X-Git-Url: http://lists.indexdata.com/cgi-bin?a=blobdiff_plain;f=src%2Fpazpar2.c;h=1a8fdfaa824de8eeb4a701e5a6494991404317bb;hb=d591758eff900dfdbe1545e2d408dbb0f061be1c;hp=3482d839a9db8b618258448a888d46886131fd9e;hpb=1ef3f77533020ff7c2373482a5b35caf10387c7a;p=pazpar2-moved-to-github.git diff --git a/src/pazpar2.c b/src/pazpar2.c index 3482d83..1a8fdfa 100644 --- a/src/pazpar2.c +++ b/src/pazpar2.c @@ -1,4 +1,4 @@ -/* $Id: pazpar2.c,v 1.11 2007-01-04 07:27:29 adam Exp $ */; +/* $Id: pazpar2.c,v 1.55 2007-03-28 04:33:41 quinn Exp $ */ #include #include @@ -11,36 +11,44 @@ #include #include +#include #include #include #include #include #include +#include #include #include +#if HAVE_CONFIG_H +#include "cconfig.h" +#endif + #define USE_TIMING 0 #if USE_TIMING #include #endif +#include + #include "pazpar2.h" #include "eventl.h" -#include "command.h" #include "http.h" #include "termlists.h" #include "reclists.h" #include "relevance.h" #include "config.h" +#include "database.h" +#include "settings.h" -#define PAZPAR2_VERSION "0.1" #define MAX_CHUNK 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); IOCHAN channel_list = 0; // Master list of connections we're handling events to @@ -48,9 +56,6 @@ 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", @@ -64,13 +69,19 @@ static char *client_states[] = { "Client_Stopped" }; +// Note: Some things in this structure will eventually move to configuration struct parameters global_parameters = { + "", + "", + "", + "", + 0, 0, 30, "81", "Index Data PazPar2 (MasterKey)", - PAZPAR2_VERSION, + VERSION, 600, // 10 minutes 60, 100, @@ -112,6 +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); @@ -127,6 +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); @@ -167,7 +191,7 @@ static void send_search(IOCHAN i) for (ndb = 0; db->databases[ndb]; ndb++) databaselist[ndb] = db->databases[ndb]; - a->u.presentRequest->preferredRecordSyntax = + a->u.searchRequest->preferredRecordSyntax = yaz_oidval_to_z3950oid(global_parameters.odr_out, CLASS_RECSYN, VAL_USMARC); a->u.searchRequest->smallSetUpperBound = &ssub; @@ -198,6 +222,8 @@ static void send_present(IOCHAN i) int start = cl->records + 1; toget = global_parameters.chunk; + if (toget > global_parameters.toget - cl->records) + toget = global_parameters.toget - cl->records; if (toget > cl->hits - cl->records) toget = cl->hits - cl->records; @@ -254,7 +280,6 @@ static void do_searchResponse(IOCHAN i, Z_APDU *a) if (r->presentStatus && !*r->presentStatus && r->records) { yaz_log(YLOG_DEBUG, "Records in search response"); - cl->records += *r->numberOfRecordsReturned; ingest_records(cl, r->records); } cl->state = Client_Idle; @@ -275,10 +300,27 @@ static void do_searchResponse(IOCHAN i, Z_APDU *a) } } -char *normalize_mergekey(char *buf) +char *normalize_mergekey(char *buf, int skiparticle) { char *p = buf, *pout = buf; + if (skiparticle) + { + char firstword[64]; + char articles[] = "the den der die des an a "; // must end in space + + while (*p && !isalnum(*p)) + p++; + pout = firstword; + while (*p && *p != ' ' && pout - firstword < 62) + *(pout++) = tolower(*(p++)); + *(pout++) = ' '; + *(pout++) = '\0'; + if (!strstr(articles, firstword)) + p = buf; + pout = buf; + } + while (*p) { while (*p && !isalnum(*p)) @@ -291,7 +333,10 @@ char *normalize_mergekey(char *buf) p++; } if (buf != pout) - *pout = '\0'; + do { + *(pout--) = '\0'; + } + while (pout > buf && *pout == ' '); return buf; } @@ -336,6 +381,8 @@ static void add_facet(struct session *s, const char *type, const char *value) { int i; + if (!*value) + return; for (i = 0; i < s->num_termlists; i++) if (!strcmp(s->termlists[i].name, type)) break; @@ -383,7 +430,7 @@ 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 @@ -391,6 +438,17 @@ 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; @@ -410,74 +468,263 @@ static xmlDoc *normalize_record(struct client *cl, Z_External *rec) if (global_parameters.dump_records) { fprintf(stderr, "Record:\n----------------\n"); +#if LIBXML_VERSION >= 20600 xmlDocFormatDump(stderr, rdoc, 1); +#else + xmlDocDump(stderr, rdoc); +#endif } return rdoc; } +// Extract what appears to be years from buf, storing highest and +// lowest values. +static int extract_years(const char *buf, int *first, int *last) +{ + *first = -1; + *last = -1; + while (*buf) + { + const char *e; + int len; + + while (*buf && !isdigit(*buf)) + buf++; + len = 0; + for (e = buf; *e && isdigit(*e); e++) + len++; + if (len == 4) + { + int value = atoi(buf); + if (*first < 0 || value < *first) + *first = value; + if (*last < 0 || value > *last) + *last = value; + } + buf = e; + } + return *first; +} + static struct record *ingest_record(struct client *cl, Z_External *rec) { xmlDoc *xdoc = normalize_record(cl, rec); xmlNode *root, *n; - struct record *res, *head; + struct record *res; + struct record_cluster *cluster; struct session *se = cl->session; xmlChar *mergekey, *mergekey_norm; + 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); return 0; } res = nmem_malloc(se->nmem, sizeof(struct record)); - res->next_cluster = 0; - res->target_offset = -1; - res->term_frequency_vec = 0; - res->title = "Unknown"; - res->relevance = 0; + 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); - res->merge_key = normalize_mergekey(mergekey_norm); + normalize_mergekey((char *) mergekey_norm, 0); - head = reclist_insert(se->reclist, res); - relevance_newrec(se->relevance, head); + 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 */ + xmlFreeDoc(xdoc); + return 0; + } + relevance_newrec(se->relevance, cluster); for (n = root->children; n; n = n->next) { - if (n->type != XML_ELEMENT_NODE) - continue; - if (!strcmp(n->name, "facet")) - { - xmlChar *type = xmlGetProp(n, "type"); - xmlChar *value = xmlNodeListGetString(xdoc, n->children, 0); - add_facet(se, type, value); - relevance_countwords(se->relevance, head, value, 1); + if (type) xmlFree(type); + if (value) xmlFree(value); - } - else if (!strcmp(n->name, "metadata")) + type = value = 0; + + if (n->type != XML_ELEMENT_NODE) + continue; + if (!strcmp((const char *) n->name, "metadata")) { - xmlChar *type = xmlGetProp(n, "type"), *value; - if (!strcmp(type, "title")) - res->title = nmem_strdup(se->nmem, - value = xmlNodeListGetString(xdoc, n->children, 0)); + struct conf_metadata *md = 0; + struct conf_sortkey *sk = 0; + struct record_metadata **wheretoput, *newm; + int imeta; + int first, last; + + type = xmlGetProp(n, (xmlChar *) "type"); + value = xmlNodeListGetString(xdoc, n->children, 0); - relevance_countwords(se->relevance, head, value, 4); + if (!type || !value) + continue; + + // First, find out what field we're looking at + for (imeta = 0; imeta < service->num_metadata; imeta++) + if (!strcmp((const char *) type, service->metadata[imeta].name)) + { + md = &service->metadata[imeta]; + if (md->sortkey_offset >= 0) + sk = &service->sortkeys[md->sortkey_offset]; + break; + } + if (!md) + { + yaz_log(YLOG_WARN, "Ignoring unknown metadata element: %s", type); + continue; + } + + // Find out where we are putting it + if (md->merge == Metadata_merge_no) + wheretoput = &res->metadata[imeta]; + else + wheretoput = &cluster->metadata[imeta]; + + // Put it there + newm = nmem_malloc(se->nmem, sizeof(struct record_metadata)); + newm->next = 0; + if (md->type == Metadata_type_generic) + { + 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((char *) value, &first, &last) < 0) + continue; + } + else + { + yaz_log(YLOG_WARN, "Unknown type in metadata element %s", type); + continue; + } + if (md->type == Metadata_type_year && md->merge != Metadata_merge_range) + { + yaz_log(YLOG_WARN, "Only range merging supported for years"); + continue; + } + if (md->merge == Metadata_merge_unique) + { + struct record_metadata *mnode; + for (mnode = *wheretoput; mnode; mnode = mnode->next) + if (!strcmp((const char *) mnode->data.text, newm->data.text)) + break; + if (!mnode) + { + newm->next = *wheretoput; + *wheretoput = newm; + } + } + else if (md->merge == Metadata_merge_longest) + { + if (!*wheretoput || + strlen(newm->data.text) > strlen((*wheretoput)->data.text)) + { + *wheretoput = newm; + if (sk) + { + char *s = nmem_strdup(se->nmem, newm->data.text); + if (!cluster->sortkeys[md->sortkey_offset]) + cluster->sortkeys[md->sortkey_offset] = + nmem_malloc(se->nmem, sizeof(union data_types)); + normalize_mergekey(s, + (sk->type == Metadata_sortkey_skiparticle)); + cluster->sortkeys[md->sortkey_offset]->text = s; + } + } + } + else if (md->merge == Metadata_merge_all || md->merge == Metadata_merge_no) + { + newm->next = *wheretoput; + *wheretoput = newm; + } + else if (md->merge == Metadata_merge_range) + { + assert(md->type == Metadata_type_year); + if (!*wheretoput) + { + *wheretoput = newm; + (*wheretoput)->data.number.min = first; + (*wheretoput)->data.number.max = last; + if (sk) + cluster->sortkeys[md->sortkey_offset] = &newm->data; + } + else + { + if (first < (*wheretoput)->data.number.min) + (*wheretoput)->data.number.min = first; + if (last > (*wheretoput)->data.number.max) + (*wheretoput)->data.number.max = last; + } +#ifdef GAGA + if (sk) + { + union data_types *sdata = cluster->sortkeys[md->sortkey_offset]; + yaz_log(YLOG_LOG, "SK range: %d-%d", sdata->number.min, sdata->number.max); + } +#endif + } + else + yaz_log(YLOG_WARN, "Don't know how to merge on element name %s", md->name); + + if (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, (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; } else yaz_log(YLOG_WARN, "Unexpected element %s in internal record", n->name); } + if (type) + xmlFree(type); + if (value) + xmlFree(value); xmlFreeDoc(xdoc); - relevance_donerecord(se->relevance, head); + relevance_donerecord(se->relevance, cluster); se->total_records++; return res; @@ -500,6 +747,7 @@ 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"); @@ -541,7 +789,6 @@ 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; ingest_records(cl, r->records); cl->state = Client_Idle; } @@ -714,17 +961,40 @@ 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 failed?"); - 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); @@ -811,6 +1081,8 @@ 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) @@ -827,21 +1099,26 @@ 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) { @@ -879,7 +1156,7 @@ void load_simpletargets(const char *fn) } assert(addrinfo->ai_family == PF_INET); memcpy(addrbuf, &((struct sockaddr_in*)addrinfo->ai_addr)->sin_addr.s_addr, 4); - sprintf(ipport, "%hhd.%hhd.%hhd.%hhd:%s", + sprintf(ipport, "%u.%u.%u.%u:%s", addrbuf[0], addrbuf[1], addrbuf[2], addrbuf[3], port); host->ipport = xstrdup(ipport); freeaddrinfo(addrinfo); @@ -892,6 +1169,10 @@ 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); @@ -906,6 +1187,8 @@ void load_simpletargets(const char *fn) fclose(f); } +#endif + static void pull_terms(NMEM nmem, struct ccl_rpn_node *n, char **termlist, int *num) { switch (n->kind) @@ -998,40 +1281,25 @@ 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) @@ -1049,22 +1317,62 @@ 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++; } @@ -1076,7 +1384,7 @@ char *search(struct session *se, char *query) se->reclist = reclist_create(se->nmem, maxrecs); extract_terms(se->nmem, query, p); se->relevance = relevance_create(se->nmem, (const char **) p, maxrecs); - se->total_records = se->total_hits = 0; + se->total_records = se->total_hits = se->total_merged = 0; se->expected_maxrecs = maxrecs; } else @@ -1091,7 +1399,7 @@ 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() @@ -1117,8 +1425,6 @@ struct session *new_session() session->watchlist[i].fun = 0; } - select_targets(session); - return session; } @@ -1130,7 +1436,8 @@ 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; @@ -1147,14 +1454,12 @@ 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; } -#ifdef REPORT_NMEM -// conditional compilation by SH: This lead to a warning with currently installed -// YAZ header files on us1 +#ifdef MISSING_HEADERS void report_nmem_stats(void) { size_t in_use, is_free; @@ -1167,16 +1472,35 @@ void report_nmem_stats(void) } #endif -struct record **show(struct session *s, int start, int *num, int *total, - int *sumhits, NMEM nmem_show) +struct record_cluster *show_single(struct session *s, int id) { - struct record **recs = nmem_malloc(nmem_show, *num - * sizeof(struct record *)); + struct record_cluster *r; + + reclist_rewind(s->reclist); + while ((r = reclist_read_record(s->reclist))) + if (r->recid == id) + return r; + return 0; +} + +struct record_cluster **show(struct session *s, struct reclist_sortparms *sp, int start, + int *num, int *total, int *sumhits, NMEM nmem_show) +{ + struct record_cluster **recs = nmem_malloc(nmem_show, *num + * sizeof(struct record_cluster *)); + struct reclist_sortparms *spp; int i; #if USE_TIMING yaz_timing_t t = yaz_timing_create(); #endif - relevance_prepare_read(s->relevance, s->reclist); + + for (spp = sp; spp; spp = spp->next) + if (spp->type == Metadata_sortkey_relevance) + { + relevance_prepare_read(s->relevance, s->reclist); + break; + } + reclist_sort(s->reclist, sp); *total = s->reclist->num_records; *sumhits = s->total_hits; @@ -1191,7 +1515,7 @@ struct record **show(struct session *s, int start, int *num, int *total, for (i = 0; i < *num; i++) { - struct record *r = reclist_read_record(s->reclist); + struct record_cluster *r = reclist_read_record(s->reclist); if (!r) { *num = i; @@ -1214,7 +1538,7 @@ void statistics(struct session *se, struct statistics *stat) struct client *cl; int count = 0; - bzero(stat, sizeof(*stat)); + memset(stat, 0, sizeof(*stat)); for (cl = se->clients; cl; cl = cl->next) { if (!cl->connection) @@ -1249,37 +1573,115 @@ static CCL_bibset load_cclfile(const char *fn) return res; } +static void start_http_listener(void) +{ + char hp[128] = ""; + struct conf_server *ser = global_parameters.server; + + if (*global_parameters.listener_override) + strcpy(hp, global_parameters.listener_override); + else + { + strcpy(hp, ser->host ? ser->host : ""); + if (ser->port) + { + if (*hp) + strcat(hp, ":"); + sprintf(hp + strlen(hp), "%d", ser->port); + } + } + http_init(hp); +} + +static void start_proxy(void) +{ + char hp[128] = ""; + struct conf_server *ser = global_parameters.server; + + if (*global_parameters.proxy_override) + strcpy(hp, global_parameters.proxy_override); + else if (ser->proxy_host || ser->proxy_port) + { + strcpy(hp, ser->proxy_host ? ser->proxy_host : ""); + if (ser->proxy_port) + { + if (*hp) + strcat(hp, ":"); + sprintf(hp + strlen(hp), "%d", ser->proxy_port); + } + } + else + return; + + 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; char *arg; - int setport = 0; - if (signal(SIGPIPE, SIG_IGN) < 0) + if (signal(SIGPIPE, SIG_IGN) == SIG_ERR) yaz_log(YLOG_WARN|YLOG_ERRNO, "signal"); yaz_log_init(YLOG_DEFAULT_LEVEL, "pazpar2", 0); - while ((ret = options("f:x:c:h:p:C:s:d", argv, argc, &arg)) != -2) + while ((ret = options("t:f:x:h:p:z:C:s:d", argv, argc, &arg)) != -2) { switch (ret) { case 'f': if (!read_config(arg)) exit(1); break; - case 'c': - command_init(atoi(arg)); - setport++; - break; case 'h': - http_init(arg); - setport++; + strcpy(global_parameters.listener_override, arg); break; case 'C': global_parameters.ccl_filter = load_cclfile(arg); break; case 'p': - http_set_proxyaddr(arg); + strcpy(global_parameters.proxy_override, arg); + break; + case 'z': + strcpy(global_parameters.zproxy_override, arg); + break; + case 't': + strcpy(global_parameters.settings_path, arg); break; case 's': load_simpletargets(arg); @@ -1291,21 +1693,30 @@ int main(int argc, char **argv) fprintf(stderr, "Usage: pazpar2\n" " -f configfile\n" " -h [host:]port (REST protocol listener)\n" - " -c cmdport (telnet-style)\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); } } - if (!setport) + if (!config) { - fprintf(stderr, "Set command port with -h or -c\n"); + yaz_log(YLOG_FATAL, "Load config with -f"); exit(1); } + global_parameters.server = config->servers; + + start_http_listener(); + start_proxy(); + start_zproxy(); - global_parameters.ccl_filter = load_cclfile("../etc/default.bib"); + if (*global_parameters.settings_path) + settings_read(global_parameters.settings_path); + if (!global_parameters.ccl_filter) + global_parameters.ccl_filter = load_cclfile("../etc/default.bib"); global_parameters.yaz_marc = yaz_marc_create(); yaz_marc_subfield_str(global_parameters.yaz_marc, "\t"); global_parameters.odr_in = odr_createmem(ODR_DECODE);