X-Git-Url: http://git.indexdata.com/?a=blobdiff_plain;f=src%2Fpazpar2.c;h=8b2638bb75b47ac9e5d99b573a56f20f23b0bc29;hb=dff850d3cf2c6f70ce1725b6d4d510bbb37b08d2;hp=1f23d6cbce26c8b7a7cbb22f881b9dfe2e35e3e1;hpb=df1fd6502c261c3939c5799b23cbb1bf66c13c0c;p=pazpar2-moved-to-github.git diff --git a/src/pazpar2.c b/src/pazpar2.c index 1f23d6c..8b2638b 100644 --- a/src/pazpar2.c +++ b/src/pazpar2.c @@ -1,4 +1,4 @@ -/* $Id: pazpar2.c,v 1.35 2007-01-16 03:38:49 quinn Exp $ */ +/* $Id: pazpar2.c,v 1.63 2007-04-04 22:43:10 marc Exp $ */ #include #include @@ -17,6 +17,7 @@ #include #include #include +#include #include #include @@ -38,6 +39,8 @@ #include "reclists.h" #include "relevance.h" #include "config.h" +#include "database.h" +#include "settings.h" #define MAX_CHUNK 15 @@ -45,17 +48,15 @@ static void client_fatal(struct client *cl); static void connection_destroy(struct connection *co); static int client_prep_connection(struct client *cl); static void ingest_records(struct client *cl, Z_Records *r); -static struct conf_retrievalprofile *database_retrieval_profile(struct database *db); +//static struct conf_retrievalprofile *database_retrieval_profile(struct database *db); void session_alert_watch(struct session *s, int what); +char *session_setting_oneval(struct session *s, struct database *db, const char *name); IOCHAN channel_list = 0; // Master list of connections we're handling events to static struct connection *connection_freelist = 0; static struct client *client_freelist = 0; -static struct host *hosts = 0; // The hosts we know about -static struct database *databases = 0; // The databases we know about - static char *client_states[] = { "Client_Connecting", "Client_Connected", @@ -74,6 +75,8 @@ struct parameters global_parameters = { "", "", + "", + "", 0, 0, 30, @@ -86,7 +89,6 @@ struct parameters global_parameters = MAX_CHUNK, 0, 0, - 0, 0 }; @@ -121,6 +123,7 @@ static int send_apdu(struct client *c, Z_APDU *a) static void send_init(IOCHAN i) { + struct connection *co = iochan_getdata(i); struct client *cl = co->client; Z_APDU *a = zget_APDU(global_parameters.odr_out, Z_APDU_initRequest); @@ -136,6 +139,18 @@ static void send_init(IOCHAN i) ODR_MASK_SET(a->u.initRequest->protocolVersion, Z_ProtocolVersion_1); ODR_MASK_SET(a->u.initRequest->protocolVersion, Z_ProtocolVersion_2); ODR_MASK_SET(a->u.initRequest->protocolVersion, Z_ProtocolVersion_3); + + + /* add virtual host if tunneling through Z39.50 proxy */ + + if (0 < strlen(global_parameters.zproxy_override) + && 0 < strlen(cl->database->url)) + yaz_oi_set_string_oidval(&a->u.initRequest->otherInfo, + global_parameters.odr_out, VAL_PROXY, + 1, cl->database->url); + + + if (send_apdu(cl, a) >= 0) { iochan_setflags(i, EVENT_INPUT); @@ -146,6 +161,34 @@ static void send_init(IOCHAN i) odr_reset(global_parameters.odr_out); } +static void pull_terms(NMEM nmem, struct ccl_rpn_node *n, char **termlist, int *num) +{ + switch (n->kind) + { + case CCL_RPN_AND: + case CCL_RPN_OR: + case CCL_RPN_NOT: + case CCL_RPN_PROX: + pull_terms(nmem, n->u.p[0], termlist, num); + pull_terms(nmem, n->u.p[1], termlist, num); + break; + case CCL_RPN_TERM: + termlist[(*num)++] = nmem_strdup(nmem, n->u.t.term); + break; + default: // NOOP + break; + } +} + +// Extract terms from query into null-terminated termlist +static void extract_terms(NMEM nmem, struct ccl_rpn_node *query, char **termlist) +{ + int num = 0; + + pull_terms(nmem, query, termlist, &num); + termlist[num] = 0; +} + static void send_search(IOCHAN i) { struct connection *co = iochan_getdata(i); @@ -158,12 +201,24 @@ static void send_search(IOCHAN i) Z_Query *zquery; struct ccl_rpn_node *cn; int ssub = 0, lslb = 100000, mspn = 10; + char *recsyn; + char *piggyback; yaz_log(YLOG_DEBUG, "Sending search"); - cn = ccl_find_str(global_parameters.ccl_filter, se->query, &cerror, &cpos); + cn = ccl_find_str(db->ccl_map, se->query, &cerror, &cpos); if (!cn) return; + + if (!se->relevance) + { + // Initialize relevance structure with query terms + char *p[512]; + extract_terms(se->nmem, cn, p); + se->relevance = relevance_create(se->nmem, (const char **) p, + se->expected_maxrecs); + } + a->u.searchRequest->query = zquery = odr_malloc(global_parameters.odr_out, sizeof(Z_Query)); zquery->which = Z_Query_type_1; @@ -176,12 +231,16 @@ static void send_search(IOCHAN i) for (ndb = 0; db->databases[ndb]; ndb++) databaselist[ndb] = db->databases[ndb]; - a->u.presentRequest->preferredRecordSyntax = - yaz_oidval_to_z3950oid(global_parameters.odr_out, - CLASS_RECSYN, VAL_USMARC); - a->u.searchRequest->smallSetUpperBound = &ssub; - a->u.searchRequest->largeSetLowerBound = &lslb; - a->u.searchRequest->mediumSetPresentNumber = &mspn; + if (!(piggyback = session_setting_oneval(se, db, "pz:piggyback")) || *piggyback == '1') + { + if ((recsyn = session_setting_oneval(se, db, "pz:syntax"))) + a->u.searchRequest->preferredRecordSyntax = + yaz_str_to_z3950oid(global_parameters.odr_out, + CLASS_RECSYN, recsyn); + a->u.searchRequest->smallSetUpperBound = &ssub; + a->u.searchRequest->largeSetLowerBound = &lslb; + a->u.searchRequest->mediumSetPresentNumber = &mspn; + } a->u.searchRequest->resultSetName = "Default"; a->u.searchRequest->databaseNames = databaselist; a->u.searchRequest->num_databaseNames = ndb; @@ -202,9 +261,12 @@ static void send_present(IOCHAN i) { struct connection *co = iochan_getdata(i); struct client *cl = co->client; + struct session *se = cl->session; + struct database *db = cl->database; Z_APDU *a = zget_APDU(global_parameters.odr_out, Z_APDU_presentRequest); int toget; int start = cl->records + 1; + char *recsyn; toget = global_parameters.chunk; if (toget > global_parameters.toget - cl->records) @@ -219,9 +281,10 @@ static void send_present(IOCHAN i) a->u.presentRequest->resultSetId = "Default"; - a->u.presentRequest->preferredRecordSyntax = - yaz_oidval_to_z3950oid(global_parameters.odr_out, - CLASS_RECSYN, VAL_USMARC); + if ((recsyn = session_setting_oneval(se, db, "pz:syntax"))) + a->u.presentRequest->preferredRecordSyntax = + yaz_str_to_z3950oid(global_parameters.odr_out, + CLASS_RECSYN, recsyn); if (send_apdu(cl, a) >= 0) { @@ -239,7 +302,7 @@ static void do_initResponse(IOCHAN i, Z_APDU *a) struct client *cl = co->client; Z_InitResponse *r = a->u.initResponse; - yaz_log(YLOG_DEBUG, "Received init response"); + yaz_log(YLOG_DEBUG, "Init response %s", cl->database->url); if (*r->result) { @@ -256,7 +319,8 @@ static void do_searchResponse(IOCHAN i, Z_APDU *a) struct session *se = cl->session; Z_SearchResponse *r = a->u.searchResponse; - yaz_log(YLOG_DEBUG, "Searchresponse (status=%d)", *r->searchStatus); + yaz_log(YLOG_DEBUG, "Search response %s (status=%d)", + cl->database->url, *r->searchStatus); if (*r->searchStatus) { @@ -264,8 +328,8 @@ static void do_searchResponse(IOCHAN i, Z_APDU *a) se->total_hits += cl->hits; if (r->presentStatus && !*r->presentStatus && r->records) { - yaz_log(YLOG_DEBUG, "Records in search response"); - cl->records += *r->numberOfRecordsReturned; + yaz_log(YLOG_DEBUG, "Records in search response %s", + cl->database->url); ingest_records(cl, r->records); } cl->state = Client_Idle; @@ -278,7 +342,9 @@ static void do_searchResponse(IOCHAN i, Z_APDU *a) Z_Records *recs = r->records; if (recs->which == Z_Records_NSD) { - yaz_log(YLOG_WARN, "Non-surrogate diagnostic"); + yaz_log(YLOG_WARN, + "Search response: Non-surrogate diagnostic %s", + cl->database->url); cl->diagnostic = *recs->u.nonSurrogateDiagnostic->condition; cl->state = Client_Error; } @@ -286,6 +352,19 @@ static void do_searchResponse(IOCHAN i, Z_APDU *a) } } +static void do_closeResponse(IOCHAN i, Z_APDU *a) +{ + struct connection *co = iochan_getdata(i); + struct client *cl = co->client; + /* Z_Close *r = a->u.close; */ + + yaz_log(YLOG_WARN, "Close response %s", cl->database->url); + + cl->state = Client_Failed; + connection_destroy(co); +} + + char *normalize_mergekey(char *buf, int skiparticle) { char *p = buf, *pout = buf; @@ -327,42 +406,6 @@ char *normalize_mergekey(char *buf, int skiparticle) return buf; } - -#ifdef GAGA -// FIXME needs to be generalized. Should flexibly generate X lists per search -static void extract_subject(struct session *s, const char *rec) -{ - const char *field, *subfield; - - while ((field = find_field(rec, "650"))) - { - rec = field; - if ((subfield = find_subfield(field, 'a'))) - { - char *e, *ef; - char buf[1024]; - int len; - - ef = index(subfield, '\n'); - if (!ef) - return; - if ((e = index(subfield, '\t')) && e < ef) - ef = e; - while (ef > subfield && !isalpha(*(ef - 1)) && *(ef - 1) != ')') - ef--; - len = ef - subfield; - assert(len < 1023); - memcpy(buf, subfield, len); - buf[len] = '\0'; -#ifdef FIXME - if (*buf) - termlist_insert(s->termlist, buf); -#endif - } - } -} -#endif - static void add_facet(struct session *s, const char *type, const char *value) { int i; @@ -386,8 +429,6 @@ static void add_facet(struct session *s, const char *type, const char *value) termlist_insert(s->termlists[i].termlist, value); } -int yaz_marc_write_xml(); - static xmlDoc *normalize_record(struct client *cl, Z_External *rec) { struct conf_retrievalprofile *rprofile = cl->database->rprofile; @@ -402,23 +443,26 @@ static xmlDoc *normalize_record(struct client *cl, Z_External *rec) int len; if (rec->which != Z_External_octet) { - yaz_log(YLOG_WARN, "Unexpected external branch, probably BER"); + yaz_log(YLOG_WARN, "Unexpected external branch, probably BER %s", + cl->database->url); return 0; } buf = (char*) rec->u.octet_aligned->buf; len = rec->u.octet_aligned->len; if (yaz_marc_read_iso2709(rprofile->yaz_marc, buf, len) < 0) { - yaz_log(YLOG_WARN, "Failed to decode MARC"); + yaz_log(YLOG_WARN, "Failed to decode MARC %s", + cl->database->url); return 0; } if (yaz_marc_write_xml(rprofile->yaz_marc, &res, "http://www.loc.gov/MARC21/slim", 0, 0) < 0) { - yaz_log(YLOG_WARN, "Failed to encode as XML"); + yaz_log(YLOG_WARN, "Failed to encode as XML %s", + cl->database->url); return 0; } - rdoc = xmlNewDoc("1.0"); + rdoc = xmlNewDoc((xmlChar *) "1.0"); xmlDocSetRootElement(rdoc, res); } else @@ -426,6 +470,17 @@ static xmlDoc *normalize_record(struct client *cl, Z_External *rec) yaz_log(YLOG_FATAL, "Unknown native_syntax in normalize_record"); exit(1); } + + if (global_parameters.dump_records) + { + fprintf(stderr, "Input Record (normalized):\n----------------\n"); +#if LIBXML_VERSION >= 20600 + xmlDocFormatDump(stderr, rdoc, 1); +#else + xmlDocDump(stderr, rdoc); +#endif + } + for (m = rprofile->maplist; m; m = m->next) { xmlDoc *new; @@ -491,15 +546,15 @@ static struct record *ingest_record(struct client *cl, Z_External *rec) struct record_cluster *cluster; struct session *se = cl->session; xmlChar *mergekey, *mergekey_norm; - xmlChar *type; - xmlChar *value; + xmlChar *type = 0; + xmlChar *value = 0; struct conf_service *service = global_parameters.server->service; if (!xdoc) return 0; root = xmlDocGetRootElement(xdoc); - if (!(mergekey = xmlGetProp(root, "mergekey"))) + if (!(mergekey = xmlGetProp(root, (xmlChar *) "mergekey"))) { yaz_log(YLOG_WARN, "No mergekey found in record"); xmlFreeDoc(xdoc); @@ -508,15 +563,20 @@ static struct record *ingest_record(struct client *cl, Z_External *rec) res = nmem_malloc(se->nmem, sizeof(struct record)); res->next = 0; + res->client = cl; res->metadata = nmem_malloc(se->nmem, sizeof(struct record_metadata*) * service->num_metadata); memset(res->metadata, 0, sizeof(struct record_metadata*) * service->num_metadata); - mergekey_norm = nmem_strdup(se->nmem, (char*) mergekey); + mergekey_norm = (xmlChar *) nmem_strdup(se->nmem, (char*) mergekey); xmlFree(mergekey); - normalize_mergekey(mergekey_norm, 0); + normalize_mergekey((char *) mergekey_norm, 0); - cluster = reclist_insert(se->reclist, res, mergekey_norm, &se->total_merged); + cluster = reclist_insert(se->reclist, res, (char *) mergekey_norm, + &se->total_merged); + if (global_parameters.dump_records) + yaz_log(YLOG_LOG, "Cluster id %d from %s (#%d)", cluster->recid, + cl->database->url, cl->records); if (!cluster) { /* no room for record */ @@ -525,7 +585,6 @@ static struct record *ingest_record(struct client *cl, Z_External *rec) } relevance_newrec(se->relevance, cluster); - type = value = 0; for (n = root->children; n; n = n->next) { if (type) @@ -536,7 +595,7 @@ static struct record *ingest_record(struct client *cl, Z_External *rec) if (n->type != XML_ELEMENT_NODE) continue; - if (!strcmp(n->name, "metadata")) + if (!strcmp((const char *) n->name, "metadata")) { struct conf_metadata *md = 0; struct conf_sortkey *sk = 0; @@ -544,11 +603,15 @@ static struct record *ingest_record(struct client *cl, Z_External *rec) int imeta; int first, last; - type = xmlGetProp(n, "type"); + type = xmlGetProp(n, (xmlChar *) "type"); value = xmlNodeListGetString(xdoc, n->children, 0); + + if (!type || !value) + continue; + // First, find out what field we're looking at for (imeta = 0; imeta < service->num_metadata; imeta++) - if (!strcmp(type, service->metadata[imeta].name)) + if (!strcmp((const char *) type, service->metadata[imeta].name)) { md = &service->metadata[imeta]; if (md->sortkey_offset >= 0) @@ -572,16 +635,18 @@ static struct record *ingest_record(struct client *cl, Z_External *rec) newm->next = 0; if (md->type == Metadata_type_generic) { - char *p; - newm->data.text = nmem_strdup(se->nmem, value); - for (p = newm->data.text + strlen(newm->data.text) - 1; - p > newm->data.text && strchr(" ,/.", *p); p--) - *p = '\0'; + char *p, *pe; + for (p = (char *) value; *p && isspace(*p); p++) + ; + for (pe = p + strlen(p) - 1; + pe > p && strchr(" ,/.:([", *pe); pe--) + *pe = '\0'; + newm->data.text = nmem_strdup(se->nmem, p); } else if (md->type == Metadata_type_year) { - if (extract_years(value, &first, &last) < 0) + if (extract_years((char *) value, &first, &last) < 0) continue; } else @@ -598,7 +663,7 @@ static struct record *ingest_record(struct client *cl, Z_External *rec) { struct record_metadata *mnode; for (mnode = *wheretoput; mnode; mnode = mnode->next) - if (!strcmp(mnode->data.text, newm->data.text)) + if (!strcmp((const char *) mnode->data.text, newm->data.text)) break; if (!mnode) { @@ -646,12 +711,6 @@ static struct record *ingest_record(struct client *cl, Z_External *rec) (*wheretoput)->data.number.min = first; if (last > (*wheretoput)->data.number.max) (*wheretoput)->data.number.max = last; - if (sk) - { - union data_types *sdata = cluster->sortkeys[md->sortkey_offset]; - sdata->number.min = first; - sdata->number.max = last; - } } #ifdef GAGA if (sk) @@ -665,9 +724,24 @@ static struct record *ingest_record(struct client *cl, Z_External *rec) yaz_log(YLOG_WARN, "Don't know how to merge on element name %s", md->name); if (md->rank) - relevance_countwords(se->relevance, cluster, value, md->rank); + relevance_countwords(se->relevance, cluster, + (char *) value, md->rank); if (md->termlist) - add_facet(se, type, value); + { + if (md->type == Metadata_type_year) + { + char year[64]; + sprintf(year, "%d", last); + add_facet(se, (char *) type, year); + if (first != last) + { + sprintf(year, "%d", first); + add_facet(se, (char *) type, year); + } + } + else + add_facet(se, (char *) type, (char *) value); + } xmlFree(type); xmlFree(value); type = value = 0; @@ -675,6 +749,10 @@ static struct record *ingest_record(struct client *cl, Z_External *rec) else yaz_log(YLOG_WARN, "Unexpected element %s in internal record", n->name); } + if (type) + xmlFree(type); + if (value) + xmlFree(value); xmlFreeDoc(xdoc); @@ -684,6 +762,19 @@ static struct record *ingest_record(struct client *cl, Z_External *rec) return res; } +// Retrieve first defined value for 'name' for given database. +// Will be extended to take into account user associated with session +char *session_setting_oneval(struct session *s, struct database *db, const char *name) +{ + int offset = settings_offset(name); + + if (offset < 0) + return 0; + if (!db->settings[offset]) + return 0; + return db->settings[offset]->value; +} + static void ingest_records(struct client *cl, Z_Records *r) { #if USE_TIMING @@ -701,9 +792,12 @@ static void ingest_records(struct client *cl, Z_Records *r) { Z_NamePlusRecord *npr = rlist->records[i]; + cl->records++; if (npr->which != Z_NamePlusRecord_databaseRecord) { - yaz_log(YLOG_WARN, "Unexpected record type, probably diagnostic"); + yaz_log(YLOG_WARN, + "Unexpected record type, probably diagnostic %s", + cl->database->url); continue; } @@ -733,7 +827,8 @@ static void do_presentResponse(IOCHAN i, Z_APDU *a) Z_Records *recs = r->records; if (recs->which == Z_Records_NSD) { - yaz_log(YLOG_WARN, "Non-surrogate diagnostic"); + yaz_log(YLOG_WARN, "Non-surrogate diagnostic %s", + cl->database->url); cl->diagnostic = *recs->u.nonSurrogateDiagnostic->condition; cl->state = Client_Error; } @@ -741,14 +836,15 @@ static void do_presentResponse(IOCHAN i, Z_APDU *a) if (!*r->presentStatus && cl->state != Client_Error) { - yaz_log(YLOG_DEBUG, "Good Present response"); - cl->records += *r->numberOfRecordsReturned; + yaz_log(YLOG_DEBUG, "Good Present response %s", + cl->database->url); ingest_records(cl, r->records); cl->state = Client_Idle; } else if (*r->presentStatus) { - yaz_log(YLOG_WARN, "Bad Present response"); + yaz_log(YLOG_WARN, "Bad Present response %s", + cl->database->url); cl->state = Client_Error; } } @@ -794,13 +890,14 @@ static void handler(IOCHAN i, int event) if (len < 0) { - yaz_log(YLOG_WARN|YLOG_ERRNO, "Error reading from Z server"); + yaz_log(YLOG_WARN|YLOG_ERRNO, "Error reading from %s", + cl->database->url); connection_destroy(co); return; } else if (len == 0) { - yaz_log(YLOG_WARN, "EOF reading from Z server"); + yaz_log(YLOG_WARN, "EOF reading from %s", cl->database->url); connection_destroy(co); return; } @@ -830,8 +927,13 @@ static void handler(IOCHAN i, int event) case Z_APDU_presentResponse: do_presentResponse(i, a); break; + case Z_APDU_close: + do_closeResponse(i, a); + break; default: - yaz_log(YLOG_WARN, "Unexpected result from server"); + yaz_log(YLOG_WARN, + "Unexpected Z39.50 response from %s", + cl->database->url); client_fatal(cl); return; } @@ -915,18 +1017,37 @@ static struct connection *connection_create(struct client *cl) int res; void *addr; - yaz_log(YLOG_DEBUG, "Connection create %s", cl->database->url); + if (!(link = cs_create(tcpip_type, 0, PROTO_Z3950))) - { - yaz_log(YLOG_FATAL|YLOG_ERRNO, "Failed to create comstack"); - exit(1); - } + { + yaz_log(YLOG_FATAL|YLOG_ERRNO, "Failed to create comstack"); + exit(1); + } + + if (0 == strlen(global_parameters.zproxy_override)){ + /* no Z39.50 proxy needed - direct connect */ + yaz_log(YLOG_DEBUG, "Connection create %s", cl->database->url); + + if (!(addr = cs_straddr(link, cl->database->host->ipport))) + { + yaz_log(YLOG_WARN|YLOG_ERRNO, + "Lookup of IP address %s failed", + cl->database->host->ipport); + return 0; + } + + } else { + /* Z39.50 proxy connect */ + yaz_log(YLOG_DEBUG, "Connection create %s proxy %s", + cl->database->url, global_parameters.zproxy_override); - if (!(addr = cs_straddr(link, cl->database->host->ipport))) - { - yaz_log(YLOG_WARN|YLOG_ERRNO, "Lookup of IP address %s failed?", - cl->database->host->ipport); - return 0; + if (!(addr = cs_straddr(link, global_parameters.zproxy_override))) + { + yaz_log(YLOG_WARN|YLOG_ERRNO, + "Lookup of IP address %s failed", + global_parameters.zproxy_override); + return 0; + } } res = cs_connect(link, addr); @@ -952,7 +1073,7 @@ static struct connection *connection_create(struct client *cl) cl->connection = new; new->link = link; - new->iochan = iochan_create(cs_fileno(link), handler, 0); + new->iochan = iochan_create(cs_fileno(link), 0, handler, 0); iochan_setdata(new->iochan, new); new->iochan->next = channel_list; channel_list = new->iochan; @@ -1013,6 +1134,8 @@ static int client_prep_connection(struct client *cl) return 0; } +#ifdef GAGA // Moved to database.c + // This function will most likely vanish when a proper target profile mechanism is // introduced. void load_simpletargets(const char *fn) @@ -1029,21 +1152,26 @@ void load_simpletargets(const char *fn) while (fgets(line, 255, f)) { char *url, *db; + char *name; struct host *host; struct database *database; if (strncmp(line, "target ", 7)) continue; + line[strlen(line) - 1] = '\0'; + + if ((name = strchr(line, ';'))) + *(name++) = '\0'; + url = line + 7; - url[strlen(url) - 1] = '\0'; - yaz_log(YLOG_DEBUG, "Target: %s", url); if ((db = strchr(url, '/'))) *(db++) = '\0'; else db = "Default"; + yaz_log(YLOG_LOG, "Target: %s, '%s'", url, db); for (host = hosts; host; host = host->next) - if (!strcmp(url, host->hostport)) + if (!strcmp((const char *) url, host->hostport)) break; if (!host) { @@ -1094,6 +1222,10 @@ void load_simpletargets(const char *fn) strcpy(database->url, url); strcat(database->url, "/"); strcat(database->url, db); + if (name) + database->name = xstrdup(name); + else + database->name = 0; database->databases = xmalloc(2 * sizeof(char *)); database->databases[0] = xstrdup(db); @@ -1108,40 +1240,7 @@ void load_simpletargets(const char *fn) fclose(f); } -static void pull_terms(NMEM nmem, struct ccl_rpn_node *n, char **termlist, int *num) -{ - switch (n->kind) - { - case CCL_RPN_AND: - case CCL_RPN_OR: - case CCL_RPN_NOT: - case CCL_RPN_PROX: - pull_terms(nmem, n->u.p[0], termlist, num); - pull_terms(nmem, n->u.p[1], termlist, num); - break; - case CCL_RPN_TERM: - termlist[(*num)++] = nmem_strdup(nmem, n->u.t.term); - break; - default: // NOOP - break; - } -} - -// Extract terms from query into null-terminated termlist -static int extract_terms(NMEM nmem, char *query, char **termlist) -{ - int error, pos; - struct ccl_rpn_node *n; - int num = 0; - - n = ccl_find_str(global_parameters.ccl_filter, query, &error, &pos); - if (!n) - return -1; - pull_terms(nmem, n, termlist, &num); - termlist[num] = 0; - ccl_rpn_delete(n); - return 0; -} +#endif static struct client *client_create(void) { @@ -1200,40 +1299,25 @@ void session_alert_watch(struct session *s, int what) s->watchlist[what].data = 0; } -// This needs to be extended with selection criteria -static struct conf_retrievalprofile *database_retrieval_profile(struct database *db) +//callback for grep_databases +static void select_targets_callback(void *context, struct database *db) { - if (!config) - { - yaz_log(YLOG_FATAL, "Must load configuration (-f)"); - exit(1); - } - if (!config->retrievalprofiles) - { - yaz_log(YLOG_FATAL, "No retrieval profiles defined"); - } - return config->retrievalprofiles; + struct session *se = (struct session*) context; + struct client *cl = client_create(); + cl->database = db; + cl->session = se; + cl->next = se->clients; + se->clients = cl; } // This should be extended with parameters to control selection criteria // Associates a set of clients with a session; -int select_targets(struct session *se) +int select_targets(struct session *se, struct database_criterion *crit) { - struct database *db; - int c = 0; - while (se->clients) client_destroy(se->clients); - for (db = databases; db; db = db->next) - { - struct client *cl = client_create(); - cl->database = db; - cl->session = se; - cl->next = se->clients; - se->clients = cl; - c++; - } - return c; + + return grep_databases(se, crit, select_targets_callback); } int session_active_clients(struct session *s) @@ -1251,33 +1335,72 @@ int session_active_clients(struct session *s) return res; } -char *search(struct session *se, char *query) +// parses crit1=val1,crit2=val2|val3,... +static struct database_criterion *parse_filter(NMEM m, const char *buf) +{ + struct database_criterion *res = 0; + char **values; + int num; + int i; + + if (!buf || !*buf) + return 0; + nmem_strsplit(m, ",", buf, &values, &num); + for (i = 0; i < num; i++) + { + char **subvalues; + int subnum; + int subi; + struct database_criterion *new = nmem_malloc(m, sizeof(*new)); + char *eq = strchr(values[i], '='); + if (!eq) + { + yaz_log(YLOG_WARN, "Missing equal-sign in filter"); + return 0; + } + *(eq++) = '\0'; + new->name = values[i]; + nmem_strsplit(m, "|", eq, &subvalues, &subnum); + new->values = 0; + for (subi = 0; subi < subnum; subi++) + { + struct database_criterion_value *newv = nmem_malloc(m, sizeof(*newv)); + newv->value = subvalues[subi]; + newv->next = new->values; + new->values = newv; + } + new->next = res; + res = new; + } + return res; +} + +char *search(struct session *se, char *query, char *filter) { int live_channels = 0; struct client *cl; + struct database_criterion *criteria; yaz_log(YLOG_DEBUG, "Search"); + nmem_reset(se->nmem); + criteria = parse_filter(se->nmem, filter); strcpy(se->query, query); se->requestid++; - nmem_reset(se->nmem); + // Release any existing clients + select_targets(se, criteria); for (cl = se->clients; cl; cl = cl->next) { - cl->hits = -1; - cl->records = 0; - cl->diagnostic = 0; - if (client_prep_connection(cl)) live_channels++; } if (live_channels) { - char *p[512]; int maxrecs = live_channels * global_parameters.toget; se->num_termlists = 0; se->reclist = reclist_create(se->nmem, maxrecs); - extract_terms(se->nmem, query, p); - se->relevance = relevance_create(se->nmem, (const char **) p, maxrecs); + // This will be initialized in send_search() + se->relevance = 0; se->total_records = se->total_hits = se->total_merged = 0; se->expected_maxrecs = maxrecs; } @@ -1293,7 +1416,7 @@ void destroy_session(struct session *s) while (s->clients) client_destroy(s->clients); nmem_destroy(s->nmem); - wrbuf_free(s->wrbuf, 1); + wrbuf_destroy(s->wrbuf); } struct session *new_session() @@ -1319,8 +1442,6 @@ struct session *new_session() session->watchlist[i].fun = 0; } - select_targets(session); - return session; } @@ -1332,7 +1453,8 @@ struct hitsbytarget *hitsbytarget(struct session *se, int *count) *count = 0; for (cl = se->clients; cl; cl = cl->next) { - strcpy(res[*count].id, cl->database->host->hostport); + res[*count].id = cl->database->url; + res[*count].name = cl->database->name; res[*count].hits = cl->hits; res[*count].records = cl->records; res[*count].diagnostic = cl->diagnostic; @@ -1349,7 +1471,7 @@ struct termlist_score **termlist(struct session *s, const char *name, int *num) int i; for (i = 0; i < s->num_termlists; i++) - if (!strcmp(s->termlists[i].name, name)) + if (!strcmp((const char *) s->termlists[i].name, name)) return termlist_highscore(s->termlists[i].termlist, num); return 0; } @@ -1457,17 +1579,6 @@ void statistics(struct session *se, struct statistics *stat) stat->num_clients = count; } -static CCL_bibset load_cclfile(const char *fn) -{ - CCL_bibset res = ccl_qual_mk(); - if (ccl_qual_fname(res, fn) < 0) - { - yaz_log(YLOG_FATAL|YLOG_ERRNO, "%s", fn); - exit(1); - } - return res; -} - static void start_http_listener(void) { char hp[128] = ""; @@ -1488,6 +1599,36 @@ static void start_http_listener(void) http_init(hp); } +// Initialize CCL map for a target +// Note: This approach ignores user-specific CCL maps, for which I +// don't presently see any application. +static void prepare_cclmap(void *context, struct database *db) +{ + struct setting *s; + + if (!db->settings) + return; + db->ccl_map = ccl_qual_mk(); + for (s = db->settings[PZ_CCLMAP]; s; s = s->next) + if (!*s->user) + { + char *p = strchr(s->name + 3, ':'); + if (!p) + { + yaz_log(YLOG_FATAL, "Malformed cclmap name: %s", s->name); + exit(1); + } + p++; + ccl_qual_fitem(db->ccl_map, s->value, p); + } +} + +// Read settings for each database, and prepare a CCL map for that database +static void prepare_cclmaps(void) +{ + grep_databases(0, 0, prepare_cclmap); +} + static void start_proxy(void) { char hp[128] = ""; @@ -1508,9 +1649,44 @@ static void start_proxy(void) else return; - http_set_proxyaddr(hp); + http_set_proxyaddr(hp, ser->myurl ? ser->myurl : ""); } +static void start_zproxy(void) +{ + struct conf_server *ser = global_parameters.server; + + if (*global_parameters.zproxy_override){ + yaz_log(YLOG_LOG, "Z39.50 proxy %s", + global_parameters.zproxy_override); + return; + } + + else if (ser->zproxy_host || ser->zproxy_port) + { + char hp[128] = ""; + + strcpy(hp, ser->zproxy_host ? ser->zproxy_host : ""); + if (ser->zproxy_port) + { + if (*hp) + strcat(hp, ":"); + else + strcat(hp, "@:"); + + sprintf(hp + strlen(hp), "%d", ser->zproxy_port); + } + strcpy(global_parameters.zproxy_override, hp); + yaz_log(YLOG_LOG, "Z39.50 proxy %s", + global_parameters.zproxy_override); + + } + else + return; +} + + + int main(int argc, char **argv) { int ret; @@ -1521,7 +1697,7 @@ int main(int argc, char **argv) yaz_log_init(YLOG_DEFAULT_LEVEL, "pazpar2", 0); - while ((ret = options("f:x:h:p:C:s:d", argv, argc, &arg)) != -2) + while ((ret = options("t:f:x:h:p:z:s:d", argv, argc, &arg)) != -2) { switch (ret) { case 'f': @@ -1531,12 +1707,15 @@ int main(int argc, char **argv) case 'h': strcpy(global_parameters.listener_override, arg); break; - case 'C': - global_parameters.ccl_filter = load_cclfile(arg); - break; case 'p': strcpy(global_parameters.proxy_override, arg); break; + case 'z': + strcpy(global_parameters.zproxy_override, arg); + break; + case 't': + strcpy(global_parameters.settings_path_override, arg); + break; case 's': load_simpletargets(arg); break; @@ -1549,7 +1728,9 @@ int main(int argc, char **argv) " -h [host:]port (REST protocol listener)\n" " -C cclconfig\n" " -s simpletargetfile\n" - " -p hostname[:portno] (HTTP proxy)\n"); + " -p hostname[:portno] (HTTP proxy)\n" + " -z hostname[:portno] (Z39.50 proxy)\n" + " -d (show internal records)\n"); exit(1); } } @@ -1563,7 +1744,15 @@ int main(int argc, char **argv) start_http_listener(); start_proxy(); - global_parameters.ccl_filter = load_cclfile("../etc/default.bib"); + start_zproxy(); + + if (*global_parameters.settings_path_override) + settings_read(global_parameters.settings_path_override); + else if (global_parameters.server->settings) + settings_read(global_parameters.server->settings); + else + yaz_log(YLOG_WARN, "No settings-directory specified. Problems may ensue!"); + prepare_cclmaps(); global_parameters.yaz_marc = yaz_marc_create(); yaz_marc_subfield_str(global_parameters.yaz_marc, "\t"); global_parameters.odr_in = odr_createmem(ODR_DECODE);