sin_addr.s_addr, 4);
 898             sprintf(ipport, "%hhd.%hhd.%hhd.%hhd:%s",
 899                     addrbuf[0], addrbuf[1], addrbuf[2], addrbuf[3], port);
 900             host->ipport = xstrdup(ipport);
 901             freeaddrinfo(addrinfo);
 902             host->next = hosts;
 903             hosts = host;
 904         }
 905         database = xmalloc(sizeof(struct database));
 906         database->host = host;
 907         database->url = xmalloc(strlen(url) + strlen(db) + 2);
 908         strcpy(database->url, url);
 909         strcat(database->url, "/");
 910         strcat(database->url, db);
 911
 912         database->databases = xmalloc(2 * sizeof(char *));
 913         database->databases[0] = xstrdup(db);
 914         database->databases[1] = 0;
 915         database->errors = 0;
 916         database->qprofile = 0;
 917         database->rprofile = database_retrieval_profile(database);
 918         database->next = databases;
 919         databases = database;
 920
 921     }
 922     fclose(f);
 923 }
 924
 925 static void pull_terms(NMEM nmem, struct ccl_rpn_node *n, char **termlist, int *num)
 926 {
 927     switch (n->kind)
 928     {
 929         case CCL_RPN_AND:
 930         case CCL_RPN_OR:
 931         case CCL_RPN_NOT:
 932         case CCL_RPN_PROX:
 933             pull_terms(nmem, n->u.p[0], termlist, num);
 934             pull_terms(nmem, n->u.p[1], termlist, num);
 935             break;
 936         case CCL_RPN_TERM:
 937             termlist[(*num)++] = nmem_strdup(nmem, n->u.t.term);
 938             break;
 939         default: // NOOP
 940             break;
 941     }
 942 }
 943
 944 // Extract terms from query into null-terminated termlist
 945 static int extract_terms(NMEM nmem, char *query, char **termlist)
 946 {
 947     int error, pos;
 948     struct ccl_rpn_node *n;
 949     int num = 0;
 950
 951     n = ccl_find_str(global_parameters.ccl_filter, query, &error, &pos);
 952     if (!n)
 953         return -1;
 954     pull_terms(nmem, n, termlist, &num);
 955     termlist[num] = 0;
 956     ccl_rpn_delete(n);
 957     return 0;
 958 }
 959
 960 static struct client *client_create(void)
 961 {
 962     struct client *r;
 963     if (client_freelist)
 964     {
 965         r = client_freelist;
 966         client_freelist = client_freelist->next;
 967     }
 968     else
 969         r = xmalloc(sizeof(struct client));
 970     r->database = 0;
 971     r->connection = 0;
 972     r->session = 0;
 973     r->hits = 0;
 974     r->records = 0;
 975     r->setno = 0;
 976     r->requestid = -1;
 977     r->diagnostic = 0;
 978     r->state = Client_Disconnected;
 979     r->next = 0;
 980     return r;
 981 }
 982
 983 void client_destroy(struct client *c)
 984 {
 985     struct session *se = c->session;
 986     if (c == se->clients)
 987         se->clients = c->next;
 988     else
 989     {
 990         struct client *cc;
 991         for (cc = se->clients; cc && cc->next != c; cc = cc->next)
 992             ;
 993         if (cc)
 994             cc->next = c->next;
 995     }
 996     if (c->connection)
 997         connection_release(c->connection);
 998     c->next = client_freelist;
 999     client_freelist = c;
1000 }
1001
1002 void session_set_watch(struct session *s, int what, session_watchfun fun, void *data)
1003 {
1004     s->watchlist[what].fun = fun;
1005     s->watchlist[what].data = data;
1006 }
1007
1008 void session_alert_watch(struct session *s, int what)
1009 {
1010     if (!s->watchlist[what].fun)
1011         return;
1012     (*s->watchlist[what].fun)(s->watchlist[what].data);
1013     s->watchlist[what].fun = 0;
1014     s->watchlist[what].data = 0;
1015 }
1016
1017 // This needs to be extended with selection criteria
1018 static struct conf_retrievalprofile *database_retrieval_profile(struct database *db)
1019 {
1020     if (!config)
1021     {
1022         yaz_log(YLOG_FATAL, "Must load configuration (-f)");
1023         exit(1);
1024     }
1025     if (!config->retrievalprofiles)
1026     {
1027         yaz_log(YLOG_FATAL, "No retrieval profiles defined");
1028     }
1029     return config->retrievalprofiles;
1030 }
1031
1032 // This should be extended with parameters to control selection criteria
1033 // Associates a set of clients with a session;
1034 int select_targets(struct session *se)
1035 {
1036     struct database *db;
1037     int c = 0;
1038
1039     while (se->clients)
1040         client_destroy(se->clients);
1041     for (db = databases; db; db = db->next)
1042     {
1043         struct client *cl = client_create();
1044         cl->database = db;
1045         cl->session = se;
1046         cl->next = se->clients;
1047         se->clients = cl;
1048         c++;
1049     }
1050     return c;
1051 }
1052
1053 int session_active_clients(struct session *s)
1054 {
1055     struct client *c;
1056     int res = 0;
1057
1058     for (c = s->clients; c; c = c->next)
1059         if (c->connection && (c->state == Client_Connecting ||
1060                     c->state == Client_Initializing ||
1061                     c->state == Client_Searching ||
1062                     c->state == Client_Presenting))
1063             res++;
1064
1065     return res;
1066 }
1067
1068 char *search(struct session *se, char *query)
1069 {
1070     int live_channels = 0;
1071     struct client *cl;
1072
1073     yaz_log(YLOG_DEBUG, "Search");
1074
1075     strcpy(se->query, query);
1076     se->requestid++;
1077     nmem_reset(se->nmem);
1078     for (cl = se->clients; cl; cl = cl->next)
1079     {
1080         cl->hits = -1;
1081         cl->records = 0;
1082         cl->diagnostic = 0;
1083
1084         if (client_prep_connection(cl))
1085             live_channels++;
1086     }
1087     if (live_channels)
1088     {
1089         char *p[512];
1090         int maxrecs = live_channels * global_parameters.toget;
1091         se->num_termlists = 0;
1092         se->reclist = reclist_create(se->nmem, maxrecs);
1093         extract_terms(se->nmem, query, p);
1094         se->relevance = relevance_create(se->nmem, (const char **) p, maxrecs);
1095         se->total_records = se->total_hits = 0;
1096         se->expected_maxrecs = maxrecs;
1097     }
1098     else
1099         return "NOTARGETS";
1100
1101     return 0;
1102 }
1103
1104 void destroy_session(struct session *s)
1105 {
1106     yaz_log(YLOG_LOG, "Destroying session");
1107     while (s->clients)
1108         client_destroy(s->clients);
1109     nmem_destroy(s->nmem);
1110     wrbuf_free(s->wrbuf, 1);
1111 }
1112
1113 struct session *new_session()
1114 {
1115     int i;
1116     struct session *session = xmalloc(sizeof(*session));
1117
1118     yaz_log(YLOG_DEBUG, "New pazpar2 session");
1119
1120     session->total_hits = 0;
1121     session->total_records = 0;
1122     session->num_termlists = 0;
1123     session->reclist = 0;
1124     session->requestid = -1;
1125     session->clients = 0;
1126     session->expected_maxrecs = 0;
1127     session->query[0] = '\0';
1128     session->nmem = nmem_create();
1129     session->wrbuf = wrbuf_alloc();
1130     for (i = 0; i <= SESSION_WATCH_MAX; i++)
1131     {
1132         session->watchlist[i].data = 0;
1133         session->watchlist[i].fun = 0;
1134     }
1135
1136     select_targets(session);
1137
1138     return session;
1139 }
1140
1141 struct hitsbytarget *hitsbytarget(struct session *se, int *count)
1142 {
1143     static struct hitsbytarget res[1000]; // FIXME MM
1144     struct client *cl;
1145
1146     *count = 0;
1147     for (cl = se->clients; cl; cl = cl->next)
1148     {
1149         strcpy(res[*count].id, cl->database->host->hostport);
1150         res[*count].hits = cl->hits;
1151         res[*count].records = cl->records;
1152         res[*count].diagnostic = cl->diagnostic;
1153         res[*count].state = client_states[cl->state];
1154         res[*count].connected  = cl->connection ? 1 : 0;
1155         (*count)++;
1156     }
1157
1158     return res;
1159 }
1160
1161 struct termlist_score **termlist(struct session *s, const char *name, int *num)
1162 {
1163     int i;
1164
1165     for (i = 0; i < s->num_termlists; i++)
1166         if (!strcmp(s->termlists[i].name, name))
1167             return termlist_highscore(s->termlists[i].termlist, num);
1168     return 0;
1169 }
1170
1171 #ifdef MISSING_HEADERS
1172 void report_nmem_stats(void)
1173 {
1174     size_t in_use, is_free;
1175
1176     nmem_get_memory_in_use(&in_use);
1177     nmem_get_memory_free(&is_free);
1178
1179     yaz_log(YLOG_LOG, "nmem stat: use=%ld free=%ld",
1180             (long) in_use, (long) is_free);
1181 }
1182 #endif
1183
1184 struct record **show(struct session *s, int start, int *num, int *total,
1185                      int *sumhits, NMEM nmem_show)
1186 {
1187     struct record **recs = nmem_malloc(nmem_show, *num
1188                                        * sizeof(struct record *));
1189     int i;
1190 #if USE_TIMING
1191     yaz_timing_t t = yaz_timing_create();
1192 #endif
1193     relevance_prepare_read(s->relevance, s->reclist);
1194
1195     *total = s->reclist->num_records;
1196     *sumhits = s->total_hits;
1197
1198     for (i = 0; i < start; i++)
1199         if (!reclist_read_record(s->reclist))
1200         {
1201             *num = 0;
1202             recs = 0;
1203             break;
1204         }
1205
1206     for (i = 0; i < *num; i++)
1207     {
1208         struct record *r = reclist_read_record(s->reclist);
1209         if (!r)
1210         {
1211             *num = i;
1212             break;
1213         }
1214         recs[i] = r;
1215     }
1216 #if USE_TIMING
1217     yaz_timing_stop(t);
1218     yaz_log(YLOG_LOG, "show %6.5f %3.2f %3.2f",
1219             yaz_timing_get_real(t), yaz_timing_get_user(t),
1220             yaz_timing_get_sys(t));
1221     yaz_timing_destroy(&t);
1222 #endif
1223     return recs;
1224 }
1225
1226 void statistics(struct session *se, struct statistics *stat)
1227 {
1228     struct client *cl;
1229     int count = 0;
1230
1231     bzero(stat, sizeof(*stat));
1232     for (cl = se->clients; cl; cl = cl->next)
1233     {
1234         if (!cl->connection)
1235             stat->num_no_connection++;
1236         switch (cl->state)
1237         {
1238             case Client_Connecting: stat->num_connecting++; break;
1239             case Client_Initializing: stat->num_initializing++; break;
1240             case Client_Searching: stat->num_searching++; break;
1241             case Client_Presenting: stat->num_presenting++; break;
1242             case Client_Idle: stat->num_idle++; break;
1243             case Client_Failed: stat->num_failed++; break;
1244             case Client_Error: stat->num_error++; break;
1245             default: break;
1246         }
1247         count++;
1248     }
1249     stat->num_hits = se->total_hits;
1250     stat->num_records = se->total_records;
1251
1252     stat->num_clients = count;
1253 }
1254
1255 static CCL_bibset load_cclfile(const char *fn)
1256 {
1257     CCL_bibset res = ccl_qual_mk();
1258     if (ccl_qual_fname(res, fn) < 0)
1259     {
1260         yaz_log(YLOG_FATAL|YLOG_ERRNO, "%s", fn);
1261         exit(1);
1262     }
1263     return res;
1264 }
1265
1266 int main(int argc, char **argv)
1267 {
1268     int ret;
1269     char *arg;
1270     int setport = 0;
1271
1272     if (signal(SIGPIPE, SIG_IGN) < 0)
1273         yaz_log(YLOG_WARN|YLOG_ERRNO, "signal");
1274
1275     yaz_log_init(YLOG_DEFAULT_LEVEL, "pazpar2", 0);
1276
1277     while ((ret = options("f:x:h:p:C:s:d", argv, argc, &arg)) != -2)
1278     {
1279         switch (ret) {
1280             case 'f':
1281                 if (!read_config(arg))
1282                     exit(1);
1283                 break;
1284             case 'h':
1285                 http_init(arg);
1286                 setport++;
1287                 break;
1288             case 'C':
1289                 global_parameters.ccl_filter = load_cclfile(arg);
1290                 break;
1291             case 'p':
1292                 http_set_proxyaddr(arg);
1293                 break;
1294             case 's':
1295                 load_simpletargets(arg);
1296                 break;
1297             case 'd':
1298                 global_parameters.dump_records = 1;
1299                 break;
1300             default:
1301                 fprintf(stderr, "Usage: pazpar2\n"
1302                         "    -f configfile\n"
1303                         "    -h [host:]port          (REST protocol listener)\n"
1304                         "    -C cclconfig\n"
1305                         "    -s simpletargetfile\n"
1306                         "    -p hostname[:portno]    (HTTP proxy)\n");
1307                 exit(1);
1308         }
1309     }
1310
1311     if (!setport)
1312     {
1313         fprintf(stderr, "Set command port with -h\n");
1314         exit(1);
1315     }
1316
1317     global_parameters.ccl_filter = load_cclfile("../etc/default.bib");
1318     global_parameters.yaz_marc = yaz_marc_create();
1319     yaz_marc_subfield_str(global_parameters.yaz_marc, "\t");
1320     global_parameters.odr_in = odr_createmem(ODR_DECODE);
1321     global_parameters.odr_out = odr_createmem(ODR_ENCODE);
1322
1323     event_loop(&channel_list);
1324
1325     return 0;
1326 }
1327
1328 /*
1329  * Local variables:
1330  * c-basic-offset: 4
1331  * indent-tabs-mode: nil
1332  * End:
1333  * vim: shiftwidth=4 tabstop=8 expandtab
1334  */