X-Git-Url: http://lists.indexdata.com/cgi-bin?a=blobdiff_plain;f=src%2Fsession.c;h=105c064323944b539e02ac58f72800f1881f11ac;hb=eb673212c641c36afba4cd502e283247de52a015;hp=78aad616886117ce0b2fd7344a8378224c75d5aa;hpb=116eaa2065a4b4fec0dcc55985170e3364132995;p=pazpar2-moved-to-github.git diff --git a/src/session.c b/src/session.c index 78aad61..105c064 100644 --- a/src/session.c +++ b/src/session.c @@ -76,6 +76,8 @@ Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA #include "settings.h" #include "normalize7bit.h" +#include + #define TERMLIST_HIGH_SCORE 25 #define MAX_CHUNK 15 @@ -406,40 +408,43 @@ const char *session_setting_oneval(struct session_database *db, int offset) // setting. However, this is not a realistic use scenario. static int prepare_map(struct session *se, struct session_database *sdb) { - const char *s; - - if (sdb->settings && sdb->settings[PZ_XSLT] && !sdb->map && - (s = session_setting_oneval(sdb, PZ_XSLT))) + if (sdb->settings && !sdb->map) { - char auto_stylesheet[256]; + const char *s; - if (!strcmp(s, "auto")) + if (sdb->settings[PZ_XSLT] && + (s = session_setting_oneval(sdb, PZ_XSLT))) { - const char *request_syntax = session_setting_oneval( - sdb, PZ_REQUESTSYNTAX); - if (request_syntax) + char auto_stylesheet[256]; + + if (!strcmp(s, "auto")) { - char *cp; - yaz_snprintf(auto_stylesheet, sizeof(auto_stylesheet), - "%s.xsl", request_syntax); - for (cp = auto_stylesheet; *cp; cp++) + const char *request_syntax = session_setting_oneval( + sdb, PZ_REQUESTSYNTAX); + if (request_syntax) { - /* deliberately only consider ASCII */ - if (*cp > 32 && *cp < 127) - *cp = tolower(*cp); + char *cp; + yaz_snprintf(auto_stylesheet, sizeof(auto_stylesheet), + "%s.xsl", request_syntax); + for (cp = auto_stylesheet; *cp; cp++) + { + /* deliberately only consider ASCII */ + if (*cp > 32 && *cp < 127) + *cp = tolower(*cp); + } + s = auto_stylesheet; + } + else + { + session_log(se, YLOG_WARN, + "No pz:requestsyntax for auto stylesheet"); } - s = auto_stylesheet; - } - else - { - session_log(se, YLOG_WARN, - "No pz:requestsyntax for auto stylesheet"); } + sdb->map = normalize_cache_get(se->normalize_cache, + se->service, s); + if (!sdb->map) + return -1; } - sdb->map = normalize_cache_get(se->normalize_cache, - se->service, s); - if (!sdb->map) - return -1; } return 0; } @@ -617,43 +622,70 @@ int session_is_preferred_clients_ready(struct session *s) return res == 0; } -void session_sort(struct session *se, const char *field, int increasing) +static void session_clear_set(struct session *se, + const char *sort_field, int increasing) +{ + reclist_destroy(se->reclist); + se->reclist = 0; + if (nmem_total(se->nmem)) + session_log(se, YLOG_DEBUG, "NMEN operation usage %zd", + nmem_total(se->nmem)); + nmem_reset(se->nmem); + se->total_records = se->total_merged = 0; + se->num_termlists = 0; + + /* reset list of sorted results and clear to relevance search */ + se->sorted_results = nmem_malloc(se->nmem, sizeof(*se->sorted_results)); + se->sorted_results->field = nmem_strdup(se->nmem, sort_field); + se->sorted_results->increasing = increasing; + se->sorted_results->next = 0; + + se->reclist = reclist_create(se->nmem); +} + +void session_sort(struct session *se, const char *field, int increasing, + int clear_set) { struct session_sorted_results *sr; struct client_list *l; session_enter(se); - /* see if we already have sorted for this critieria */ - for (sr = se->sorted_results; sr; sr = sr->next) + yaz_log(YLOG_LOG, "session_sort field=%s", field); + if (clear_set) { - if (!strcmp(field, sr->field) && increasing == sr->increasing) - break; + session_clear_set(se, field, increasing); } - if (sr) + else { - yaz_log(YLOG_LOG, "search_sort: field=%s increasing=%d already fetched", - field, increasing); - session_leave(se); - return; + /* see if we already have sorted for this critieria */ + for (sr = se->sorted_results; sr; sr = sr->next) + { + if (!strcmp(field, sr->field) && increasing == sr->increasing) + break; + } + if (sr) + { + session_log(se, YLOG_DEBUG, "search_sort: field=%s increasing=%d already fetched", + field, increasing); + session_leave(se); + return; + } + session_log(se, YLOG_DEBUG, "search_sort: field=%s increasing=%d must fetch", + field, increasing); + sr = nmem_malloc(se->nmem, sizeof(*sr)); + sr->field = nmem_strdup(se->nmem, field); + sr->increasing = increasing; + sr->next = se->sorted_results; + se->sorted_results = sr; } - yaz_log(YLOG_LOG, "search_sort: field=%s increasing=%d must fetch", - field, increasing); - sr = nmem_malloc(se->nmem, sizeof(*sr)); - sr->field = nmem_strdup(se->nmem, field); - sr->increasing = increasing; - sr->next = se->sorted_results; - se->sorted_results = sr; - + for (l = se->clients_active; l; l = l->next) { struct client *cl = l->client; - struct timeval tval; - int ret = client_prep_connection(cl, se->service->z3950_operation_timeout, - se->service->z3950_session_timeout, - se->service->server->iochan_man, - &tval, 1); - if (ret) + if (client_get_state(cl) == Client_Connecting || + client_get_state(cl) == Client_Idle || + client_get_state(cl) == Client_Working) client_start_search(cl); } session_leave(se); @@ -686,27 +718,16 @@ enum pazpar2_error_code session_search(struct session *se, session_reset_active_clients(se, 0); session_enter(se); - reclist_destroy(se->reclist); - se->reclist = 0; se->settings_modified = 0; + session_clear_set(se, sort_field, increasing); relevance_destroy(&se->relevance); - nmem_reset(se->nmem); - se->total_records = se->total_merged = 0; - se->num_termlists = 0; - /* reset list of sorted results and clear to relevance search */ - se->sorted_results = nmem_malloc(se->nmem, sizeof(*se->sorted_results)); - se->sorted_results->field = nmem_strdup(se->nmem, sort_field); - se->sorted_results->increasing = increasing; - se->sorted_results->next = 0; - live_channels = select_targets(se, filter); if (!live_channels) { session_leave(se); return PAZPAR2_NO_TARGETS; } - se->reclist = reclist_create(se->nmem); yaz_gettimeofday(&tval); @@ -744,15 +765,15 @@ enum pazpar2_error_code session_search(struct session *se, client_prep_connection(cl, se->service->z3950_operation_timeout, se->service->z3950_session_timeout, se->service->server->iochan_man, - &tval, 0); + &tval); if (parse_ret == 1 && r == 2) { - session_log(se, YLOG_LOG, "client REUSE %s", client_get_id(cl)); + session_log(se, YLOG_LOG, "client %s REUSE result", client_get_id(cl)); client_reingest(cl); } - else + else if (r) { - session_log(se, YLOG_LOG, "client NEW %s", client_get_id(cl)); + session_log(se, YLOG_LOG, "client %s NEW search", client_get_id(cl)); client_start_search(cl); } no_working++; @@ -776,7 +797,7 @@ enum pazpar2_error_code session_search(struct session *se, else return PAZPAR2_NO_TARGETS; } - yaz_log(YLOG_LOG, "session_start_search done"); + session_log(se, YLOG_LOG, "session_start_search done"); return PAZPAR2_NO_ERROR; } @@ -822,8 +843,7 @@ void session_init_databases(struct session *se) static struct session_database *load_session_database(struct session *se, char *id) { - struct database *db = new_database(id, se->session_nmem); - + struct database *db = new_database_inherit_settings(id, se->session_nmem, se->service->settings); session_init_databases_fun((void*) se, db); // New sdb is head of se->databases list @@ -887,6 +907,10 @@ void session_destroy(struct session *se) normalize_cache_destroy(se->normalize_cache); relevance_destroy(&se->relevance); reclist_destroy(se->reclist); + if (nmem_total(se->nmem)) + session_log(se, YLOG_DEBUG, "NMEN operation usage %zd", nmem_total(se->nmem)); + if (nmem_total(se->session_nmem)) + session_log(se, YLOG_DEBUG, "NMEN session usage %zd", nmem_total(se->session_nmem)); nmem_destroy(se->nmem); service_destroy(se->service); yaz_mutex_destroy(&se->session_mutex); @@ -964,7 +988,9 @@ static struct hitsbytarget *hitsbytarget_nb(struct session *se, res[*count].id = client_get_id(cl); res[*count].name = *name ? name : "Unknown"; res[*count].hits = client_get_hits(cl); + res[*count].approximation = client_get_approximation(cl); res[*count].records = client_get_num_records(cl); + res[*count].filtered = client_get_num_records_filtered(cl); res[*count].diagnostic = client_get_diagnostic(cl, &res[*count].addinfo); res[*count].state = client_get_state_str(cl); @@ -1014,14 +1040,25 @@ static int cmp_ht(const void *p1, const void *p2) return h2->hits - h1->hits; } +// Compares two hitsbytarget nodes by hitcount +static int cmp_ht_approx(const void *p1, const void *p2) +{ + const struct hitsbytarget *h1 = p1; + const struct hitsbytarget *h2 = p2; + return h2->approximation - h1->approximation; +} + static int targets_termlist_nb(WRBUF wrbuf, struct session *se, int num, - NMEM nmem) + NMEM nmem, int version) { struct hitsbytarget *ht; int count, i; ht = hitsbytarget_nb(se, &count, nmem); - qsort(ht, count, sizeof(struct hitsbytarget), cmp_ht); + if (version >= 2) + qsort(ht, count, sizeof(struct hitsbytarget), cmp_ht_approx); + else + qsort(ht, count, sizeof(struct hitsbytarget), cmp_ht); for (i = 0; i < count && i < num && ht[i].hits > 0; i++) { @@ -1042,7 +1079,14 @@ static int targets_termlist_nb(WRBUF wrbuf, struct session *se, int num, wrbuf_printf(wrbuf, "" ODR_INT_PRINTF "\n", ht[i].hits); - + + if (version >= 2) { + // Should not print if we know it isn't a approximation. + wrbuf_printf(wrbuf, "" ODR_INT_PRINTF "\n", ht[i].approximation); + wrbuf_printf(wrbuf, "%d\n", ht[i].records - ht[i].filtered); + wrbuf_printf(wrbuf, "%d\n", ht[i].filtered); + } + wrbuf_puts(wrbuf, ""); wrbuf_xmlputs(wrbuf, ht[i].state); wrbuf_puts(wrbuf, "\n"); @@ -1055,7 +1099,7 @@ static int targets_termlist_nb(WRBUF wrbuf, struct session *se, int num, } void perform_termlist(struct http_channel *c, struct session *se, - const char *name, int num) + const char *name, int num, int version) { int i, j; NMEM nmem_tmp = nmem_create(); @@ -1072,7 +1116,8 @@ void perform_termlist(struct http_channel *c, struct session *se, for (j = 0; j < num_names; j++) { const char *tname; - + int must_generate_empty = 1; /* bug 5350 */ + for (i = 0; i < se->num_termlists; i++) { tname = se->termlists[i].name; @@ -1084,6 +1129,7 @@ void perform_termlist(struct http_channel *c, struct session *se, wrbuf_puts(c->wrbuf, "wrbuf, tname); wrbuf_puts(c->wrbuf, "\">\n"); + must_generate_empty = 0; p = termlist_highscore(se->termlists[i].termlist, &len); if (p) @@ -1116,8 +1162,15 @@ void perform_termlist(struct http_channel *c, struct session *se, wrbuf_xmlputs(c->wrbuf, tname); wrbuf_puts(c->wrbuf, "\">\n"); - targets_termlist_nb(c->wrbuf, se, num, c->nmem); + targets_termlist_nb(c->wrbuf, se, num, c->nmem, version); wrbuf_puts(c->wrbuf, "\n"); + must_generate_empty = 0; + } + if (must_generate_empty) + { + wrbuf_puts(c->wrbuf, "wrbuf, names[j]); + wrbuf_puts(c->wrbuf, "\"/>\n"); } } session_leave(se); @@ -1172,7 +1225,7 @@ void show_single_stop(struct session *se, struct record_cluster *rec) struct record_cluster **show_range_start(struct session *se, struct reclist_sortparms *sp, - int start, int *num, int *total, Odr_int *sumhits) + int start, int *num, int *total, Odr_int *sumhits, Odr_int *approx_hits) { struct record_cluster **recs; struct reclist_sortparms *spp; @@ -1186,7 +1239,8 @@ struct record_cluster **show_range_start(struct session *se, { *num = 0; *total = 0; - *sumhits = 0; + *sumhits = 0; + *approx_hits = 0; recs = 0; } else @@ -1205,9 +1259,11 @@ struct record_cluster **show_range_start(struct session *se, *total = reclist_get_num_records(se->reclist); *sumhits = 0; - for (l = se->clients_active; l; l = l->next) + *approx_hits = 0; + for (l = se->clients_active; l; l = l->next) { *sumhits += client_get_hits(l->client); - + *approx_hits += client_get_approximation(l->client); + } for (i = 0; i < start; i++) if (!reclist_read_record(se->reclist)) { @@ -1535,8 +1591,7 @@ int ingest_record(struct client *cl, const char *rec, if (!check_record_filter(root, sdb)) { - session_log(se, YLOG_LOG, "Filtered out record no %d from %s", - record_no, sdb->database->id); + session_log(se, YLOG_LOG, "Filtered out record no %d from %s", record_no, sdb->database->id); xmlFreeDoc(xdoc); return -2; } @@ -1557,6 +1612,7 @@ int ingest_record(struct client *cl, const char *rec, return ret; } +// Skip record on non-zero static int check_limit_local(struct client *cl, struct record *record, int record_no) @@ -1575,9 +1631,7 @@ static int check_limit_local(struct client *cl, char **values = 0; int i, num_v = 0; - const char *name = - client_get_facet_limit_local(cl, sdb, &l, nmem_tmp, &num_v, - &values); + const char *name = client_get_facet_limit_local(cl, sdb, &l, nmem_tmp, &num_v, &values); if (!name) break; @@ -1589,7 +1643,7 @@ static int check_limit_local(struct client *cl, } ser_md = &service->metadata[md_field_id]; rec_md = record->metadata[md_field_id]; - yaz_log(YLOG_LOG, "check limit local %s", name); + yaz_log(YLOG_DEBUG, "check limit local %s", name); for (i = 0; i < num_v; ) { if (rec_md) @@ -1604,10 +1658,10 @@ static int check_limit_local(struct client *cl, } else { - yaz_log(YLOG_LOG, "cmp: '%s' '%s'", - rec_md->data.text.disp, values[i]); + yaz_log(YLOG_DEBUG, "cmp: '%s' '%s'", rec_md->data.text.disp, values[i]); if (!strcmp(rec_md->data.text.disp, values[i])) { + // Value equals, should not be filtered. break; } } @@ -1619,6 +1673,7 @@ static int check_limit_local(struct client *cl, i++; } } + // At end , not match if (i == num_v) { skip_record = 1; @@ -1757,6 +1812,8 @@ static int ingest_to_cluster(struct client *cl, struct record_metadata *rec_md = 0; int md_field_id = -1; int sk_field_id = -1; + int rank = 0; + xmlChar *rank_str = 0; type = xmlGetProp(n, (xmlChar *) "type"); value = xmlNodeListGetString(xdoc, n->children, 1); @@ -1770,6 +1827,15 @@ static int ingest_to_cluster(struct client *cl, continue; ser_md = &service->metadata[md_field_id]; + + rank_str = xmlGetProp(n, (xmlChar *) "rank"); + if (rank_str) + { + rank = atoi((const char *) rank_str); + xmlFree(rank_str); + } + else + rank = ser_md->rank; if (ser_md->sortkey_offset >= 0) { @@ -1869,10 +1935,11 @@ static int ingest_to_cluster(struct client *cl, // ranking of _all_ fields enabled ... - if (ser_md->rank) + if (rank) + { relevance_countwords(se->relevance, cluster, - (char *) value, ser_md->rank, - ser_md->name); + (char *) value, rank, ser_md->name); + } // construct facets ... unless the client already has reported them if (ser_md->termlist && !client_has_facet(cl, (char *) type))