1 /* $Id: pazpar2.c,v 1.53 2007-03-26 14:00:21 marc 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"
45 static void client_fatal(struct client *cl);
46 static void connection_destroy(struct connection *co);
47 static int client_prep_connection(struct client *cl);
48 static void ingest_records(struct client *cl, Z_Records *r);
49 //static struct conf_retrievalprofile *database_retrieval_profile(struct database *db);
50 void session_alert_watch(struct session *s, int what);
52 IOCHAN channel_list = 0; // Master list of connections we're handling events to
54 static struct connection *connection_freelist = 0;
55 static struct client *client_freelist = 0;
57 static char *client_states[] = {
61 "Client_Initializing",
66 "Client_Disconnected",
70 // Note: Some things in this structure will eventually move to configuration
71 struct parameters global_parameters =
80 "Index Data PazPar2 (MasterKey)",
92 static int send_apdu(struct client *c, Z_APDU *a)
94 struct connection *co = c->connection;
98 if (!z_APDU(global_parameters.odr_out, &a, 0, 0))
100 odr_perror(global_parameters.odr_out, "Encoding APDU");
103 buf = odr_getbuf(global_parameters.odr_out, &len, 0);
104 r = cs_put(co->link, buf, len);
107 yaz_log(YLOG_WARN, "cs_put: %s", cs_errmsg(cs_errno(co->link)));
112 fprintf(stderr, "cs_put incomplete (ParaZ does not handle that)\n");
115 odr_reset(global_parameters.odr_out); /* release the APDU structure */
116 co->state = Conn_Waiting;
121 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);
141 /* add virtual host if tunneling through Z39.50 proxy */
143 if (0 < strlen(global_parameters.zproxy_override)
144 && 0 < strlen(cl->database->url))
145 yaz_oi_set_string_oidval(&a->u.initRequest->otherInfo,
146 global_parameters.odr_out, VAL_PROXY,
147 1, cl->database->url);
151 if (send_apdu(cl, a) >= 0)
153 iochan_setflags(i, EVENT_INPUT);
154 cl->state = Client_Initializing;
157 cl->state = Client_Error;
158 odr_reset(global_parameters.odr_out);
161 static void send_search(IOCHAN i)
163 struct connection *co = iochan_getdata(i);
164 struct client *cl = co->client;
165 struct session *se = cl->session;
166 struct database *db = cl->database;
167 Z_APDU *a = zget_APDU(global_parameters.odr_out, Z_APDU_searchRequest);
168 int ndb, cerror, cpos;
171 struct ccl_rpn_node *cn;
172 int ssub = 0, lslb = 100000, mspn = 10;
174 yaz_log(YLOG_DEBUG, "Sending search");
176 cn = ccl_find_str(global_parameters.ccl_filter, se->query, &cerror, &cpos);
179 a->u.searchRequest->query = zquery = odr_malloc(global_parameters.odr_out,
181 zquery->which = Z_Query_type_1;
182 zquery->u.type_1 = ccl_rpn_query(global_parameters.odr_out, cn);
185 for (ndb = 0; db->databases[ndb]; ndb++)
187 databaselist = odr_malloc(global_parameters.odr_out, sizeof(char*) * ndb);
188 for (ndb = 0; db->databases[ndb]; ndb++)
189 databaselist[ndb] = db->databases[ndb];
191 a->u.searchRequest->preferredRecordSyntax =
192 yaz_oidval_to_z3950oid(global_parameters.odr_out,
193 CLASS_RECSYN, VAL_USMARC);
194 a->u.searchRequest->smallSetUpperBound = &ssub;
195 a->u.searchRequest->largeSetLowerBound = &lslb;
196 a->u.searchRequest->mediumSetPresentNumber = &mspn;
197 a->u.searchRequest->resultSetName = "Default";
198 a->u.searchRequest->databaseNames = databaselist;
199 a->u.searchRequest->num_databaseNames = ndb;
201 if (send_apdu(cl, a) >= 0)
203 iochan_setflags(i, EVENT_INPUT);
204 cl->state = Client_Searching;
205 cl->requestid = se->requestid;
208 cl->state = Client_Error;
210 odr_reset(global_parameters.odr_out);
213 static void send_present(IOCHAN i)
215 struct connection *co = iochan_getdata(i);
216 struct client *cl = co->client;
217 Z_APDU *a = zget_APDU(global_parameters.odr_out, Z_APDU_presentRequest);
219 int start = cl->records + 1;
221 toget = global_parameters.chunk;
222 if (toget > global_parameters.toget - cl->records)
223 toget = global_parameters.toget - cl->records;
224 if (toget > cl->hits - cl->records)
225 toget = cl->hits - cl->records;
227 yaz_log(YLOG_DEBUG, "Trying to present %d records\n", toget);
229 a->u.presentRequest->resultSetStartPoint = &start;
230 a->u.presentRequest->numberOfRecordsRequested = &toget;
232 a->u.presentRequest->resultSetId = "Default";
234 a->u.presentRequest->preferredRecordSyntax =
235 yaz_oidval_to_z3950oid(global_parameters.odr_out,
236 CLASS_RECSYN, VAL_USMARC);
238 if (send_apdu(cl, a) >= 0)
240 iochan_setflags(i, EVENT_INPUT);
241 cl->state = Client_Presenting;
244 cl->state = Client_Error;
245 odr_reset(global_parameters.odr_out);
248 static void do_initResponse(IOCHAN i, Z_APDU *a)
250 struct connection *co = iochan_getdata(i);
251 struct client *cl = co->client;
252 Z_InitResponse *r = a->u.initResponse;
254 yaz_log(YLOG_DEBUG, "Received init response");
258 cl->state = Client_Idle;
261 cl->state = Client_Failed; // FIXME need to do something to the connection
264 static void do_searchResponse(IOCHAN i, Z_APDU *a)
266 struct connection *co = iochan_getdata(i);
267 struct client *cl = co->client;
268 struct session *se = cl->session;
269 Z_SearchResponse *r = a->u.searchResponse;
271 yaz_log(YLOG_DEBUG, "Searchresponse (status=%d)", *r->searchStatus);
273 if (*r->searchStatus)
275 cl->hits = *r->resultCount;
276 se->total_hits += cl->hits;
277 if (r->presentStatus && !*r->presentStatus && r->records)
279 yaz_log(YLOG_DEBUG, "Records in search response");
280 ingest_records(cl, r->records);
282 cl->state = Client_Idle;
287 cl->state = Client_Error;
289 Z_Records *recs = r->records;
290 if (recs->which == Z_Records_NSD)
292 yaz_log(YLOG_WARN, "Non-surrogate diagnostic");
293 cl->diagnostic = *recs->u.nonSurrogateDiagnostic->condition;
294 cl->state = Client_Error;
300 char *normalize_mergekey(char *buf, int skiparticle)
302 char *p = buf, *pout = buf;
307 char articles[] = "the den der die des an a "; // must end in space
309 while (*p && !isalnum(*p))
312 while (*p && *p != ' ' && pout - firstword < 62)
313 *(pout++) = tolower(*(p++));
316 if (!strstr(articles, firstword))
323 while (*p && !isalnum(*p))
326 *(pout++) = tolower(*(p++));
329 while (*p && !isalnum(*p))
336 while (pout > buf && *pout == ' ');
343 // FIXME needs to be generalized. Should flexibly generate X lists per search
344 static void extract_subject(struct session *s, const char *rec)
346 const char *field, *subfield;
348 while ((field = find_field(rec, "650")))
351 if ((subfield = find_subfield(field, 'a')))
357 ef = index(subfield, '\n');
360 if ((e = index(subfield, '\t')) && e < ef)
362 while (ef > subfield && !isalpha(*(ef - 1)) && *(ef - 1) != ')')
366 memcpy(buf, subfield, len);
370 termlist_insert(s->termlist, buf);
377 static void add_facet(struct session *s, const char *type, const char *value)
383 for (i = 0; i < s->num_termlists; i++)
384 if (!strcmp(s->termlists[i].name, type))
386 if (i == s->num_termlists)
388 if (i == SESSION_MAX_TERMLISTS)
390 yaz_log(YLOG_FATAL, "Too many termlists");
393 s->termlists[i].name = nmem_strdup(s->nmem, type);
394 s->termlists[i].termlist = termlist_create(s->nmem, s->expected_maxrecs, 15);
395 s->num_termlists = i + 1;
397 termlist_insert(s->termlists[i].termlist, value);
400 static xmlDoc *normalize_record(struct client *cl, Z_External *rec)
402 struct conf_retrievalprofile *rprofile = cl->database->rprofile;
403 struct conf_retrievalmap *m;
407 // First normalize to XML
408 if (rprofile->native_syntax == Nativesyn_iso2709)
412 if (rec->which != Z_External_octet)
414 yaz_log(YLOG_WARN, "Unexpected external branch, probably BER");
417 buf = (char*) rec->u.octet_aligned->buf;
418 len = rec->u.octet_aligned->len;
419 if (yaz_marc_read_iso2709(rprofile->yaz_marc, buf, len) < 0)
421 yaz_log(YLOG_WARN, "Failed to decode MARC");
424 if (yaz_marc_write_xml(rprofile->yaz_marc, &res,
425 "http://www.loc.gov/MARC21/slim", 0, 0) < 0)
427 yaz_log(YLOG_WARN, "Failed to encode as XML");
430 rdoc = xmlNewDoc("1.0");
431 xmlDocSetRootElement(rdoc, res);
435 yaz_log(YLOG_FATAL, "Unknown native_syntax in normalize_record");
439 if (global_parameters.dump_records)
441 fprintf(stderr, "Input Record (normalized):\n----------------\n");
442 #if LIBXML_VERSION >= 20600
443 xmlDocFormatDump(stderr, rdoc, 1);
445 xmlDocDump(stderr, rdoc);
449 for (m = rprofile->maplist; m; m = m->next)
452 if (m->type != Map_xslt)
454 yaz_log(YLOG_WARN, "Unknown map type");
457 if (!(new = xsltApplyStylesheet(m->stylesheet, rdoc, 0)))
459 yaz_log(YLOG_WARN, "XSLT transformation failed");
465 if (global_parameters.dump_records)
467 fprintf(stderr, "Record:\n----------------\n");
468 #if LIBXML_VERSION >= 20600
469 xmlDocFormatDump(stderr, rdoc, 1);
471 xmlDocDump(stderr, rdoc);
477 // Extract what appears to be years from buf, storing highest and
479 static int extract_years(const char *buf, int *first, int *last)
488 while (*buf && !isdigit(*buf))
491 for (e = buf; *e && isdigit(*e); e++)
495 int value = atoi(buf);
496 if (*first < 0 || value < *first)
498 if (*last < 0 || value > *last)
506 static struct record *ingest_record(struct client *cl, Z_External *rec)
508 xmlDoc *xdoc = normalize_record(cl, rec);
511 struct record_cluster *cluster;
512 struct session *se = cl->session;
513 xmlChar *mergekey, *mergekey_norm;
516 struct conf_service *service = global_parameters.server->service;
521 root = xmlDocGetRootElement(xdoc);
522 if (!(mergekey = xmlGetProp(root, "mergekey")))
524 yaz_log(YLOG_WARN, "No mergekey found in record");
529 res = nmem_malloc(se->nmem, sizeof(struct record));
532 res->metadata = nmem_malloc(se->nmem,
533 sizeof(struct record_metadata*) * service->num_metadata);
534 memset(res->metadata, 0, sizeof(struct record_metadata*) * service->num_metadata);
536 mergekey_norm = nmem_strdup(se->nmem, (char*) mergekey);
538 normalize_mergekey(mergekey_norm, 0);
540 cluster = reclist_insert(se->reclist, res, mergekey_norm, &se->total_merged);
541 if (global_parameters.dump_records)
542 yaz_log(YLOG_LOG, "Cluster id %d from %s (#%d)", cluster->recid,
543 cl->database->url, cl->records);
546 /* no room for record */
550 relevance_newrec(se->relevance, cluster);
552 for (n = root->children; n; n = n->next)
560 if (n->type != XML_ELEMENT_NODE)
562 if (!strcmp(n->name, "metadata"))
564 struct conf_metadata *md = 0;
565 struct conf_sortkey *sk = 0;
566 struct record_metadata **wheretoput, *newm;
570 type = xmlGetProp(n, "type");
571 value = xmlNodeListGetString(xdoc, n->children, 0);
576 // First, find out what field we're looking at
577 for (imeta = 0; imeta < service->num_metadata; imeta++)
578 if (!strcmp(type, service->metadata[imeta].name))
580 md = &service->metadata[imeta];
581 if (md->sortkey_offset >= 0)
582 sk = &service->sortkeys[md->sortkey_offset];
587 yaz_log(YLOG_WARN, "Ignoring unknown metadata element: %s", type);
591 // Find out where we are putting it
592 if (md->merge == Metadata_merge_no)
593 wheretoput = &res->metadata[imeta];
595 wheretoput = &cluster->metadata[imeta];
598 newm = nmem_malloc(se->nmem, sizeof(struct record_metadata));
600 if (md->type == Metadata_type_generic)
603 for (p = value; *p && isspace(*p); p++)
605 for (pe = p + strlen(p) - 1;
606 pe > p && strchr(" ,/.:([", *pe); pe--)
608 newm->data.text = nmem_strdup(se->nmem, p);
611 else if (md->type == Metadata_type_year)
613 if (extract_years(value, &first, &last) < 0)
618 yaz_log(YLOG_WARN, "Unknown type in metadata element %s", type);
621 if (md->type == Metadata_type_year && md->merge != Metadata_merge_range)
623 yaz_log(YLOG_WARN, "Only range merging supported for years");
626 if (md->merge == Metadata_merge_unique)
628 struct record_metadata *mnode;
629 for (mnode = *wheretoput; mnode; mnode = mnode->next)
630 if (!strcmp(mnode->data.text, newm->data.text))
634 newm->next = *wheretoput;
638 else if (md->merge == Metadata_merge_longest)
641 strlen(newm->data.text) > strlen((*wheretoput)->data.text))
646 char *s = nmem_strdup(se->nmem, newm->data.text);
647 if (!cluster->sortkeys[md->sortkey_offset])
648 cluster->sortkeys[md->sortkey_offset] =
649 nmem_malloc(se->nmem, sizeof(union data_types));
650 normalize_mergekey(s,
651 (sk->type == Metadata_sortkey_skiparticle));
652 cluster->sortkeys[md->sortkey_offset]->text = s;
656 else if (md->merge == Metadata_merge_all || md->merge == Metadata_merge_no)
658 newm->next = *wheretoput;
661 else if (md->merge == Metadata_merge_range)
663 assert(md->type == Metadata_type_year);
667 (*wheretoput)->data.number.min = first;
668 (*wheretoput)->data.number.max = last;
670 cluster->sortkeys[md->sortkey_offset] = &newm->data;
674 if (first < (*wheretoput)->data.number.min)
675 (*wheretoput)->data.number.min = first;
676 if (last > (*wheretoput)->data.number.max)
677 (*wheretoput)->data.number.max = last;
682 union data_types *sdata = cluster->sortkeys[md->sortkey_offset];
683 yaz_log(YLOG_LOG, "SK range: %d-%d", sdata->number.min, sdata->number.max);
688 yaz_log(YLOG_WARN, "Don't know how to merge on element name %s", md->name);
691 relevance_countwords(se->relevance, cluster, value, md->rank);
694 if (md->type == Metadata_type_year)
697 sprintf(year, "%d", last);
698 add_facet(se, type, year);
701 sprintf(year, "%d", first);
702 add_facet(se, type, year);
706 add_facet(se, type, value);
713 yaz_log(YLOG_WARN, "Unexpected element %s in internal record", n->name);
722 relevance_donerecord(se->relevance, cluster);
728 static void ingest_records(struct client *cl, Z_Records *r)
731 yaz_timing_t t = yaz_timing_create();
734 struct session *s = cl->session;
735 Z_NamePlusRecordList *rlist;
738 if (r->which != Z_Records_DBOSD)
740 rlist = r->u.databaseOrSurDiagnostics;
741 for (i = 0; i < rlist->num_records; i++)
743 Z_NamePlusRecord *npr = rlist->records[i];
746 if (npr->which != Z_NamePlusRecord_databaseRecord)
748 yaz_log(YLOG_WARN, "Unexpected record type, probably diagnostic");
752 rec = ingest_record(cl, npr->u.databaseRecord);
756 if (s->watchlist[SESSION_WATCH_RECORDS].fun && rlist->num_records)
757 session_alert_watch(s, SESSION_WATCH_RECORDS);
761 yaz_log(YLOG_LOG, "ingest_records %6.5f %3.2f %3.2f",
762 yaz_timing_get_real(t), yaz_timing_get_user(t),
763 yaz_timing_get_sys(t));
764 yaz_timing_destroy(&t);
768 static void do_presentResponse(IOCHAN i, Z_APDU *a)
770 struct connection *co = iochan_getdata(i);
771 struct client *cl = co->client;
772 Z_PresentResponse *r = a->u.presentResponse;
775 Z_Records *recs = r->records;
776 if (recs->which == Z_Records_NSD)
778 yaz_log(YLOG_WARN, "Non-surrogate diagnostic");
779 cl->diagnostic = *recs->u.nonSurrogateDiagnostic->condition;
780 cl->state = Client_Error;
784 if (!*r->presentStatus && cl->state != Client_Error)
786 yaz_log(YLOG_DEBUG, "Good Present response");
787 ingest_records(cl, r->records);
788 cl->state = Client_Idle;
790 else if (*r->presentStatus)
792 yaz_log(YLOG_WARN, "Bad Present response");
793 cl->state = Client_Error;
797 static void handler(IOCHAN i, int event)
799 struct connection *co = iochan_getdata(i);
800 struct client *cl = co->client;
801 struct session *se = 0;
807 yaz_log(YLOG_WARN, "Destroying orphan connection");
808 connection_destroy(co);
812 if (co->state == Conn_Connecting && event & EVENT_OUTPUT)
815 socklen_t errlen = sizeof(errcode);
817 if (getsockopt(cs_fileno(co->link), SOL_SOCKET, SO_ERROR, &errcode,
818 &errlen) < 0 || errcode != 0)
825 yaz_log(YLOG_DEBUG, "Connect OK");
826 co->state = Conn_Open;
828 cl->state = Client_Connected;
832 else if (event & EVENT_INPUT)
834 int len = cs_get(co->link, &co->ibuf, &co->ibufsize);
838 yaz_log(YLOG_WARN|YLOG_ERRNO, "Error reading from Z server");
839 connection_destroy(co);
844 yaz_log(YLOG_WARN, "EOF reading from Z server");
845 connection_destroy(co);
848 else if (len > 1) // We discard input if we have no connection
850 co->state = Conn_Open;
852 if (cl && (cl->requestid == se->requestid || cl->state == Client_Initializing))
856 odr_reset(global_parameters.odr_in);
857 odr_setbuf(global_parameters.odr_in, co->ibuf, len, 0);
858 if (!z_APDU(global_parameters.odr_in, &a, 0, 0))
865 case Z_APDU_initResponse:
866 do_initResponse(i, a);
868 case Z_APDU_searchResponse:
869 do_searchResponse(i, a);
871 case Z_APDU_presentResponse:
872 do_presentResponse(i, a);
875 yaz_log(YLOG_WARN, "Unexpected result from server");
879 // We aren't expecting staggered output from target
880 // if (cs_more(t->link))
881 // iochan_setevent(i, EVENT_INPUT);
883 else // we throw away response and go to idle mode
885 yaz_log(YLOG_DEBUG, "Ignoring result of expired operation");
886 cl->state = Client_Idle;
889 /* if len==1 we do nothing but wait for more input */
892 if (cl->state == Client_Connected) {
896 if (cl->state == Client_Idle)
898 if (cl->requestid != se->requestid && *se->query) {
901 else if (cl->hits > 0 && cl->records < global_parameters.toget &&
902 cl->records < cl->hits) {
908 // Disassociate connection from client
909 static void connection_release(struct connection *co)
911 struct client *cl = co->client;
913 yaz_log(YLOG_DEBUG, "Connection release %s", co->host->hostport);
920 // Close connection and recycle structure
921 static void connection_destroy(struct connection *co)
923 struct host *h = co->host;
925 iochan_destroy(co->iochan);
927 yaz_log(YLOG_DEBUG, "Connection destroy %s", co->host->hostport);
928 if (h->connections == co)
929 h->connections = co->next;
932 struct connection *pco;
933 for (pco = h->connections; pco && pco->next != co; pco = pco->next)
936 pco->next = co->next;
942 if (co->client->state != Client_Idle)
943 co->client->state = Client_Disconnected;
944 co->client->connection = 0;
946 co->next = connection_freelist;
947 connection_freelist = co;
950 // Creates a new connection for client, associated with the host of
952 static struct connection *connection_create(struct client *cl)
954 struct connection *new;
960 if (!(link = cs_create(tcpip_type, 0, PROTO_Z3950)))
962 yaz_log(YLOG_FATAL|YLOG_ERRNO, "Failed to create comstack");
966 if (0 == strlen(global_parameters.zproxy_override)){
967 /* no Z39.50 proxy needed - direct connect */
968 yaz_log(YLOG_DEBUG, "Connection create %s", cl->database->url);
970 if (!(addr = cs_straddr(link, cl->database->host->ipport)))
972 yaz_log(YLOG_WARN|YLOG_ERRNO,
973 "Lookup of IP address %s failed",
974 cl->database->host->ipport);
979 /* Z39.50 proxy connect */
980 yaz_log(YLOG_DEBUG, "Connection create %s proxy %s",
981 cl->database->url, global_parameters.zproxy_override);
983 yaz_log(YLOG_LOG, "Connection cs_create_host %s proxy %s",
984 cl->database->url, global_parameters.zproxy_override);
986 if (!(addr = cs_straddr(link, global_parameters.zproxy_override)))
988 yaz_log(YLOG_WARN|YLOG_ERRNO,
989 "Lookup of IP address %s failed",
990 global_parameters.zproxy_override);
995 res = cs_connect(link, addr);
998 yaz_log(YLOG_WARN|YLOG_ERRNO, "cs_connect %s", cl->database->url);
1002 if ((new = connection_freelist))
1003 connection_freelist = new->next;
1006 new = xmalloc(sizeof (struct connection));
1010 new->state = Conn_Connecting;
1011 new->host = cl->database->host;
1012 new->next = new->host->connections;
1013 new->host->connections = new;
1015 cl->connection = new;
1018 new->iochan = iochan_create(cs_fileno(link), handler, 0);
1019 iochan_setdata(new->iochan, new);
1020 new->iochan->next = channel_list;
1021 channel_list = new->iochan;
1025 // Close connection and set state to error
1026 static void client_fatal(struct client *cl)
1028 yaz_log(YLOG_WARN, "Fatal error from %s", cl->database->url);
1029 connection_destroy(cl->connection);
1030 cl->state = Client_Error;
1033 // Ensure that client has a connection associated
1034 static int client_prep_connection(struct client *cl)
1036 struct connection *co;
1037 struct session *se = cl->session;
1038 struct host *host = cl->database->host;
1040 co = cl->connection;
1042 yaz_log(YLOG_DEBUG, "Client prep %s", cl->database->url);
1046 // See if someone else has an idle connection
1047 // We should look at timestamps here to select the longest-idle connection
1048 for (co = host->connections; co; co = co->next)
1049 if (co->state == Conn_Open && (!co->client || co->client->session != se))
1053 connection_release(co);
1054 cl->connection = co;
1058 co = connection_create(cl);
1062 if (co->state == Conn_Connecting)
1064 cl->state = Client_Connecting;
1065 iochan_setflag(co->iochan, EVENT_OUTPUT);
1067 else if (co->state == Conn_Open)
1069 if (cl->state == Client_Error || cl->state == Client_Disconnected)
1070 cl->state = Client_Idle;
1071 iochan_setflag(co->iochan, EVENT_OUTPUT);
1079 #ifdef GAGA // Moved to database.c
1081 // This function will most likely vanish when a proper target profile mechanism is
1083 void load_simpletargets(const char *fn)
1085 FILE *f = fopen(fn, "r");
1090 yaz_log(YLOG_WARN|YLOG_ERRNO, "open %s", fn);
1094 while (fgets(line, 255, f))
1099 struct database *database;
1101 if (strncmp(line, "target ", 7))
1103 line[strlen(line) - 1] = '\0';
1105 if ((name = strchr(line, ';')))
1109 if ((db = strchr(url, '/')))
1114 yaz_log(YLOG_LOG, "Target: %s, '%s'", url, db);
1115 for (host = hosts; host; host = host->next)
1116 if (!strcmp(url, host->hostport))
1120 struct addrinfo *addrinfo, hints;
1123 unsigned char addrbuf[4];
1126 host = xmalloc(sizeof(struct host));
1127 host->hostport = xstrdup(url);
1128 host->connections = 0;
1130 if ((port = strchr(url, ':')))
1136 hints.ai_family = PF_INET;
1137 hints.ai_socktype = SOCK_STREAM;
1138 hints.ai_protocol = IPPROTO_TCP;
1139 hints.ai_addrlen = 0;
1141 hints.ai_canonname = 0;
1143 // This is not robust code. It assumes that getaddrinfo returns AF_INET
1145 if ((res = getaddrinfo(url, port, &hints, &addrinfo)))
1147 yaz_log(YLOG_WARN, "Failed to resolve %s: %s", url, gai_strerror(res));
1148 xfree(host->hostport);
1152 assert(addrinfo->ai_family == PF_INET);
1153 memcpy(addrbuf, &((struct sockaddr_in*)addrinfo->ai_addr)->sin_addr.s_addr, 4);
1154 sprintf(ipport, "%u.%u.%u.%u:%s",
1155 addrbuf[0], addrbuf[1], addrbuf[2], addrbuf[3], port);
1156 host->ipport = xstrdup(ipport);
1157 freeaddrinfo(addrinfo);
1161 database = xmalloc(sizeof(struct database));
1162 database->host = host;
1163 database->url = xmalloc(strlen(url) + strlen(db) + 2);
1164 strcpy(database->url, url);
1165 strcat(database->url, "/");
1166 strcat(database->url, db);
1168 database->name = xstrdup(name);
1172 database->databases = xmalloc(2 * sizeof(char *));
1173 database->databases[0] = xstrdup(db);
1174 database->databases[1] = 0;
1175 database->errors = 0;
1176 database->qprofile = 0;
1177 database->rprofile = database_retrieval_profile(database);
1178 database->next = databases;
1179 databases = database;
1187 static void pull_terms(NMEM nmem, struct ccl_rpn_node *n, char **termlist, int *num)
1195 pull_terms(nmem, n->u.p[0], termlist, num);
1196 pull_terms(nmem, n->u.p[1], termlist, num);
1199 termlist[(*num)++] = nmem_strdup(nmem, n->u.t.term);
1206 // Extract terms from query into null-terminated termlist
1207 static int extract_terms(NMEM nmem, char *query, char **termlist)
1210 struct ccl_rpn_node *n;
1213 n = ccl_find_str(global_parameters.ccl_filter, query, &error, &pos);
1216 pull_terms(nmem, n, termlist, &num);
1222 static struct client *client_create(void)
1225 if (client_freelist)
1227 r = client_freelist;
1228 client_freelist = client_freelist->next;
1231 r = xmalloc(sizeof(struct client));
1240 r->state = Client_Disconnected;
1245 void client_destroy(struct client *c)
1247 struct session *se = c->session;
1248 if (c == se->clients)
1249 se->clients = c->next;
1253 for (cc = se->clients; cc && cc->next != c; cc = cc->next)
1259 connection_release(c->connection);
1260 c->next = client_freelist;
1261 client_freelist = c;
1264 void session_set_watch(struct session *s, int what, session_watchfun fun, void *data)
1266 s->watchlist[what].fun = fun;
1267 s->watchlist[what].data = data;
1270 void session_alert_watch(struct session *s, int what)
1272 if (!s->watchlist[what].fun)
1274 (*s->watchlist[what].fun)(s->watchlist[what].data);
1275 s->watchlist[what].fun = 0;
1276 s->watchlist[what].data = 0;
1279 //callback for grep_databases
1280 static void select_targets_callback(void *context, struct database *db)
1282 struct session *se = (struct session*) context;
1283 struct client *cl = client_create();
1286 cl->next = se->clients;
1290 // This should be extended with parameters to control selection criteria
1291 // Associates a set of clients with a session;
1292 int select_targets(struct session *se, struct database_criterion *crit)
1295 client_destroy(se->clients);
1297 return grep_databases(se, crit, select_targets_callback);
1300 int session_active_clients(struct session *s)
1305 for (c = s->clients; c; c = c->next)
1306 if (c->connection && (c->state == Client_Connecting ||
1307 c->state == Client_Initializing ||
1308 c->state == Client_Searching ||
1309 c->state == Client_Presenting))
1315 // parses crit1=val1,crit2=val2|val3,...
1316 static struct database_criterion *parse_filter(NMEM m, const char *buf)
1318 struct database_criterion *res = 0;
1325 nmem_strsplit(m, ",", buf, &values, &num);
1326 for (i = 0; i < num; i++)
1331 struct database_criterion *new = nmem_malloc(m, sizeof(*new));
1332 char *eq = strchr(values[i], '=');
1335 yaz_log(YLOG_WARN, "Missing equal-sign in filter");
1339 new->name = values[i];
1340 nmem_strsplit(m, "|", eq, &subvalues, &subnum);
1342 for (subi = 0; subi < subnum; subi++)
1344 struct database_criterion_value *newv = nmem_malloc(m, sizeof(*newv));
1345 newv->value = subvalues[subi];
1346 newv->next = new->values;
1355 char *search(struct session *se, char *query, char *filter)
1357 int live_channels = 0;
1359 struct database_criterion *criteria;
1361 yaz_log(YLOG_DEBUG, "Search");
1363 nmem_reset(se->nmem);
1364 criteria = parse_filter(se->nmem, filter);
1365 strcpy(se->query, query);
1367 // Release any existing clients
1368 select_targets(se, criteria);
1369 for (cl = se->clients; cl; cl = cl->next)
1371 if (client_prep_connection(cl))
1377 int maxrecs = live_channels * global_parameters.toget;
1378 se->num_termlists = 0;
1379 se->reclist = reclist_create(se->nmem, maxrecs);
1380 extract_terms(se->nmem, query, p);
1381 se->relevance = relevance_create(se->nmem, (const char **) p, maxrecs);
1382 se->total_records = se->total_hits = se->total_merged = 0;
1383 se->expected_maxrecs = maxrecs;
1391 void destroy_session(struct session *s)
1393 yaz_log(YLOG_LOG, "Destroying session");
1395 client_destroy(s->clients);
1396 nmem_destroy(s->nmem);
1397 wrbuf_destroy(s->wrbuf);
1400 struct session *new_session()
1403 struct session *session = xmalloc(sizeof(*session));
1405 yaz_log(YLOG_DEBUG, "New pazpar2 session");
1407 session->total_hits = 0;
1408 session->total_records = 0;
1409 session->num_termlists = 0;
1410 session->reclist = 0;
1411 session->requestid = -1;
1412 session->clients = 0;
1413 session->expected_maxrecs = 0;
1414 session->query[0] = '\0';
1415 session->nmem = nmem_create();
1416 session->wrbuf = wrbuf_alloc();
1417 for (i = 0; i <= SESSION_WATCH_MAX; i++)
1419 session->watchlist[i].data = 0;
1420 session->watchlist[i].fun = 0;
1426 struct hitsbytarget *hitsbytarget(struct session *se, int *count)
1428 static struct hitsbytarget res[1000]; // FIXME MM
1432 for (cl = se->clients; cl; cl = cl->next)
1434 res[*count].id = cl->database->url;
1435 res[*count].name = cl->database->name;
1436 res[*count].hits = cl->hits;
1437 res[*count].records = cl->records;
1438 res[*count].diagnostic = cl->diagnostic;
1439 res[*count].state = client_states[cl->state];
1440 res[*count].connected = cl->connection ? 1 : 0;
1447 struct termlist_score **termlist(struct session *s, const char *name, int *num)
1451 for (i = 0; i < s->num_termlists; i++)
1452 if (!strcmp(s->termlists[i].name, name))
1453 return termlist_highscore(s->termlists[i].termlist, num);
1457 #ifdef MISSING_HEADERS
1458 void report_nmem_stats(void)
1460 size_t in_use, is_free;
1462 nmem_get_memory_in_use(&in_use);
1463 nmem_get_memory_free(&is_free);
1465 yaz_log(YLOG_LOG, "nmem stat: use=%ld free=%ld",
1466 (long) in_use, (long) is_free);
1470 struct record_cluster *show_single(struct session *s, int id)
1472 struct record_cluster *r;
1474 reclist_rewind(s->reclist);
1475 while ((r = reclist_read_record(s->reclist)))
1481 struct record_cluster **show(struct session *s, struct reclist_sortparms *sp, int start,
1482 int *num, int *total, int *sumhits, NMEM nmem_show)
1484 struct record_cluster **recs = nmem_malloc(nmem_show, *num
1485 * sizeof(struct record_cluster *));
1486 struct reclist_sortparms *spp;
1489 yaz_timing_t t = yaz_timing_create();
1492 for (spp = sp; spp; spp = spp->next)
1493 if (spp->type == Metadata_sortkey_relevance)
1495 relevance_prepare_read(s->relevance, s->reclist);
1498 reclist_sort(s->reclist, sp);
1500 *total = s->reclist->num_records;
1501 *sumhits = s->total_hits;
1503 for (i = 0; i < start; i++)
1504 if (!reclist_read_record(s->reclist))
1511 for (i = 0; i < *num; i++)
1513 struct record_cluster *r = reclist_read_record(s->reclist);
1523 yaz_log(YLOG_LOG, "show %6.5f %3.2f %3.2f",
1524 yaz_timing_get_real(t), yaz_timing_get_user(t),
1525 yaz_timing_get_sys(t));
1526 yaz_timing_destroy(&t);
1531 void statistics(struct session *se, struct statistics *stat)
1536 memset(stat, 0, sizeof(*stat));
1537 for (cl = se->clients; cl; cl = cl->next)
1539 if (!cl->connection)
1540 stat->num_no_connection++;
1543 case Client_Connecting: stat->num_connecting++; break;
1544 case Client_Initializing: stat->num_initializing++; break;
1545 case Client_Searching: stat->num_searching++; break;
1546 case Client_Presenting: stat->num_presenting++; break;
1547 case Client_Idle: stat->num_idle++; break;
1548 case Client_Failed: stat->num_failed++; break;
1549 case Client_Error: stat->num_error++; break;
1554 stat->num_hits = se->total_hits;
1555 stat->num_records = se->total_records;
1557 stat->num_clients = count;
1560 static CCL_bibset load_cclfile(const char *fn)
1562 CCL_bibset res = ccl_qual_mk();
1563 if (ccl_qual_fname(res, fn) < 0)
1565 yaz_log(YLOG_FATAL|YLOG_ERRNO, "%s", fn);
1571 static void start_http_listener(void)
1574 struct conf_server *ser = global_parameters.server;
1576 if (*global_parameters.listener_override)
1577 strcpy(hp, global_parameters.listener_override);
1580 strcpy(hp, ser->host ? ser->host : "");
1585 sprintf(hp + strlen(hp), "%d", ser->port);
1591 static void start_proxy(void)
1594 struct conf_server *ser = global_parameters.server;
1596 if (*global_parameters.proxy_override)
1597 strcpy(hp, global_parameters.proxy_override);
1598 else if (ser->proxy_host || ser->proxy_port)
1600 strcpy(hp, ser->proxy_host ? ser->proxy_host : "");
1601 if (ser->proxy_port)
1605 sprintf(hp + strlen(hp), "%d", ser->proxy_port);
1611 http_set_proxyaddr(hp, ser->myurl ? ser->myurl : "");
1614 int main(int argc, char **argv)
1619 if (signal(SIGPIPE, SIG_IGN) == SIG_ERR)
1620 yaz_log(YLOG_WARN|YLOG_ERRNO, "signal");
1622 yaz_log_init(YLOG_DEFAULT_LEVEL, "pazpar2", 0);
1624 while ((ret = options("f:x:h:p:z:C:s:d", argv, argc, &arg)) != -2)
1628 if (!read_config(arg))
1632 strcpy(global_parameters.listener_override, arg);
1635 global_parameters.ccl_filter = load_cclfile(arg);
1638 strcpy(global_parameters.proxy_override, arg);
1641 strcpy(global_parameters.zproxy_override, arg);
1644 load_simpletargets(arg);
1647 global_parameters.dump_records = 1;
1650 fprintf(stderr, "Usage: pazpar2\n"
1652 " -h [host:]port (REST protocol listener)\n"
1654 " -s simpletargetfile\n"
1655 " -p hostname[:portno] (HTTP proxy)\n"
1656 " -z hostname[:portno] (Z39.50 proxy)\n"
1657 " -d (show internal records)\n");
1664 yaz_log(YLOG_FATAL, "Load config with -f");
1667 global_parameters.server = config->servers;
1669 start_http_listener();
1671 if (!global_parameters.ccl_filter)
1672 global_parameters.ccl_filter = load_cclfile("../etc/default.bib");
1673 global_parameters.yaz_marc = yaz_marc_create();
1674 yaz_marc_subfield_str(global_parameters.yaz_marc, "\t");
1675 global_parameters.odr_in = odr_createmem(ODR_DECODE);
1676 global_parameters.odr_out = odr_createmem(ODR_ENCODE);
1678 event_loop(&channel_list);
1686 * indent-tabs-mode: nil
1688 * vim: shiftwidth=4 tabstop=8 expandtab