1 /* $Id: pazpar2.c,v 1.41 2007-01-17 15:32:39 quinn Exp $ */
8 #include <sys/socket.h>
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>
29 #include <yaz/timing.h>
32 #include <netinet/in.h>
37 #include "termlists.h"
39 #include "relevance.h"
44 static void client_fatal(struct client *cl);
45 static void connection_destroy(struct connection *co);
46 static int client_prep_connection(struct client *cl);
47 static void ingest_records(struct client *cl, Z_Records *r);
48 static struct conf_retrievalprofile *database_retrieval_profile(struct database *db);
49 void session_alert_watch(struct session *s, int what);
51 IOCHAN channel_list = 0; // Master list of connections we're handling events to
53 static struct connection *connection_freelist = 0;
54 static struct client *client_freelist = 0;
56 static struct host *hosts = 0; // The hosts we know about
57 static struct database *databases = 0; // The databases we know about
59 static char *client_states[] = {
63 "Client_Initializing",
68 "Client_Disconnected",
72 // Note: Some things in this structure will eventually move to configuration
73 struct parameters global_parameters =
81 "Index Data PazPar2 (MasterKey)",
93 static int send_apdu(struct client *c, Z_APDU *a)
95 struct connection *co = c->connection;
99 if (!z_APDU(global_parameters.odr_out, &a, 0, 0))
101 odr_perror(global_parameters.odr_out, "Encoding APDU");
104 buf = odr_getbuf(global_parameters.odr_out, &len, 0);
105 r = cs_put(co->link, buf, len);
108 yaz_log(YLOG_WARN, "cs_put: %s", cs_errmsg(cs_errno(co->link)));
113 fprintf(stderr, "cs_put incomplete (ParaZ does not handle that)\n");
116 odr_reset(global_parameters.odr_out); /* release the APDU structure */
117 co->state = Conn_Waiting;
122 static void send_init(IOCHAN i)
124 struct connection *co = iochan_getdata(i);
125 struct client *cl = co->client;
126 Z_APDU *a = zget_APDU(global_parameters.odr_out, Z_APDU_initRequest);
128 a->u.initRequest->implementationId = global_parameters.implementationId;
129 a->u.initRequest->implementationName = global_parameters.implementationName;
130 a->u.initRequest->implementationVersion =
131 global_parameters.implementationVersion;
132 ODR_MASK_SET(a->u.initRequest->options, Z_Options_search);
133 ODR_MASK_SET(a->u.initRequest->options, Z_Options_present);
134 ODR_MASK_SET(a->u.initRequest->options, Z_Options_namedResultSets);
136 ODR_MASK_SET(a->u.initRequest->protocolVersion, Z_ProtocolVersion_1);
137 ODR_MASK_SET(a->u.initRequest->protocolVersion, Z_ProtocolVersion_2);
138 ODR_MASK_SET(a->u.initRequest->protocolVersion, Z_ProtocolVersion_3);
139 if (send_apdu(cl, a) >= 0)
141 iochan_setflags(i, EVENT_INPUT);
142 cl->state = Client_Initializing;
145 cl->state = Client_Error;
146 odr_reset(global_parameters.odr_out);
149 static void send_search(IOCHAN i)
151 struct connection *co = iochan_getdata(i);
152 struct client *cl = co->client;
153 struct session *se = cl->session;
154 struct database *db = cl->database;
155 Z_APDU *a = zget_APDU(global_parameters.odr_out, Z_APDU_searchRequest);
156 int ndb, cerror, cpos;
159 struct ccl_rpn_node *cn;
160 int ssub = 0, lslb = 100000, mspn = 10;
162 yaz_log(YLOG_DEBUG, "Sending search");
164 cn = ccl_find_str(global_parameters.ccl_filter, se->query, &cerror, &cpos);
167 a->u.searchRequest->query = zquery = odr_malloc(global_parameters.odr_out,
169 zquery->which = Z_Query_type_1;
170 zquery->u.type_1 = ccl_rpn_query(global_parameters.odr_out, cn);
173 for (ndb = 0; db->databases[ndb]; ndb++)
175 databaselist = odr_malloc(global_parameters.odr_out, sizeof(char*) * ndb);
176 for (ndb = 0; db->databases[ndb]; ndb++)
177 databaselist[ndb] = db->databases[ndb];
179 a->u.presentRequest->preferredRecordSyntax =
180 yaz_oidval_to_z3950oid(global_parameters.odr_out,
181 CLASS_RECSYN, VAL_USMARC);
182 a->u.searchRequest->smallSetUpperBound = &ssub;
183 a->u.searchRequest->largeSetLowerBound = &lslb;
184 a->u.searchRequest->mediumSetPresentNumber = &mspn;
185 a->u.searchRequest->resultSetName = "Default";
186 a->u.searchRequest->databaseNames = databaselist;
187 a->u.searchRequest->num_databaseNames = ndb;
189 if (send_apdu(cl, a) >= 0)
191 iochan_setflags(i, EVENT_INPUT);
192 cl->state = Client_Searching;
193 cl->requestid = se->requestid;
196 cl->state = Client_Error;
198 odr_reset(global_parameters.odr_out);
201 static void send_present(IOCHAN i)
203 struct connection *co = iochan_getdata(i);
204 struct client *cl = co->client;
205 Z_APDU *a = zget_APDU(global_parameters.odr_out, Z_APDU_presentRequest);
207 int start = cl->records + 1;
209 toget = global_parameters.chunk;
210 if (toget > global_parameters.toget - cl->records)
211 toget = global_parameters.toget - cl->records;
212 if (toget > cl->hits - cl->records)
213 toget = cl->hits - cl->records;
215 yaz_log(YLOG_DEBUG, "Trying to present %d records\n", toget);
217 a->u.presentRequest->resultSetStartPoint = &start;
218 a->u.presentRequest->numberOfRecordsRequested = &toget;
220 a->u.presentRequest->resultSetId = "Default";
222 a->u.presentRequest->preferredRecordSyntax =
223 yaz_oidval_to_z3950oid(global_parameters.odr_out,
224 CLASS_RECSYN, VAL_USMARC);
226 if (send_apdu(cl, a) >= 0)
228 iochan_setflags(i, EVENT_INPUT);
229 cl->state = Client_Presenting;
232 cl->state = Client_Error;
233 odr_reset(global_parameters.odr_out);
236 static void do_initResponse(IOCHAN i, Z_APDU *a)
238 struct connection *co = iochan_getdata(i);
239 struct client *cl = co->client;
240 Z_InitResponse *r = a->u.initResponse;
242 yaz_log(YLOG_DEBUG, "Received init response");
246 cl->state = Client_Idle;
249 cl->state = Client_Failed; // FIXME need to do something to the connection
252 static void do_searchResponse(IOCHAN i, Z_APDU *a)
254 struct connection *co = iochan_getdata(i);
255 struct client *cl = co->client;
256 struct session *se = cl->session;
257 Z_SearchResponse *r = a->u.searchResponse;
259 yaz_log(YLOG_DEBUG, "Searchresponse (status=%d)", *r->searchStatus);
261 if (*r->searchStatus)
263 cl->hits = *r->resultCount;
264 se->total_hits += cl->hits;
265 if (r->presentStatus && !*r->presentStatus && r->records)
267 yaz_log(YLOG_DEBUG, "Records in search response");
268 cl->records += *r->numberOfRecordsReturned;
269 ingest_records(cl, r->records);
271 cl->state = Client_Idle;
276 cl->state = Client_Error;
278 Z_Records *recs = r->records;
279 if (recs->which == Z_Records_NSD)
281 yaz_log(YLOG_WARN, "Non-surrogate diagnostic");
282 cl->diagnostic = *recs->u.nonSurrogateDiagnostic->condition;
283 cl->state = Client_Error;
289 char *normalize_mergekey(char *buf, int skiparticle)
291 char *p = buf, *pout = buf;
296 char articles[] = "the den der die des an a "; // must end in space
298 while (*p && !isalnum(*p))
301 while (*p && *p != ' ' && pout - firstword < 62)
302 *(pout++) = tolower(*(p++));
305 if (!strstr(articles, firstword))
312 while (*p && !isalnum(*p))
315 *(pout++) = tolower(*(p++));
318 while (*p && !isalnum(*p))
325 while (pout > buf && *pout == ' ');
332 // FIXME needs to be generalized. Should flexibly generate X lists per search
333 static void extract_subject(struct session *s, const char *rec)
335 const char *field, *subfield;
337 while ((field = find_field(rec, "650")))
340 if ((subfield = find_subfield(field, 'a')))
346 ef = index(subfield, '\n');
349 if ((e = index(subfield, '\t')) && e < ef)
351 while (ef > subfield && !isalpha(*(ef - 1)) && *(ef - 1) != ')')
355 memcpy(buf, subfield, len);
359 termlist_insert(s->termlist, buf);
366 static void add_facet(struct session *s, const char *type, const char *value)
372 for (i = 0; i < s->num_termlists; i++)
373 if (!strcmp(s->termlists[i].name, type))
375 if (i == s->num_termlists)
377 if (i == SESSION_MAX_TERMLISTS)
379 yaz_log(YLOG_FATAL, "Too many termlists");
382 s->termlists[i].name = nmem_strdup(s->nmem, type);
383 s->termlists[i].termlist = termlist_create(s->nmem, s->expected_maxrecs, 15);
384 s->num_termlists = i + 1;
386 termlist_insert(s->termlists[i].termlist, value);
389 int yaz_marc_write_xml();
391 static xmlDoc *normalize_record(struct client *cl, Z_External *rec)
393 struct conf_retrievalprofile *rprofile = cl->database->rprofile;
394 struct conf_retrievalmap *m;
398 // First normalize to XML
399 if (rprofile->native_syntax == Nativesyn_iso2709)
403 if (rec->which != Z_External_octet)
405 yaz_log(YLOG_WARN, "Unexpected external branch, probably BER");
408 buf = (char*) rec->u.octet_aligned->buf;
409 len = rec->u.octet_aligned->len;
410 if (yaz_marc_read_iso2709(rprofile->yaz_marc, buf, len) < 0)
412 yaz_log(YLOG_WARN, "Failed to decode MARC");
415 if (yaz_marc_write_xml(rprofile->yaz_marc, &res,
416 "http://www.loc.gov/MARC21/slim", 0, 0) < 0)
418 yaz_log(YLOG_WARN, "Failed to encode as XML");
421 rdoc = xmlNewDoc("1.0");
422 xmlDocSetRootElement(rdoc, res);
426 yaz_log(YLOG_FATAL, "Unknown native_syntax in normalize_record");
429 for (m = rprofile->maplist; m; m = m->next)
432 if (m->type != Map_xslt)
434 yaz_log(YLOG_WARN, "Unknown map type");
437 if (!(new = xsltApplyStylesheet(m->stylesheet, rdoc, 0)))
439 yaz_log(YLOG_WARN, "XSLT transformation failed");
445 if (global_parameters.dump_records)
447 fprintf(stderr, "Record:\n----------------\n");
448 #if LIBXML_VERSION >= 20600
449 xmlDocFormatDump(stderr, rdoc, 1);
451 xmlDocDump(stderr, rdoc);
457 // Extract what appears to be years from buf, storing highest and
459 static int extract_years(const char *buf, int *first, int *last)
468 while (*buf && !isdigit(*buf))
471 for (e = buf; *e && isdigit(*e); e++)
475 int value = atoi(buf);
476 if (*first < 0 || value < *first)
478 if (*last < 0 || value > *last)
486 static struct record *ingest_record(struct client *cl, Z_External *rec)
488 xmlDoc *xdoc = normalize_record(cl, rec);
491 struct record_cluster *cluster;
492 struct session *se = cl->session;
493 xmlChar *mergekey, *mergekey_norm;
496 struct conf_service *service = global_parameters.server->service;
501 root = xmlDocGetRootElement(xdoc);
502 if (!(mergekey = xmlGetProp(root, "mergekey")))
504 yaz_log(YLOG_WARN, "No mergekey found in record");
509 res = nmem_malloc(se->nmem, sizeof(struct record));
511 res->metadata = nmem_malloc(se->nmem,
512 sizeof(struct record_metadata*) * service->num_metadata);
513 memset(res->metadata, 0, sizeof(struct record_metadata*) * service->num_metadata);
515 mergekey_norm = nmem_strdup(se->nmem, (char*) mergekey);
517 normalize_mergekey(mergekey_norm, 0);
519 cluster = reclist_insert(se->reclist, res, mergekey_norm, &se->total_merged);
520 if (global_parameters.dump_records)
521 yaz_log(YLOG_LOG, "Cluster id %d from %s (#%d)", cluster->recid,
522 cl->database->url, cl->records);
525 /* no room for record */
529 relevance_newrec(se->relevance, cluster);
531 for (n = root->children; n; n = n->next)
539 if (n->type != XML_ELEMENT_NODE)
541 if (!strcmp(n->name, "metadata"))
543 struct conf_metadata *md = 0;
544 struct conf_sortkey *sk = 0;
545 struct record_metadata **wheretoput, *newm;
549 type = xmlGetProp(n, "type");
550 value = xmlNodeListGetString(xdoc, n->children, 0);
555 // First, find out what field we're looking at
556 for (imeta = 0; imeta < service->num_metadata; imeta++)
557 if (!strcmp(type, service->metadata[imeta].name))
559 md = &service->metadata[imeta];
560 if (md->sortkey_offset >= 0)
561 sk = &service->sortkeys[md->sortkey_offset];
566 yaz_log(YLOG_WARN, "Ignoring unknown metadata element: %s", type);
570 // Find out where we are putting it
571 if (md->merge == Metadata_merge_no)
572 wheretoput = &res->metadata[imeta];
574 wheretoput = &cluster->metadata[imeta];
577 newm = nmem_malloc(se->nmem, sizeof(struct record_metadata));
579 if (md->type == Metadata_type_generic)
582 newm->data.text = nmem_strdup(se->nmem, value);
583 for (p = newm->data.text + strlen(newm->data.text) - 1;
584 p > newm->data.text && strchr(" ,/.", *p); p--)
588 else if (md->type == Metadata_type_year)
590 if (extract_years(value, &first, &last) < 0)
595 yaz_log(YLOG_WARN, "Unknown type in metadata element %s", type);
598 if (md->type == Metadata_type_year && md->merge != Metadata_merge_range)
600 yaz_log(YLOG_WARN, "Only range merging supported for years");
603 if (md->merge == Metadata_merge_unique)
605 struct record_metadata *mnode;
606 for (mnode = *wheretoput; mnode; mnode = mnode->next)
607 if (!strcmp(mnode->data.text, newm->data.text))
611 newm->next = *wheretoput;
615 else if (md->merge == Metadata_merge_longest)
618 strlen(newm->data.text) > strlen((*wheretoput)->data.text))
623 char *s = nmem_strdup(se->nmem, newm->data.text);
624 if (!cluster->sortkeys[md->sortkey_offset])
625 cluster->sortkeys[md->sortkey_offset] =
626 nmem_malloc(se->nmem, sizeof(union data_types));
627 normalize_mergekey(s,
628 (sk->type == Metadata_sortkey_skiparticle));
629 cluster->sortkeys[md->sortkey_offset]->text = s;
633 else if (md->merge == Metadata_merge_all || md->merge == Metadata_merge_no)
635 newm->next = *wheretoput;
638 else if (md->merge == Metadata_merge_range)
640 assert(md->type == Metadata_type_year);
644 (*wheretoput)->data.number.min = first;
645 (*wheretoput)->data.number.max = last;
647 cluster->sortkeys[md->sortkey_offset] = &newm->data;
651 if (first < (*wheretoput)->data.number.min)
652 (*wheretoput)->data.number.min = first;
653 if (last > (*wheretoput)->data.number.max)
654 (*wheretoput)->data.number.max = last;
659 union data_types *sdata = cluster->sortkeys[md->sortkey_offset];
660 yaz_log(YLOG_LOG, "SK range: %d-%d", sdata->number.min, sdata->number.max);
665 yaz_log(YLOG_WARN, "Don't know how to merge on element name %s", md->name);
668 relevance_countwords(se->relevance, cluster, value, md->rank);
671 if (md->type == Metadata_type_year)
674 sprintf(year, "%d", last);
675 add_facet(se, type, year);
678 sprintf(year, "%d", first);
679 add_facet(se, type, year);
683 add_facet(se, type, value);
690 yaz_log(YLOG_WARN, "Unexpected element %s in internal record", n->name);
699 relevance_donerecord(se->relevance, cluster);
705 static void ingest_records(struct client *cl, Z_Records *r)
708 yaz_timing_t t = yaz_timing_create();
711 struct session *s = cl->session;
712 Z_NamePlusRecordList *rlist;
715 if (r->which != Z_Records_DBOSD)
717 rlist = r->u.databaseOrSurDiagnostics;
718 for (i = 0; i < rlist->num_records; i++)
720 Z_NamePlusRecord *npr = rlist->records[i];
723 if (npr->which != Z_NamePlusRecord_databaseRecord)
725 yaz_log(YLOG_WARN, "Unexpected record type, probably diagnostic");
729 rec = ingest_record(cl, npr->u.databaseRecord);
733 if (s->watchlist[SESSION_WATCH_RECORDS].fun && rlist->num_records)
734 session_alert_watch(s, SESSION_WATCH_RECORDS);
738 yaz_log(YLOG_LOG, "ingest_records %6.5f %3.2f %3.2f",
739 yaz_timing_get_real(t), yaz_timing_get_user(t),
740 yaz_timing_get_sys(t));
741 yaz_timing_destroy(&t);
745 static void do_presentResponse(IOCHAN i, Z_APDU *a)
747 struct connection *co = iochan_getdata(i);
748 struct client *cl = co->client;
749 Z_PresentResponse *r = a->u.presentResponse;
752 Z_Records *recs = r->records;
753 if (recs->which == Z_Records_NSD)
755 yaz_log(YLOG_WARN, "Non-surrogate diagnostic");
756 cl->diagnostic = *recs->u.nonSurrogateDiagnostic->condition;
757 cl->state = Client_Error;
761 if (!*r->presentStatus && cl->state != Client_Error)
763 yaz_log(YLOG_DEBUG, "Good Present response");
764 ingest_records(cl, r->records);
765 cl->state = Client_Idle;
767 else if (*r->presentStatus)
769 yaz_log(YLOG_WARN, "Bad Present response");
770 cl->state = Client_Error;
774 static void handler(IOCHAN i, int event)
776 struct connection *co = iochan_getdata(i);
777 struct client *cl = co->client;
778 struct session *se = 0;
784 yaz_log(YLOG_WARN, "Destroying orphan connection");
785 connection_destroy(co);
789 if (co->state == Conn_Connecting && event & EVENT_OUTPUT)
792 socklen_t errlen = sizeof(errcode);
794 if (getsockopt(cs_fileno(co->link), SOL_SOCKET, SO_ERROR, &errcode,
795 &errlen) < 0 || errcode != 0)
802 yaz_log(YLOG_DEBUG, "Connect OK");
803 co->state = Conn_Open;
805 cl->state = Client_Connected;
809 else if (event & EVENT_INPUT)
811 int len = cs_get(co->link, &co->ibuf, &co->ibufsize);
815 yaz_log(YLOG_WARN|YLOG_ERRNO, "Error reading from Z server");
816 connection_destroy(co);
821 yaz_log(YLOG_WARN, "EOF reading from Z server");
822 connection_destroy(co);
825 else if (len > 1) // We discard input if we have no connection
827 co->state = Conn_Open;
829 if (cl && (cl->requestid == se->requestid || cl->state == Client_Initializing))
833 odr_reset(global_parameters.odr_in);
834 odr_setbuf(global_parameters.odr_in, co->ibuf, len, 0);
835 if (!z_APDU(global_parameters.odr_in, &a, 0, 0))
842 case Z_APDU_initResponse:
843 do_initResponse(i, a);
845 case Z_APDU_searchResponse:
846 do_searchResponse(i, a);
848 case Z_APDU_presentResponse:
849 do_presentResponse(i, a);
852 yaz_log(YLOG_WARN, "Unexpected result from server");
856 // We aren't expecting staggered output from target
857 // if (cs_more(t->link))
858 // iochan_setevent(i, EVENT_INPUT);
860 else // we throw away response and go to idle mode
862 yaz_log(YLOG_DEBUG, "Ignoring result of expired operation");
863 cl->state = Client_Idle;
866 /* if len==1 we do nothing but wait for more input */
869 if (cl->state == Client_Connected) {
873 if (cl->state == Client_Idle)
875 if (cl->requestid != se->requestid && *se->query) {
878 else if (cl->hits > 0 && cl->records < global_parameters.toget &&
879 cl->records < cl->hits) {
885 // Disassociate connection from client
886 static void connection_release(struct connection *co)
888 struct client *cl = co->client;
890 yaz_log(YLOG_DEBUG, "Connection release %s", co->host->hostport);
897 // Close connection and recycle structure
898 static void connection_destroy(struct connection *co)
900 struct host *h = co->host;
902 iochan_destroy(co->iochan);
904 yaz_log(YLOG_DEBUG, "Connection destroy %s", co->host->hostport);
905 if (h->connections == co)
906 h->connections = co->next;
909 struct connection *pco;
910 for (pco = h->connections; pco && pco->next != co; pco = pco->next)
913 pco->next = co->next;
919 if (co->client->state != Client_Idle)
920 co->client->state = Client_Disconnected;
921 co->client->connection = 0;
923 co->next = connection_freelist;
924 connection_freelist = co;
927 // Creates a new connection for client, associated with the host of
929 static struct connection *connection_create(struct client *cl)
931 struct connection *new;
936 yaz_log(YLOG_DEBUG, "Connection create %s", cl->database->url);
937 if (!(link = cs_create(tcpip_type, 0, PROTO_Z3950)))
939 yaz_log(YLOG_FATAL|YLOG_ERRNO, "Failed to create comstack");
943 if (!(addr = cs_straddr(link, cl->database->host->ipport)))
945 yaz_log(YLOG_WARN|YLOG_ERRNO, "Lookup of IP address %s failed?",
946 cl->database->host->ipport);
950 res = cs_connect(link, addr);
953 yaz_log(YLOG_WARN|YLOG_ERRNO, "cs_connect %s", cl->database->url);
957 if ((new = connection_freelist))
958 connection_freelist = new->next;
961 new = xmalloc(sizeof (struct connection));
965 new->state = Conn_Connecting;
966 new->host = cl->database->host;
967 new->next = new->host->connections;
968 new->host->connections = new;
970 cl->connection = new;
973 new->iochan = iochan_create(cs_fileno(link), handler, 0);
974 iochan_setdata(new->iochan, new);
975 new->iochan->next = channel_list;
976 channel_list = new->iochan;
980 // Close connection and set state to error
981 static void client_fatal(struct client *cl)
983 yaz_log(YLOG_WARN, "Fatal error from %s", cl->database->url);
984 connection_destroy(cl->connection);
985 cl->state = Client_Error;
988 // Ensure that client has a connection associated
989 static int client_prep_connection(struct client *cl)
991 struct connection *co;
992 struct session *se = cl->session;
993 struct host *host = cl->database->host;
997 yaz_log(YLOG_DEBUG, "Client prep %s", cl->database->url);
1001 // See if someone else has an idle connection
1002 // We should look at timestamps here to select the longest-idle connection
1003 for (co = host->connections; co; co = co->next)
1004 if (co->state == Conn_Open && (!co->client || co->client->session != se))
1008 connection_release(co);
1009 cl->connection = co;
1013 co = connection_create(cl);
1017 if (co->state == Conn_Connecting)
1019 cl->state = Client_Connecting;
1020 iochan_setflag(co->iochan, EVENT_OUTPUT);
1022 else if (co->state == Conn_Open)
1024 if (cl->state == Client_Error || cl->state == Client_Disconnected)
1025 cl->state = Client_Idle;
1026 iochan_setflag(co->iochan, EVENT_OUTPUT);
1034 // This function will most likely vanish when a proper target profile mechanism is
1036 void load_simpletargets(const char *fn)
1038 FILE *f = fopen(fn, "r");
1043 yaz_log(YLOG_WARN|YLOG_ERRNO, "open %s", fn);
1047 while (fgets(line, 255, f))
1051 struct database *database;
1053 if (strncmp(line, "target ", 7))
1056 url[strlen(url) - 1] = '\0';
1057 yaz_log(YLOG_DEBUG, "Target: %s", url);
1058 if ((db = strchr(url, '/')))
1063 for (host = hosts; host; host = host->next)
1064 if (!strcmp(url, host->hostport))
1068 struct addrinfo *addrinfo, hints;
1071 unsigned char addrbuf[4];
1074 host = xmalloc(sizeof(struct host));
1075 host->hostport = xstrdup(url);
1076 host->connections = 0;
1078 if ((port = strchr(url, ':')))
1084 hints.ai_family = PF_INET;
1085 hints.ai_socktype = SOCK_STREAM;
1086 hints.ai_protocol = IPPROTO_TCP;
1087 hints.ai_addrlen = 0;
1089 hints.ai_canonname = 0;
1091 // This is not robust code. It assumes that getaddrinfo returns AF_INET
1093 if ((res = getaddrinfo(url, port, &hints, &addrinfo)))
1095 yaz_log(YLOG_WARN, "Failed to resolve %s: %s", url, gai_strerror(res));
1096 xfree(host->hostport);
1100 assert(addrinfo->ai_family == PF_INET);
1101 memcpy(addrbuf, &((struct sockaddr_in*)addrinfo->ai_addr)->sin_addr.s_addr, 4);
1102 sprintf(ipport, "%u.%u.%u.%u:%s",
1103 addrbuf[0], addrbuf[1], addrbuf[2], addrbuf[3], port);
1104 host->ipport = xstrdup(ipport);
1105 freeaddrinfo(addrinfo);
1109 database = xmalloc(sizeof(struct database));
1110 database->host = host;
1111 database->url = xmalloc(strlen(url) + strlen(db) + 2);
1112 strcpy(database->url, url);
1113 strcat(database->url, "/");
1114 strcat(database->url, db);
1116 database->databases = xmalloc(2 * sizeof(char *));
1117 database->databases[0] = xstrdup(db);
1118 database->databases[1] = 0;
1119 database->errors = 0;
1120 database->qprofile = 0;
1121 database->rprofile = database_retrieval_profile(database);
1122 database->next = databases;
1123 databases = database;
1129 static void pull_terms(NMEM nmem, struct ccl_rpn_node *n, char **termlist, int *num)
1137 pull_terms(nmem, n->u.p[0], termlist, num);
1138 pull_terms(nmem, n->u.p[1], termlist, num);
1141 termlist[(*num)++] = nmem_strdup(nmem, n->u.t.term);
1148 // Extract terms from query into null-terminated termlist
1149 static int extract_terms(NMEM nmem, char *query, char **termlist)
1152 struct ccl_rpn_node *n;
1155 n = ccl_find_str(global_parameters.ccl_filter, query, &error, &pos);
1158 pull_terms(nmem, n, termlist, &num);
1164 static struct client *client_create(void)
1167 if (client_freelist)
1169 r = client_freelist;
1170 client_freelist = client_freelist->next;
1173 r = xmalloc(sizeof(struct client));
1182 r->state = Client_Disconnected;
1187 void client_destroy(struct client *c)
1189 struct session *se = c->session;
1190 if (c == se->clients)
1191 se->clients = c->next;
1195 for (cc = se->clients; cc && cc->next != c; cc = cc->next)
1201 connection_release(c->connection);
1202 c->next = client_freelist;
1203 client_freelist = c;
1206 void session_set_watch(struct session *s, int what, session_watchfun fun, void *data)
1208 s->watchlist[what].fun = fun;
1209 s->watchlist[what].data = data;
1212 void session_alert_watch(struct session *s, int what)
1214 if (!s->watchlist[what].fun)
1216 (*s->watchlist[what].fun)(s->watchlist[what].data);
1217 s->watchlist[what].fun = 0;
1218 s->watchlist[what].data = 0;
1221 // This needs to be extended with selection criteria
1222 static struct conf_retrievalprofile *database_retrieval_profile(struct database *db)
1226 yaz_log(YLOG_FATAL, "Must load configuration (-f)");
1229 if (!config->retrievalprofiles)
1231 yaz_log(YLOG_FATAL, "No retrieval profiles defined");
1233 return config->retrievalprofiles;
1236 // This should be extended with parameters to control selection criteria
1237 // Associates a set of clients with a session;
1238 int select_targets(struct session *se)
1240 struct database *db;
1244 client_destroy(se->clients);
1245 for (db = databases; db; db = db->next)
1247 struct client *cl = client_create();
1250 cl->next = se->clients;
1257 int session_active_clients(struct session *s)
1262 for (c = s->clients; c; c = c->next)
1263 if (c->connection && (c->state == Client_Connecting ||
1264 c->state == Client_Initializing ||
1265 c->state == Client_Searching ||
1266 c->state == Client_Presenting))
1272 char *search(struct session *se, char *query)
1274 int live_channels = 0;
1277 yaz_log(YLOG_DEBUG, "Search");
1279 strcpy(se->query, query);
1281 nmem_reset(se->nmem);
1282 for (cl = se->clients; cl; cl = cl->next)
1288 if (client_prep_connection(cl))
1294 int maxrecs = live_channels * global_parameters.toget;
1295 se->num_termlists = 0;
1296 se->reclist = reclist_create(se->nmem, maxrecs);
1297 extract_terms(se->nmem, query, p);
1298 se->relevance = relevance_create(se->nmem, (const char **) p, maxrecs);
1299 se->total_records = se->total_hits = se->total_merged = 0;
1300 se->expected_maxrecs = maxrecs;
1308 void destroy_session(struct session *s)
1310 yaz_log(YLOG_LOG, "Destroying session");
1312 client_destroy(s->clients);
1313 nmem_destroy(s->nmem);
1314 wrbuf_free(s->wrbuf, 1);
1317 struct session *new_session()
1320 struct session *session = xmalloc(sizeof(*session));
1322 yaz_log(YLOG_DEBUG, "New pazpar2 session");
1324 session->total_hits = 0;
1325 session->total_records = 0;
1326 session->num_termlists = 0;
1327 session->reclist = 0;
1328 session->requestid = -1;
1329 session->clients = 0;
1330 session->expected_maxrecs = 0;
1331 session->query[0] = '\0';
1332 session->nmem = nmem_create();
1333 session->wrbuf = wrbuf_alloc();
1334 for (i = 0; i <= SESSION_WATCH_MAX; i++)
1336 session->watchlist[i].data = 0;
1337 session->watchlist[i].fun = 0;
1340 select_targets(session);
1345 struct hitsbytarget *hitsbytarget(struct session *se, int *count)
1347 static struct hitsbytarget res[1000]; // FIXME MM
1351 for (cl = se->clients; cl; cl = cl->next)
1353 strcpy(res[*count].id, cl->database->host->hostport);
1354 res[*count].hits = cl->hits;
1355 res[*count].records = cl->records;
1356 res[*count].diagnostic = cl->diagnostic;
1357 res[*count].state = client_states[cl->state];
1358 res[*count].connected = cl->connection ? 1 : 0;
1365 struct termlist_score **termlist(struct session *s, const char *name, int *num)
1369 for (i = 0; i < s->num_termlists; i++)
1370 if (!strcmp(s->termlists[i].name, name))
1371 return termlist_highscore(s->termlists[i].termlist, num);
1375 #ifdef MISSING_HEADERS
1376 void report_nmem_stats(void)
1378 size_t in_use, is_free;
1380 nmem_get_memory_in_use(&in_use);
1381 nmem_get_memory_free(&is_free);
1383 yaz_log(YLOG_LOG, "nmem stat: use=%ld free=%ld",
1384 (long) in_use, (long) is_free);
1388 struct record_cluster *show_single(struct session *s, int id)
1390 struct record_cluster *r;
1392 reclist_rewind(s->reclist);
1393 while ((r = reclist_read_record(s->reclist)))
1399 struct record_cluster **show(struct session *s, struct reclist_sortparms *sp, int start,
1400 int *num, int *total, int *sumhits, NMEM nmem_show)
1402 struct record_cluster **recs = nmem_malloc(nmem_show, *num
1403 * sizeof(struct record_cluster *));
1404 struct reclist_sortparms *spp;
1407 yaz_timing_t t = yaz_timing_create();
1410 for (spp = sp; spp; spp = spp->next)
1411 if (spp->type == Metadata_sortkey_relevance)
1413 relevance_prepare_read(s->relevance, s->reclist);
1416 reclist_sort(s->reclist, sp);
1418 *total = s->reclist->num_records;
1419 *sumhits = s->total_hits;
1421 for (i = 0; i < start; i++)
1422 if (!reclist_read_record(s->reclist))
1429 for (i = 0; i < *num; i++)
1431 struct record_cluster *r = reclist_read_record(s->reclist);
1441 yaz_log(YLOG_LOG, "show %6.5f %3.2f %3.2f",
1442 yaz_timing_get_real(t), yaz_timing_get_user(t),
1443 yaz_timing_get_sys(t));
1444 yaz_timing_destroy(&t);
1449 void statistics(struct session *se, struct statistics *stat)
1454 memset(stat, 0, sizeof(*stat));
1455 for (cl = se->clients; cl; cl = cl->next)
1457 if (!cl->connection)
1458 stat->num_no_connection++;
1461 case Client_Connecting: stat->num_connecting++; break;
1462 case Client_Initializing: stat->num_initializing++; break;
1463 case Client_Searching: stat->num_searching++; break;
1464 case Client_Presenting: stat->num_presenting++; break;
1465 case Client_Idle: stat->num_idle++; break;
1466 case Client_Failed: stat->num_failed++; break;
1467 case Client_Error: stat->num_error++; break;
1472 stat->num_hits = se->total_hits;
1473 stat->num_records = se->total_records;
1475 stat->num_clients = count;
1478 static CCL_bibset load_cclfile(const char *fn)
1480 CCL_bibset res = ccl_qual_mk();
1481 if (ccl_qual_fname(res, fn) < 0)
1483 yaz_log(YLOG_FATAL|YLOG_ERRNO, "%s", fn);
1489 static void start_http_listener(void)
1492 struct conf_server *ser = global_parameters.server;
1494 if (*global_parameters.listener_override)
1495 strcpy(hp, global_parameters.listener_override);
1498 strcpy(hp, ser->host ? ser->host : "");
1503 sprintf(hp + strlen(hp), "%d", ser->port);
1509 static void start_proxy(void)
1512 struct conf_server *ser = global_parameters.server;
1514 if (*global_parameters.proxy_override)
1515 strcpy(hp, global_parameters.proxy_override);
1516 else if (ser->proxy_host || ser->proxy_port)
1518 strcpy(hp, ser->proxy_host ? ser->proxy_host : "");
1519 if (ser->proxy_port)
1523 sprintf(hp + strlen(hp), "%d", ser->proxy_port);
1529 http_set_proxyaddr(hp);
1532 int main(int argc, char **argv)
1537 if (signal(SIGPIPE, SIG_IGN) == SIG_ERR)
1538 yaz_log(YLOG_WARN|YLOG_ERRNO, "signal");
1540 yaz_log_init(YLOG_DEFAULT_LEVEL, "pazpar2", 0);
1542 while ((ret = options("f:x:h:p:C:s:d", argv, argc, &arg)) != -2)
1546 if (!read_config(arg))
1550 strcpy(global_parameters.listener_override, arg);
1553 global_parameters.ccl_filter = load_cclfile(arg);
1556 strcpy(global_parameters.proxy_override, arg);
1559 load_simpletargets(arg);
1562 global_parameters.dump_records = 1;
1565 fprintf(stderr, "Usage: pazpar2\n"
1567 " -h [host:]port (REST protocol listener)\n"
1569 " -s simpletargetfile\n"
1570 " -p hostname[:portno] (HTTP proxy)\n"
1571 " -d (show internal records)\n");
1578 yaz_log(YLOG_FATAL, "Load config with -f");
1581 global_parameters.server = config->servers;
1583 start_http_listener();
1585 global_parameters.ccl_filter = load_cclfile("../etc/default.bib");
1586 global_parameters.yaz_marc = yaz_marc_create();
1587 yaz_marc_subfield_str(global_parameters.yaz_marc, "\t");
1588 global_parameters.odr_in = odr_createmem(ODR_DECODE);
1589 global_parameters.odr_out = odr_createmem(ODR_ENCODE);
1591 event_loop(&channel_list);
1599 * indent-tabs-mode: nil
1601 * vim: shiftwidth=4 tabstop=8 expandtab