src/pazpar2.c

   1 /* $Id: pazpar2.c,v 1.21 2007-01-09 18:06:28 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 // Extract what appears to be years from buf, storing highest and
 428 // lowest values.
 429 static int extract_years(const char *buf, int *first, int *last)
 430 {
 431     *first = -1;
 432     *last = -1;
 433     while (*buf)
 434     {
 435         const char *e;
 436         int len;
 437
 438         while (*buf && !isdigit(*buf))
 439             buf++;
 440         len = 0;
 441         for (e = buf; *e && isdigit(*e); e++)
 442             len++;
 443         if (len == 4)
 444         {
 445             int value = atoi(buf);
 446             if (*first < 0 || value < *first)
 447                 *first = value;
 448             if (*last < 0 || value > *last)
 449                 *last = value;
 450         }
 451         buf = e;
 452     }
 453     return *first;
 454 }
 455
 456 static struct record *ingest_record(struct client *cl, Z_External *rec)
 457 {
 458     xmlDoc *xdoc = normalize_record(cl, rec);
 459     xmlNode *root, *n;
 460     struct record *res;
 461     struct record_cluster *cluster;
 462     struct session *se = cl->session;
 463     xmlChar *mergekey, *mergekey_norm;
 464     xmlChar *type;
 465     xmlChar *value;
 466     struct conf_service *service = global_parameters.server->service;
 467
 468     if (!xdoc)
 469         return 0;
 470
 471     root = xmlDocGetRootElement(xdoc);
 472     if (!(mergekey = xmlGetProp(root, "mergekey")))
 473     {
 474         yaz_log(YLOG_WARN, "No mergekey found in record");
 475         xmlFreeDoc(xdoc);
 476         return 0;
 477     }
 478
 479     res = nmem_malloc(se->nmem, sizeof(struct record));
 480     res->next = 0;
 481     res->target_offset = -1;
 482     res->term_frequency_vec = 0;
 483     res->metadata = nmem_malloc(se->nmem,
 484             sizeof(struct record_metadata*) * service->num_metadata);
 485     bzero(res->metadata, sizeof(struct record_metadata*) * service->num_metadata);
 486     res->relevance = 0;
 487
 488     mergekey_norm = nmem_strdup(se->nmem, (char*) mergekey);
 489     xmlFree(mergekey);
 490     normalize_mergekey(mergekey_norm);
 491
 492     cluster = reclist_insert(se->reclist, res, mergekey_norm);
 493     if (!cluster)
 494     {
 495         /* no room for record */
 496         xmlFreeDoc(xdoc);
 497         return 0;
 498     }
 499     relevance_newrec(se->relevance, cluster);
 500
 501     type = value = 0;
 502     for (n = root->children; n; n = n->next)
 503     {
 504         if (type)
 505             xmlFree(type);
 506         if (value)
 507             xmlFree(value);
 508         type = value = 0;
 509
 510         if (n->type != XML_ELEMENT_NODE)
 511             continue;
 512         if (!strcmp(n->name, "metadata"))
 513         {
 514             type = xmlGetProp(n, "type");
 515             value = xmlNodeListGetString(xdoc, n->children, 0);
 516             struct conf_metadata *md = 0;
 517             struct record_metadata **wheretoput, *newm;
 518             int imeta;
 519             int first, last;
 520
 521             // First, find out what field we're looking at
 522             for (imeta = 0; imeta < service->num_metadata; imeta++)
 523                 if (!strcmp(type, service->metadata[imeta].name))
 524                 {
 525                     md = &service->metadata[imeta];
 526                     break;
 527                 }
 528             if (!md)
 529             {
 530                 yaz_log(YLOG_WARN, "Ignoring unknown metadata element: %s", type);
 531                 continue;
 532             }
 533
 534             // Find out where we are putting it
 535             if (md->merge == Metadata_merge_no)
 536                 wheretoput = &res->metadata[imeta];
 537             else
 538                 wheretoput = &cluster->metadata[imeta];
 539
 540             // Put it there
 541             newm = nmem_malloc(se->nmem, sizeof(struct record_metadata));
 542             newm->next = 0;
 543             if (md->type == Metadata_type_generic)
 544             {
 545                 newm->data.text = nmem_strdup(se->nmem, value);
 546             }
 547             else if (md->type == Metadata_type_year)
 548             {
 549                 if (extract_years(value, &first, &last) < 0)
 550                     continue;
 551             }
 552             else
 553             {
 554                 yaz_log(YLOG_WARN, "Unknown type in metadata element %s", type);
 555                 continue;
 556             }
 557             if (md->type == Metadata_type_year && md->merge != Metadata_merge_range)
 558             {
 559                 yaz_log(YLOG_WARN, "Only range merging supported for years");
 560                 continue;
 561             }
 562             if (md->merge == Metadata_merge_unique)
 563             {
 564                 struct record_metadata *mnode;
 565                 for (mnode = *wheretoput; mnode; mnode = mnode->next)
 566                     if (!strcmp(mnode->data.text, mnode->data.text))
 567                         break;
 568                 if (!mnode)
 569                 {
 570                     newm->next = *wheretoput;
 571                     *wheretoput = newm;
 572                 }
 573             }
 574             else if (md->merge == Metadata_merge_longest)
 575             {
 576                 if (!*wheretoput ||
 577                         strlen(newm->data.text) > strlen((*wheretoput)->data.text))
 578                 *wheretoput = newm;
 579             }
 580             else if (md->merge == Metadata_merge_all || md->merge == Metadata_merge_no)
 581             {
 582                 newm->next = *wheretoput;
 583                 *wheretoput = newm;
 584             }
 585             else if (md->merge == Metadata_merge_range)
 586             {
 587                 assert(md->type == Metadata_type_year);
 588                 if (!*wheretoput)
 589                 {
 590                     *wheretoput = newm;
 591                     (*wheretoput)->data.year.year1 = first;
 592                     (*wheretoput)->data.year.year2 = last;
 593                 }
 594                 else
 595                 {
 596                     if (first < (*wheretoput)->data.year.year1)
 597                         (*wheretoput)->data.year.year1 = first;
 598                     if (last > (*wheretoput)->data.year.year2)
 599                         (*wheretoput)->data.year.year2 = last;
 600                 }
 601             }
 602             else
 603                 yaz_log(YLOG_WARN, "Don't know how to merge on element name %s", md->name);
 604
 605             if (md->rank)
 606                 relevance_countwords(se->relevance, cluster, value, md->rank);
 607             if (md->termlist)
 608                 add_facet(se, type, value);
 609             xmlFree(type);
 610             xmlFree(value);
 611             type = value = 0;
 612         }
 613         else
 614             yaz_log(YLOG_WARN, "Unexpected element %s in internal record", n->name);
 615     }
 616
 617     xmlFreeDoc(xdoc);
 618
 619     relevance_donerecord(se->relevance, cluster);
 620     se->total_records++;
 621
 622     return res;
 623 }
 624
 625 static void ingest_records(struct client *cl, Z_Records *r)
 626 {
 627 #if USE_TIMING
 628     yaz_timing_t t = yaz_timing_create();
 629 #endif
 630     struct record *rec;
 631     struct session *s = cl->session;
 632     Z_NamePlusRecordList *rlist;
 633     int i;
 634
 635     if (r->which != Z_Records_DBOSD)
 636         return;
 637     rlist = r->u.databaseOrSurDiagnostics;
 638     for (i = 0; i < rlist->num_records; i++)
 639     {
 640         Z_NamePlusRecord *npr = rlist->records[i];
 641
 642         if (npr->which != Z_NamePlusRecord_databaseRecord)
 643         {
 644             yaz_log(YLOG_WARN, "Unexpected record type, probably diagnostic");
 645             continue;
 646         }
 647
 648         rec = ingest_record(cl, npr->u.databaseRecord);
 649         if (!rec)
 650             continue;
 651     }
 652     if (s->watchlist[SESSION_WATCH_RECORDS].fun && rlist->num_records)
 653         session_alert_watch(s, SESSION_WATCH_RECORDS);
 654
 655 #if USE_TIMING
 656     yaz_timing_stop(t);
 657     yaz_log(YLOG_LOG, "ingest_records %6.5f %3.2f %3.2f",
 658             yaz_timing_get_real(t), yaz_timing_get_user(t),
 659             yaz_timing_get_sys(t));
 660     yaz_timing_destroy(&t);
 661 #endif
 662 }
 663
 664 static void do_presentResponse(IOCHAN i, Z_APDU *a)
 665 {
 666     struct connection *co = iochan_getdata(i);
 667     struct client *cl = co->client;
 668     Z_PresentResponse *r = a->u.presentResponse;
 669
 670     if (r->records) {
 671         Z_Records *recs = r->records;
 672         if (recs->which == Z_Records_NSD)
 673         {
 674             yaz_log(YLOG_WARN, "Non-surrogate diagnostic");
 675             cl->diagnostic = *recs->u.nonSurrogateDiagnostic->condition;
 676             cl->state = Client_Error;
 677         }
 678     }
 679
 680     if (!*r->presentStatus && cl->state != Client_Error)
 681     {
 682         yaz_log(YLOG_DEBUG, "Good Present response");
 683         cl->records += *r->numberOfRecordsReturned;
 684         ingest_records(cl, r->records);
 685         cl->state = Client_Idle;
 686     }
 687     else if (*r->presentStatus)
 688     {
 689         yaz_log(YLOG_WARN, "Bad Present response");
 690         cl->state = Client_Error;
 691     }
 692 }
 693
 694 static void handler(IOCHAN i, int event)
 695 {
 696     struct connection *co = iochan_getdata(i);
 697     struct client *cl = co->client;
 698     struct session *se = 0;
 699
 700     if (cl)
 701         se = cl->session;
 702     else
 703     {
 704         yaz_log(YLOG_WARN, "Destroying orphan connection");
 705         connection_destroy(co);
 706         return;
 707     }
 708
 709     if (co->state == Conn_Connecting && event & EVENT_OUTPUT)
 710     {
 711         int errcode;
 712         socklen_t errlen = sizeof(errcode);
 713
 714         if (getsockopt(cs_fileno(co->link), SOL_SOCKET, SO_ERROR, &errcode,
 715             &errlen) < 0 || errcode != 0)
 716         {
 717             client_fatal(cl);
 718             return;
 719         }
 720         else
 721         {
 722             yaz_log(YLOG_DEBUG, "Connect OK");
 723             co->state = Conn_Open;
 724             if (cl)
 725                 cl->state = Client_Connected;
 726         }
 727     }
 728
 729     else if (event & EVENT_INPUT)
 730     {
 731         int len = cs_get(co->link, &co->ibuf, &co->ibufsize);
 732
 733         if (len < 0)
 734         {
 735             yaz_log(YLOG_WARN|YLOG_ERRNO, "Error reading from Z server");
 736             connection_destroy(co);
 737             return;
 738         }
 739         else if (len == 0)
 740         {
 741             yaz_log(YLOG_WARN, "EOF reading from Z server");
 742             connection_destroy(co);
 743             return;
 744         }
 745         else if (len > 1) // We discard input if we have no connection
 746         {
 747             co->state = Conn_Open;
 748
 749             if (cl && (cl->requestid == se->requestid || cl->state == Client_Initializing))
 750             {
 751                 Z_APDU *a;
 752
 753                 odr_reset(global_parameters.odr_in);
 754                 odr_setbuf(global_parameters.odr_in, co->ibuf, len, 0);
 755                 if (!z_APDU(global_parameters.odr_in, &a, 0, 0))
 756                 {
 757                     client_fatal(cl);
 758                     return;
 759                 }
 760                 switch (a->which)
 761                 {
 762                     case Z_APDU_initResponse:
 763                         do_initResponse(i, a);
 764                         break;
 765                     case Z_APDU_searchResponse:
 766                         do_searchResponse(i, a);
 767                         break;
 768                     case Z_APDU_presentResponse:
 769                         do_presentResponse(i, a);
 770                         break;
 771                     default:
 772                         yaz_log(YLOG_WARN, "Unexpected result from server");
 773                         client_fatal(cl);
 774                         return;
 775                 }
 776                 // We aren't expecting staggered output from target
 777                 // if (cs_more(t->link))
 778                 //    iochan_setevent(i, EVENT_INPUT);
 779             }
 780             else  // we throw away response and go to idle mode
 781             {
 782                 yaz_log(YLOG_DEBUG, "Ignoring result of expired operation");
 783                 cl->state = Client_Idle;
 784             }
 785         }
 786         /* if len==1 we do nothing but wait for more input */
 787     }
 788
 789     if (cl->state == Client_Connected) {
 790         send_init(i);
 791     }
 792
 793     if (cl->state == Client_Idle)
 794     {
 795         if (cl->requestid != se->requestid && *se->query) {
 796             send_search(i);
 797         }
 798         else if (cl->hits > 0 && cl->records < global_parameters.toget &&
 799             cl->records < cl->hits) {
 800             send_present(i);
 801         }
 802     }
 803 }
 804
 805 // Disassociate connection from client
 806 static void connection_release(struct connection *co)
 807 {
 808     struct client *cl = co->client;
 809
 810     yaz_log(YLOG_DEBUG, "Connection release %s", co->host->hostport);
 811     if (!cl)
 812         return;
 813     cl->connection = 0;
 814     co->client = 0;
 815 }
 816
 817 // Close connection and recycle structure
 818 static void connection_destroy(struct connection *co)
 819 {
 820     struct host *h = co->host;
 821     cs_close(co->link);
 822     iochan_destroy(co->iochan);
 823
 824     yaz_log(YLOG_DEBUG, "Connection destroy %s", co->host->hostport);
 825     if (h->connections == co)
 826         h->connections = co->next;
 827     else
 828     {
 829         struct connection *pco;
 830         for (pco = h->connections; pco && pco->next != co; pco = pco->next)
 831             ;
 832         if (pco)
 833             pco->next = co->next;
 834         else
 835             abort();
 836     }
 837     if (co->client)
 838     {
 839         if (co->client->state != Client_Idle)
 840             co->client->state = Client_Disconnected;
 841         co->client->connection = 0;
 842     }
 843     co->next = connection_freelist;
 844     connection_freelist = co;
 845 }
 846
 847 // Creates a new connection for client, associated with the host of
 848 // client's database
 849 static struct connection *connection_create(struct client *cl)
 850 {
 851     struct connection *new;
 852     COMSTACK link;
 853     int res;
 854     void *addr;
 855
 856     yaz_log(YLOG_DEBUG, "Connection create %s", cl->database->url);
 857     if (!(link = cs_create(tcpip_type, 0, PROTO_Z3950)))
 858     {
 859         yaz_log(YLOG_FATAL|YLOG_ERRNO, "Failed to create comstack");
 860         exit(1);
 861     }
 862
 863     if (!(addr = cs_straddr(link, cl->database->host->ipport)))
 864     {
 865         yaz_log(YLOG_WARN|YLOG_ERRNO, "Lookup of IP address failed?");
 866         return 0;
 867     }
 868
 869     res = cs_connect(link, addr);
 870     if (res < 0)
 871     {
 872         yaz_log(YLOG_WARN|YLOG_ERRNO, "cs_connect %s", cl->database->url);
 873         return 0;
 874     }
 875
 876     if ((new = connection_freelist))
 877         connection_freelist = new->next;
 878     else
 879     {
 880         new = xmalloc(sizeof (struct connection));
 881         new->ibuf = 0;
 882         new->ibufsize = 0;
 883     }
 884     new->state = Conn_Connecting;
 885     new->host = cl->database->host;
 886     new->next = new->host->connections;
 887     new->host->connections = new;
 888     new->client = cl;
 889     cl->connection = new;
 890     new->link = link;
 891
 892     new->iochan = iochan_create(cs_fileno(link), handler, 0);
 893     iochan_setdata(new->iochan, new);
 894     new->iochan->next = channel_list;
 895     channel_list = new->iochan;
 896     return new;
 897 }
 898
 899 // Close connection and set state to error
 900 static void client_fatal(struct client *cl)
 901 {
 902     yaz_log(YLOG_WARN, "Fatal error from %s", cl->database->url);
 903     connection_destroy(cl->connection);
 904     cl->state = Client_Error;
 905 }
 906
 907 // Ensure that client has a connection associated
 908 static int client_prep_connection(struct client *cl)
 909 {
 910     struct connection *co;
 911     struct session *se = cl->session;
 912     struct host *host = cl->database->host;
 913
 914     co = cl->connection;
 915
 916     yaz_log(YLOG_DEBUG, "Client prep %s", cl->database->url);
 917
 918     if (!co)
 919     {
 920         // See if someone else has an idle connection
 921         // We should look at timestamps here to select the longest-idle connection
 922         for (co = host->connections; co; co = co->next)
 923             if (co->state == Conn_Open && (!co->client || co->client->session != se))
 924                 break;
 925         if (co)
 926         {
 927             connection_release(co);
 928             cl->connection = co;
 929             co->client = cl;
 930         }
 931         else
 932             co = connection_create(cl);
 933     }
 934     if (co)
 935     {
 936         if (co->state == Conn_Connecting)
 937         {
 938             cl->state = Client_Connecting;
 939             iochan_setflag(co->iochan, EVENT_OUTPUT);
 940         }
 941         else if (co->state == Conn_Open)
 942         {
 943             if (cl->state == Client_Error || cl->state == Client_Disconnected)
 944                 cl->state = Client_Idle;
 945             iochan_setflag(co->iochan, EVENT_OUTPUT);
 946         }
 947         return 1;
 948     }
 949     else
 950         return 0;
 951 }
 952
 953 // This function will most likely vanish when a proper target profile mechanism is
 954 // introduced.
 955 void load_simpletargets(const char *fn)
 956 {
 957     FILE *f = fopen(fn, "r");
 958     char line[256];
 959
 960     if (!f)
 961     {
 962         yaz_log(YLOG_WARN|YLOG_ERRNO, "open %s", fn);
 963         exit(1);
 964     }
 965
 966     while (fgets(line, 255, f))
 967     {
 968         char *url, *db;
 969         struct host *host;
 970         struct database *database;
 971
 972         if (strncmp(line, "target ", 7))
 973             continue;
 974         url = line + 7;
 975         url[strlen(url) - 1] = '\0';
 976         yaz_log(YLOG_DEBUG, "Target: %s", url);
 977         if ((db = strchr(url, '/')))
 978             *(db++) = '\0';
 979         else
 980             db = "Default";
 981
 982         for (host = hosts; host; host = host->next)
 983             if (!strcmp(url, host->hostport))
 984                 break;
 985         if (!host)
 986         {
 987             struct addrinfo *addrinfo, hints;
 988             char *port;
 989             char ipport[128];
 990             unsigned char addrbuf[4];
 991             int res;
 992
 993             host = xmalloc(sizeof(struct host));
 994             host->hostport = xstrdup(url);
 995             host->connections = 0;
 996
 997             if ((port = strchr(url, ':')))
 998                 *(port++) = '\0';
 999             else
1000                 port = "210";
1001
1002             hints.ai_flags = 0;
1003             hints.ai_family = PF_INET;
1004             hints.ai_socktype = SOCK_STREAM;
1005             hints.ai_protocol = IPPROTO_TCP;
1006             hints.ai_addrlen = 0;
1007             hints.ai_addr = 0;
1008             hints.ai_canonname = 0;
1009             hints.ai_next = 0;
1010             // This is not robust code. It assumes that getaddrinfo returns AF_INET
1011             // address.
1012             if ((res = getaddrinfo(url, port, &hints, &addrinfo)))
1013             {
1014                 yaz_log(YLOG_WARN, "Failed to resolve %s: %s", url, gai_strerror(res));
1015                 xfree(host->hostport);
1016                 xfree(host);
1017                 continue;
1018             }
1019             assert(addrinfo->ai_family == PF_INET);
1020             memcpy(addrbuf, &((struct sockaddr_in*)addrinfo->ai_addr)->sin_addr.s_addr, 4);
1021             sprintf(ipport, "%hhd.%hhd.%hhd.%hhd:%s",
1022                     addrbuf[0], addrbuf[1], addrbuf[2], addrbuf[3], port);
1023             host->ipport = xstrdup(ipport);
1024             freeaddrinfo(addrinfo);
1025             host->next = hosts;
1026             hosts = host;
1027         }
1028         database = xmalloc(sizeof(struct database));
1029         database->host = host;
1030         database->url = xmalloc(strlen(url) + strlen(db) + 2);
1031         strcpy(database->url, url);
1032         strcat(database->url, "/");
1033         strcat(database->url, db);
1034
1035         database->databases = xmalloc(2 * sizeof(char *));
1036         database->databases[0] = xstrdup(db);
1037         database->databases[1] = 0;
1038         database->errors = 0;
1039         database->qprofile = 0;
1040         database->rprofile = database_retrieval_profile(database);
1041         database->next = databases;
1042         databases = database;
1043
1044     }
1045     fclose(f);
1046 }
1047
1048 static void pull_terms(NMEM nmem, struct ccl_rpn_node *n, char **termlist, int *num)
1049 {
1050     switch (n->kind)
1051     {
1052         case CCL_RPN_AND:
1053         case CCL_RPN_OR:
1054         case CCL_RPN_NOT:
1055         case CCL_RPN_PROX:
1056             pull_terms(nmem, n->u.p[0], termlist, num);
1057             pull_terms(nmem, n->u.p[1], termlist, num);
1058             break;
1059         case CCL_RPN_TERM:
1060             termlist[(*num)++] = nmem_strdup(nmem, n->u.t.term);
1061             break;
1062         default: // NOOP
1063             break;
1064     }
1065 }
1066
1067 // Extract terms from query into null-terminated termlist
1068 static int extract_terms(NMEM nmem, char *query, char **termlist)
1069 {
1070     int error, pos;
1071     struct ccl_rpn_node *n;
1072     int num = 0;
1073
1074     n = ccl_find_str(global_parameters.ccl_filter, query, &error, &pos);
1075     if (!n)
1076         return -1;
1077     pull_terms(nmem, n, termlist, &num);
1078     termlist[num] = 0;
1079     ccl_rpn_delete(n);
1080     return 0;
1081 }
1082
1083 static struct client *client_create(void)
1084 {
1085     struct client *r;
1086     if (client_freelist)
1087     {
1088         r = client_freelist;
1089         client_freelist = client_freelist->next;
1090     }
1091     else
1092         r = xmalloc(sizeof(struct client));
1093     r->database = 0;
1094     r->connection = 0;
1095     r->session = 0;
1096     r->hits = 0;
1097     r->records = 0;
1098     r->setno = 0;
1099     r->requestid = -1;
1100     r->diagnostic = 0;
1101     r->state = Client_Disconnected;
1102     r->next = 0;
1103     return r;
1104 }
1105
1106 void client_destroy(struct client *c)
1107 {
1108     struct session *se = c->session;
1109     if (c == se->clients)
1110         se->clients = c->next;
1111     else
1112     {
1113         struct client *cc;
1114         for (cc = se->clients; cc && cc->next != c; cc = cc->next)
1115             ;
1116         if (cc)
1117             cc->next = c->next;
1118     }
1119     if (c->connection)
1120         connection_release(c->connection);
1121     c->next = client_freelist;
1122     client_freelist = c;
1123 }
1124
1125 void session_set_watch(struct session *s, int what, session_watchfun fun, void *data)
1126 {
1127     s->watchlist[what].fun = fun;
1128     s->watchlist[what].data = data;
1129 }
1130
1131 void session_alert_watch(struct session *s, int what)
1132 {
1133     if (!s->watchlist[what].fun)
1134         return;
1135     (*s->watchlist[what].fun)(s->watchlist[what].data);
1136     s->watchlist[what].fun = 0;
1137     s->watchlist[what].data = 0;
1138 }
1139
1140 // This needs to be extended with selection criteria
1141 static struct conf_retrievalprofile *database_retrieval_profile(struct database *db)
1142 {
1143     if (!config)
1144     {
1145         yaz_log(YLOG_FATAL, "Must load configuration (-f)");
1146         exit(1);
1147     }
1148     if (!config->retrievalprofiles)
1149     {
1150         yaz_log(YLOG_FATAL, "No retrieval profiles defined");
1151     }
1152     return config->retrievalprofiles;
1153 }
1154
1155 // This should be extended with parameters to control selection criteria
1156 // Associates a set of clients with a session;
1157 int select_targets(struct session *se)
1158 {
1159     struct database *db;
1160     int c = 0;
1161
1162     while (se->clients)
1163         client_destroy(se->clients);
1164     for (db = databases; db; db = db->next)
1165     {
1166         struct client *cl = client_create();
1167         cl->database = db;
1168         cl->session = se;
1169         cl->next = se->clients;
1170         se->clients = cl;
1171         c++;
1172     }
1173     return c;
1174 }
1175
1176 int session_active_clients(struct session *s)
1177 {
1178     struct client *c;
1179     int res = 0;
1180
1181     for (c = s->clients; c; c = c->next)
1182         if (c->connection && (c->state == Client_Connecting ||
1183                     c->state == Client_Initializing ||
1184                     c->state == Client_Searching ||
1185                     c->state == Client_Presenting))
1186             res++;
1187
1188     return res;
1189 }
1190
1191 char *search(struct session *se, char *query)
1192 {
1193     int live_channels = 0;
1194     struct client *cl;
1195
1196     yaz_log(YLOG_DEBUG, "Search");
1197
1198     strcpy(se->query, query);
1199     se->requestid++;
1200     nmem_reset(se->nmem);
1201     for (cl = se->clients; cl; cl = cl->next)
1202     {
1203         cl->hits = -1;
1204         cl->records = 0;
1205         cl->diagnostic = 0;
1206
1207         if (client_prep_connection(cl))
1208             live_channels++;
1209     }
1210     if (live_channels)
1211     {
1212         char *p[512];
1213         int maxrecs = live_channels * global_parameters.toget;
1214         se->num_termlists = 0;
1215         se->reclist = reclist_create(se->nmem, maxrecs);
1216         extract_terms(se->nmem, query, p);
1217         se->relevance = relevance_create(se->nmem, (const char **) p, maxrecs);
1218         se->total_records = se->total_hits = 0;
1219         se->expected_maxrecs = maxrecs;
1220     }
1221     else
1222         return "NOTARGETS";
1223
1224     return 0;
1225 }
1226
1227 void destroy_session(struct session *s)
1228 {
1229     yaz_log(YLOG_LOG, "Destroying session");
1230     while (s->clients)
1231         client_destroy(s->clients);
1232     nmem_destroy(s->nmem);
1233     wrbuf_free(s->wrbuf, 1);
1234 }
1235
1236 struct session *new_session()
1237 {
1238     int i;
1239     struct session *session = xmalloc(sizeof(*session));
1240
1241     yaz_log(YLOG_DEBUG, "New pazpar2 session");
1242
1243     session->total_hits = 0;
1244     session->total_records = 0;
1245     session->num_termlists = 0;
1246     session->reclist = 0;
1247     session->requestid = -1;
1248     session->clients = 0;
1249     session->expected_maxrecs = 0;
1250     session->query[0] = '\0';
1251     session->nmem = nmem_create();
1252     session->wrbuf = wrbuf_alloc();
1253     for (i = 0; i <= SESSION_WATCH_MAX; i++)
1254     {
1255         session->watchlist[i].data = 0;
1256         session->watchlist[i].fun = 0;
1257     }
1258
1259     select_targets(session);
1260
1261     return session;
1262 }
1263
1264 struct hitsbytarget *hitsbytarget(struct session *se, int *count)
1265 {
1266     static struct hitsbytarget res[1000]; // FIXME MM
1267     struct client *cl;
1268
1269     *count = 0;
1270     for (cl = se->clients; cl; cl = cl->next)
1271     {
1272         strcpy(res[*count].id, cl->database->host->hostport);
1273         res[*count].hits = cl->hits;
1274         res[*count].records = cl->records;
1275         res[*count].diagnostic = cl->diagnostic;
1276         res[*count].state = client_states[cl->state];
1277         res[*count].connected  = cl->connection ? 1 : 0;
1278         (*count)++;
1279     }
1280
1281     return res;
1282 }
1283
1284 struct termlist_score **termlist(struct session *s, const char *name, int *num)
1285 {
1286     int i;
1287
1288     for (i = 0; i < s->num_termlists; i++)
1289         if (!strcmp(s->termlists[i].name, name))
1290             return termlist_highscore(s->termlists[i].termlist, num);
1291     return 0;
1292 }
1293
1294 #ifdef MISSING_HEADERS
1295 void report_nmem_stats(void)
1296 {
1297     size_t in_use, is_free;
1298
1299     nmem_get_memory_in_use(&in_use);
1300     nmem_get_memory_free(&is_free);
1301
1302     yaz_log(YLOG_LOG, "nmem stat: use=%ld free=%ld",
1303             (long) in_use, (long) is_free);
1304 }
1305 #endif
1306
1307 struct record_cluster **show(struct session *s, int start, int *num, int *total,
1308                      int *sumhits, NMEM nmem_show)
1309 {
1310     struct record_cluster **recs = nmem_malloc(nmem_show, *num
1311                                        * sizeof(struct record_cluster *));
1312     int i;
1313 #if USE_TIMING
1314     yaz_timing_t t = yaz_timing_create();
1315 #endif
1316     relevance_prepare_read(s->relevance, s->reclist);
1317
1318     *total = s->reclist->num_records;
1319     *sumhits = s->total_hits;
1320
1321     for (i = 0; i < start; i++)
1322         if (!reclist_read_record(s->reclist))
1323         {
1324             *num = 0;
1325             recs = 0;
1326             break;
1327         }
1328
1329     for (i = 0; i < *num; i++)
1330     {
1331         struct record_cluster *r = reclist_read_record(s->reclist);
1332         if (!r)
1333         {
1334             *num = i;
1335             break;
1336         }
1337         recs[i] = r;
1338     }
1339 #if USE_TIMING
1340     yaz_timing_stop(t);
1341     yaz_log(YLOG_LOG, "show %6.5f %3.2f %3.2f",
1342             yaz_timing_get_real(t), yaz_timing_get_user(t),
1343             yaz_timing_get_sys(t));
1344     yaz_timing_destroy(&t);
1345 #endif
1346     return recs;
1347 }
1348
1349 void statistics(struct session *se, struct statistics *stat)
1350 {
1351     struct client *cl;
1352     int count = 0;
1353
1354     bzero(stat, sizeof(*stat));
1355     for (cl = se->clients; cl; cl = cl->next)
1356     {
1357         if (!cl->connection)
1358             stat->num_no_connection++;
1359         switch (cl->state)
1360         {
1361             case Client_Connecting: stat->num_connecting++; break;
1362             case Client_Initializing: stat->num_initializing++; break;
1363             case Client_Searching: stat->num_searching++; break;
1364             case Client_Presenting: stat->num_presenting++; break;
1365             case Client_Idle: stat->num_idle++; break;
1366             case Client_Failed: stat->num_failed++; break;
1367             case Client_Error: stat->num_error++; break;
1368             default: break;
1369         }
1370         count++;
1371     }
1372     stat->num_hits = se->total_hits;
1373     stat->num_records = se->total_records;
1374
1375     stat->num_clients = count;
1376 }
1377
1378 static CCL_bibset load_cclfile(const char *fn)
1379 {
1380     CCL_bibset res = ccl_qual_mk();
1381     if (ccl_qual_fname(res, fn) < 0)
1382     {
1383         yaz_log(YLOG_FATAL|YLOG_ERRNO, "%s", fn);
1384         exit(1);
1385     }
1386     return res;
1387 }
1388
1389 int main(int argc, char **argv)
1390 {
1391     int ret;
1392     char *arg;
1393     int setport = 0;
1394
1395     if (signal(SIGPIPE, SIG_IGN) < 0)
1396         yaz_log(YLOG_WARN|YLOG_ERRNO, "signal");
1397
1398     yaz_log_init(YLOG_DEFAULT_LEVEL, "pazpar2", 0);
1399
1400     while ((ret = options("f:x:h:p:C:s:d", argv, argc, &arg)) != -2)
1401     {
1402         switch (ret) {
1403             case 'f':
1404                 if (!read_config(arg))
1405                     exit(1);
1406                 break;
1407             case 'h':
1408                 http_init(arg);
1409                 setport++;
1410                 break;
1411             case 'C':
1412                 global_parameters.ccl_filter = load_cclfile(arg);
1413                 break;
1414             case 'p':
1415                 http_set_proxyaddr(arg);
1416                 break;
1417             case 's':
1418                 load_simpletargets(arg);
1419                 break;
1420             case 'd':
1421                 global_parameters.dump_records = 1;
1422                 break;
1423             default:
1424                 fprintf(stderr, "Usage: pazpar2\n"
1425                         "    -f configfile\n"
1426                         "    -h [host:]port          (REST protocol listener)\n"
1427                         "    -C cclconfig\n"
1428                         "    -s simpletargetfile\n"
1429                         "    -p hostname[:portno]    (HTTP proxy)\n");
1430                 exit(1);
1431         }
1432     }
1433
1434     if (!config)
1435     {
1436         yaz_log(YLOG_FATAL, "Load config with -f");
1437         exit(1);
1438     }
1439     global_parameters.server = config->servers;
1440
1441     if (!setport)
1442     {
1443         fprintf(stderr, "Set command port with -h\n");
1444         exit(1);
1445     }
1446
1447     global_parameters.ccl_filter = load_cclfile("../etc/default.bib");
1448     global_parameters.yaz_marc = yaz_marc_create();
1449     yaz_marc_subfield_str(global_parameters.yaz_marc, "\t");
1450     global_parameters.odr_in = odr_createmem(ODR_DECODE);
1451     global_parameters.odr_out = odr_createmem(ODR_ENCODE);
1452
1453     event_loop(&channel_list);
1454
1455     return 0;
1456 }
1457
1458 /*
1459  * Local variables:
1460  * c-basic-offset: 4
1461  * indent-tabs-mode: nil
1462  * End:
1463  * vim: shiftwidth=4 tabstop=8 expandtab
1464  */