X-Git-Url: http://lists.indexdata.com/cgi-bin?a=blobdiff_plain;f=src%2Fpazpar2.c;h=13fbb7239834ca28cf8b77523a5d7b69d8e170f4;hb=acd20da059780084ac6e0a84eab5be85cc0092fd;hp=6c3e1962d46ef893f90bce0506e4fa353fa18aa8;hpb=56d94019bbdff3b9823069b61f0f0ec6a671f923;p=pazpar2-moved-to-github.git diff --git a/src/pazpar2.c b/src/pazpar2.c index 6c3e196..13fbb72 100644 --- a/src/pazpar2.c +++ b/src/pazpar2.c @@ -1,4 +1,4 @@ -/* $Id: pazpar2.c,v 1.52 2007-03-23 03:26:22 quinn Exp $ */ +/* $Id: pazpar2.c,v 1.61 2007-04-03 04:05:01 quinn Exp $ */ #include #include @@ -17,6 +17,7 @@ #include #include #include +#include #include #include @@ -39,6 +40,7 @@ #include "relevance.h" #include "config.h" #include "database.h" +#include "settings.h" #define MAX_CHUNK 15 @@ -48,6 +50,7 @@ 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); 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 @@ -72,6 +75,8 @@ struct parameters global_parameters = { "", "", + "", + "", 0, 0, 30, @@ -84,7 +89,6 @@ struct parameters global_parameters = MAX_CHUNK, 0, 0, - 0, 0 }; @@ -119,6 +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); @@ -134,6 +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); @@ -144,6 +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); @@ -156,12 +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; @@ -174,12 +231,16 @@ static void send_search(IOCHAN i) for (ndb = 0; db->databases[ndb]; ndb++) databaselist[ndb] = db->databases[ndb]; - a->u.searchRequest->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; @@ -200,9 +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) @@ -217,9 +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) { @@ -324,42 +389,6 @@ 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; @@ -413,7 +442,7 @@ static xmlDoc *normalize_record(struct client *cl, Z_External *rec) yaz_log(YLOG_WARN, "Failed to encode as XML"); return 0; } - rdoc = xmlNewDoc("1.0"); + rdoc = xmlNewDoc((xmlChar *) "1.0"); xmlDocSetRootElement(rdoc, res); } else @@ -505,7 +534,7 @@ static struct record *ingest_record(struct client *cl, Z_External *rec) 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); @@ -519,11 +548,12 @@ static struct record *ingest_record(struct client *cl, Z_External *rec) 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); @@ -545,7 +575,7 @@ 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; @@ -553,7 +583,7 @@ 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) @@ -561,7 +591,7 @@ static struct record *ingest_record(struct client *cl, Z_External *rec) // 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) @@ -586,7 +616,7 @@ static struct record *ingest_record(struct client *cl, Z_External *rec) if (md->type == Metadata_type_generic) { char *p, *pe; - for (p = value; *p && isspace(*p); p++) + for (p = (char *) value; *p && isspace(*p); p++) ; for (pe = p + strlen(p) - 1; pe > p && strchr(" ,/.:([", *pe); pe--) @@ -596,7 +626,7 @@ static struct record *ingest_record(struct client *cl, Z_External *rec) } else if (md->type == Metadata_type_year) { - if (extract_years(value, &first, &last) < 0) + if (extract_years((char *) value, &first, &last) < 0) continue; } else @@ -613,7 +643,7 @@ 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) { @@ -674,22 +704,23 @@ 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) { if (md->type == Metadata_type_year) { char year[64]; sprintf(year, "%d", last); - add_facet(se, type, year); + add_facet(se, (char *) type, year); if (first != last) { sprintf(year, "%d", first); - add_facet(se, type, year); + add_facet(se, (char *) type, year); } } else - add_facet(se, type, value); + add_facet(se, (char *) type, (char *) value); } xmlFree(type); xmlFree(value); @@ -711,6 +742,19 @@ 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 @@ -942,18 +986,40 @@ 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; + yaz_log(YLOG_LOG, "Connection cs_create_host %s proxy %s", + cl->database->url, global_parameters.zproxy_override); + + 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); @@ -979,7 +1045,7 @@ 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; @@ -1077,7 +1143,7 @@ void load_simpletargets(const char *fn) 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) { @@ -1148,41 +1214,6 @@ void load_simpletargets(const char *fn) #endif -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; -} - static struct client *client_create(void) { struct client *r; @@ -1337,12 +1368,11 @@ char *search(struct session *se, char *query, char *filter) } 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; } @@ -1413,7 +1443,7 @@ 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; } @@ -1521,17 +1551,6 @@ 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] = ""; @@ -1552,6 +1571,36 @@ 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] = ""; @@ -1575,6 +1624,41 @@ static void start_proxy(void) 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; @@ -1585,7 +1669,7 @@ 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': @@ -1595,12 +1679,15 @@ 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; @@ -1614,6 +1701,7 @@ int main(int argc, char **argv) " -C cclconfig\n" " -s simpletargetfile\n" " -p hostname[:portno] (HTTP proxy)\n" + " -z hostname[:portno] (Z39.50 proxy)\n" " -d (show internal records)\n"); exit(1); } @@ -1628,8 +1716,15 @@ int main(int argc, char **argv) start_http_listener(); start_proxy(); - if (!global_parameters.ccl_filter) - 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);