X-Git-Url: http://lists.indexdata.com/cgi-bin?a=blobdiff_plain;f=src%2Fsession.c;h=bb8cc7f2ceccfe346ba1cdbc3dd777961c98edd0;hb=ca6a981e9db639044914a8dbf32f50b6e09bb271;hp=3fb79b9c9f084ad68886220187b174841defb5a7;hpb=5b9e96b0d4272562f4940736b83489a707ad81ce;p=pazpar2-moved-to-github.git diff --git a/src/session.c b/src/session.c index 3fb79b9..bb8cc7f 100644 --- a/src/session.c +++ b/src/session.c @@ -78,14 +78,12 @@ Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA #include -#define TERMLIST_HIGH_SCORE 25 - #define MAX_CHUNK 15 #define MAX(a,b) ((a)>(b)?(a):(b)) // Note: Some things in this structure will eventually move to configuration -struct parameters global_parameters = +struct parameters global_parameters = { 0, // dump_records 0, // debug_mode @@ -213,7 +211,7 @@ void add_facet(struct session *s, const char *type, const char *value, int count WRBUF display_wrbuf = wrbuf_alloc(); session_normalize_facet(s, type, value, display_wrbuf, facet_wrbuf); - + if (wrbuf_len(facet_wrbuf)) { int i; @@ -229,15 +227,14 @@ void add_facet(struct session *s, const char *type, const char *value, int count wrbuf_destroy(display_wrbuf); return; } - + s->termlists[i].name = nmem_strdup(s->nmem, type); - s->termlists[i].termlist - = termlist_create(s->nmem, TERMLIST_HIGH_SCORE); + s->termlists[i].termlist = termlist_create(s->nmem); s->num_termlists = i + 1; } - + #if 0 - session_log(s, YLOG_DEBUG, "Facets for %s: %s norm:%s (%d)", type, value, wrbuf_cstr(facet_wrbuf), count); + session_log(s, YLOG_LOG, "Facets for %s: %s norm:%s (%d)", type, value, wrbuf_cstr(facet_wrbuf), count); #endif termlist_insert(s->termlists[i].termlist, wrbuf_cstr(display_wrbuf), wrbuf_cstr(facet_wrbuf), count); @@ -345,9 +342,9 @@ static xmlDoc *normalize_record(struct session *se, if (rdoc) { char *parms[MAX_XSLT_ARGS*2+1]; - + insert_settings_parameters(sdb, service, parms, nmem); - + if (normalize_record_transform(sdb->map, &rdoc, (const char **)parms)) { session_log(se, YLOG_WARN, "Normalize failed from %s", @@ -356,10 +353,10 @@ static xmlDoc *normalize_record(struct session *se, else { insert_settings_values(sdb, rdoc, service); - + if (global_parameters.dump_records) { - session_log(se, YLOG_LOG, "Normalized record from %s", + session_log(se, YLOG_LOG, "Normalized record from %s", sdb->database->id); log_xml_doc(rdoc); } @@ -408,40 +405,43 @@ 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; } @@ -458,7 +458,7 @@ static void session_watch_cancel(void *data, struct http_channel *c, } // set watch. Returns 0=OK, -1 if watch is already set -int session_set_watch(struct session *s, int what, +int session_set_watch(struct session *s, int what, session_watchfun fun, void *data, struct http_channel *chan) { @@ -468,7 +468,7 @@ int session_set_watch(struct session *s, int what, ret = -1; else { - + s->watchlist[what].fun = fun; s->watchlist[what].data = data; s->watchlist[what].obs = http_add_observer(chan, &s->watchlist[what], @@ -554,7 +554,7 @@ static void session_reset_active_clients(struct session *se, { struct client_list *l_next = l->next; - client_lock(l->client); + client_lock(l->client); client_set_session(l->client, 0); /* mark client inactive */ client_unlock(l->client); @@ -588,7 +588,7 @@ static void session_remove_cached_clients(struct session *se) } // Associates a set of clients with a session; -// Note: Session-databases represent databases with per-session +// Note: Session-databases represent databases with per-session // setting overrides static int select_targets(struct session *se, const char *filter) { @@ -619,41 +619,79 @@ 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, int position) +{ + 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->position = position; + se->sorted_results->next = 0; + + session_log(se, YLOG_DEBUG, "clear_set session_sort: field=%s increasing=%d position=%d configured", + sort_field, increasing, position); + + se->reclist = reclist_create(se->nmem); +} + +void session_sort(struct session *se, const char *field, int increasing, + int position) { struct session_sorted_results *sr; struct client_list *l; session_enter(se); + yaz_log(YLOG_LOG, "session_sort field=%s increasing=%d position=%d", field, increasing, position); /* 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) + if (!strcmp(field, sr->field) && increasing == sr->increasing && sr->position == position) break; } if (sr) { - yaz_log(YLOG_LOG, "search_sort: field=%s increasing=%d already fetched", - field, increasing); + session_log(se, YLOG_DEBUG, "search_sort: field=%s increasing=%d position=%d already fetched", + field, increasing, position); session_leave(se); return; } - 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; - + session_log(se, YLOG_DEBUG, "search_sort: field=%s increasing=%d position=%d must fetch", + field, increasing, position); + if (position) + { + yaz_log(YLOG_DEBUG, "Reset results due to position"); + session_clear_set(se, field, increasing, position); + } + else { + sr = nmem_malloc(se->nmem, sizeof(*sr)); + sr->field = nmem_strdup(se->nmem, field); + sr->increasing = increasing; + sr->position = position; + sr->next = se->sorted_results; + se->sorted_results = sr; + } + yaz_log(YLOG_DEBUG, "Restarting search for clients due to change in sort order"); + for (l = se->clients_active; l; l = l->next) { struct client *cl = l->client; if (client_get_state(cl) == Client_Connecting || client_get_state(cl) == Client_Idle || - client_get_state(cl) == Client_Working) + client_get_state(cl) == Client_Working) { + yaz_log(YLOG_DEBUG, "Client %s: Restarting search due to change in sort order", client_get_id(cl)); client_start_search(cl); + } } session_leave(se); } @@ -665,7 +703,7 @@ enum pazpar2_error_code session_search(struct session *se, const char *filter, const char *limit, const char **addinfo, - const char *sort_field, int increasing) + struct reclist_sortparms *sp) { int live_channels = 0; int no_working = 0; @@ -683,32 +721,21 @@ enum pazpar2_error_code session_search(struct session *se, session_remove_cached_clients(se); else session_reset_active_clients(se, 0); - + session_enter(se); - reclist_destroy(se->reclist); - se->reclist = 0; se->settings_modified = 0; + session_clear_set(se, sp->name, sp->increasing, sp->type == Metadata_sortkey_position); 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); - + tval.tv_sec += 5; facet_limits = facet_limits_create(limit); @@ -732,11 +759,13 @@ enum pazpar2_error_code session_search(struct session *se, continue; parse_ret = client_parse_query(cl, query, facet_limits, startrecs, - maxrecs); + maxrecs, se->service->ccl_bibset); if (parse_ret == -1) no_failed_query++; else if (parse_ret == -2) no_failed_limit++; + else if (parse_ret < 0) + no_working++; /* other error, such as bad CCL map */ else { int r = @@ -746,12 +775,12 @@ enum pazpar2_error_code session_search(struct session *se, &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 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++; @@ -775,7 +804,7 @@ 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; } @@ -787,7 +816,7 @@ static void session_init_databases_fun(void *context, struct database *db) int i; new->database = db; - + new->map = 0; assert(db->settings); new->settings = nmem_malloc(se->session_nmem, @@ -818,7 +847,7 @@ void session_init_databases(struct session *se) // Probably session_init_databases_fun should be refactored instead of // called here. -static struct session_database *load_session_database(struct session *se, +static struct session_database *load_session_database(struct session *se, char *id) { struct database *db = new_database_inherit_settings(id, se->session_nmem, se->service->settings); @@ -829,7 +858,7 @@ static struct session_database *load_session_database(struct session *se, } // Find an existing session database. If not found, load it -static struct session_database *find_session_database(struct session *se, +static struct session_database *find_session_database(struct session *se, char *id) { struct session_database *sdb; @@ -885,6 +914,10 @@ 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); @@ -988,23 +1021,6 @@ struct hitsbytarget *get_hitsbytarget(struct session *se, int *count, NMEM nmem) session_leave(se); return p; } - -struct termlist_score **get_termlist_score(struct session *se, - const char *name, int *num) -{ - int i; - struct termlist_score **tl = 0; - - session_enter(se); - for (i = 0; i < se->num_termlists; i++) - if (!strcmp((const char *) se->termlists[i].name, name)) - { - tl = termlist_highscore(se->termlists[i].termlist, num); - break; - } - session_leave(se); - return tl; -} // Compares two hitsbytarget nodes by hitcount static int cmp_ht(const void *p1, const void *p2) @@ -1037,20 +1053,20 @@ static int targets_termlist_nb(WRBUF wrbuf, struct session *se, int num, { // do only print terms which have display names - + wrbuf_puts(wrbuf, "\n"); wrbuf_puts(wrbuf, ""); wrbuf_xmlputs(wrbuf, ht[i].id); wrbuf_puts(wrbuf, "\n"); - + wrbuf_puts(wrbuf, ""); if (!ht[i].name || !ht[i].name[0]) wrbuf_xmlputs(wrbuf, "NO TARGET NAME"); else wrbuf_xmlputs(wrbuf, ht[i].name); wrbuf_puts(wrbuf, "\n"); - + wrbuf_printf(wrbuf, "" ODR_INT_PRINTF "\n", ht[i].hits); @@ -1064,8 +1080,8 @@ static int targets_termlist_nb(WRBUF wrbuf, struct session *se, int num, wrbuf_puts(wrbuf, ""); wrbuf_xmlputs(wrbuf, ht[i].state); wrbuf_puts(wrbuf, "\n"); - - wrbuf_printf(wrbuf, "%d\n", + + wrbuf_printf(wrbuf, "%d\n", ht[i].diagnostic); wrbuf_puts(wrbuf, "\n"); } @@ -1105,7 +1121,8 @@ void perform_termlist(struct http_channel *c, struct session *se, wrbuf_puts(c->wrbuf, "\">\n"); must_generate_empty = 0; - p = termlist_highscore(se->termlists[i].termlist, &len); + p = termlist_highscore(se->termlists[i].termlist, &len, + nmem_tmp); if (p) { int i; @@ -1114,14 +1131,14 @@ void perform_termlist(struct http_channel *c, struct session *se, // prevent sending empty term elements if (!p[i]->display_term || !p[i]->display_term[0]) continue; - + wrbuf_puts(c->wrbuf, ""); wrbuf_puts(c->wrbuf, ""); wrbuf_xmlputs(c->wrbuf, p[i]->display_term); wrbuf_puts(c->wrbuf, ""); - - wrbuf_printf(c->wrbuf, - "%d", + + wrbuf_printf(c->wrbuf, + "%d", p[i]->frequency); wrbuf_puts(c->wrbuf, "\n"); } @@ -1159,7 +1176,7 @@ void report_nmem_stats(void) nmem_get_memory_in_use(&in_use); nmem_get_memory_free(&is_free); - yaz_log(YLOG_LOG, "nmem stat: use=%ld free=%ld", + yaz_log(YLOG_LOG, "nmem stat: use=%ld free=%ld", (long) in_use, (long) is_free); } #endif @@ -1198,13 +1215,13 @@ void show_single_stop(struct session *se, struct record_cluster *rec) } struct record_cluster **show_range_start(struct session *se, - struct reclist_sortparms *sp, + struct reclist_sortparms *sp, int start, int *num, int *total, Odr_int *sumhits, Odr_int *approx_hits) { struct record_cluster **recs; struct reclist_sortparms *spp; int i; -#if USE_TIMING +#if USE_TIMING yaz_timing_t t = yaz_timing_create(); #endif session_enter(se); @@ -1213,14 +1230,14 @@ struct record_cluster **show_range_start(struct session *se, { *num = 0; *total = 0; - *sumhits = 0; + *sumhits = 0; *approx_hits = 0; recs = 0; } else { struct client_list *l; - + for (spp = sp; spp; spp = spp->next) if (spp->type == Metadata_sortkey_relevance) { @@ -1228,7 +1245,7 @@ struct record_cluster **show_range_start(struct session *se, break; } reclist_sort(se->reclist, sp); - + reclist_enter(se->reclist); *total = reclist_get_num_records(se->reclist); @@ -1245,7 +1262,7 @@ struct record_cluster **show_range_start(struct session *se, recs = 0; break; } - + for (i = 0; i < *num; i++) { struct record_cluster *r = reclist_read_record(se->reclist); @@ -1260,7 +1277,7 @@ struct record_cluster **show_range_start(struct session *se, } #if USE_TIMING yaz_timing_stop(t); - yaz_log(YLOG_LOG, "show %6.5f %3.2f %3.2f", + yaz_log(YLOG_LOG, "show %6.5f %3.2f %3.2f", yaz_timing_get_real(t), yaz_timing_get_user(t), yaz_timing_get_sys(t)); yaz_timing_destroy(&t); @@ -1308,7 +1325,7 @@ static struct record_metadata *record_metadata_init( { struct record_metadata *rec_md = record_metadata_create(nmem); struct record_metadata_attr **attrp = &rec_md->attributes; - + for (; attr; attr = attr->next) { if (attr->children && attr->children->content) @@ -1333,7 +1350,7 @@ static struct record_metadata *record_metadata_init( char *p = nmem_strdup(nmem, value); p = normalize7bit_generic(p, " ,/.:(["); - + rec_md->data.text.disp = p; rec_md->data.text.sort = 0; } @@ -1378,7 +1395,7 @@ static int get_mergekey_from_doc(xmlDoc *doc, xmlNode *root, const char *name, const char *norm_str; pp2_charset_token_t prt = pp2_charset_token_create(service->charsets, "mergekey"); - + pp2_charset_token_first(prt, (const char *) value, 0); if (wrbuf_len(norm_wr) > 0) wrbuf_puts(norm_wr, " "); @@ -1444,7 +1461,7 @@ static const char *get_mergekey(xmlDoc *doc, struct client *cl, int record_no, service, norm_wr); if (r == 0 && ser_md->mergekey == Metadata_mergekey_required) { - /* no mergekey on this one and it is required.. + /* no mergekey on this one and it is required.. Generate unique key instead */ wrbuf_rewind(norm_wr); break; @@ -1470,7 +1487,7 @@ static const char *get_mergekey(xmlDoc *doc, struct client *cl, int record_no, return mergekey_norm; } -/** \brief see if metadata for pz:recordfilter exists +/** \brief see if metadata for pz:recordfilter exists \param root xml root element of normalized record \param sdb session database for client \retval 0 if there is no metadata for pz:recordfilter @@ -1479,7 +1496,7 @@ static const char *get_mergekey(xmlDoc *doc, struct client *cl, int record_no, If there is no pz:recordfilter defined, this function returns 1 as well. */ - + static int check_record_filter(xmlNode *root, struct session_database *sdb) { int match = 0; @@ -1557,19 +1574,19 @@ int ingest_record(struct client *cl, const char *rec, xmlDoc *xdoc = normalize_record(se, sdb, service, rec, nmem); xmlNode *root; const char *mergekey_norm; - + if (!xdoc) return -1; - + root = xmlDocGetRootElement(xdoc); - + if (!check_record_filter(root, sdb)) { session_log(se, YLOG_LOG, "Filtered out record no %d from %s", record_no, sdb->database->id); xmlFreeDoc(xdoc); return -2; } - + mergekey_norm = get_mergekey(xdoc, cl, record_no, service, nmem); if (!mergekey_norm) { @@ -1581,7 +1598,7 @@ int ingest_record(struct client *cl, const char *rec, if (client_get_session(cl) == se) ret = ingest_to_cluster(cl, xdoc, root, record_no, mergekey_norm); session_leave(se); - + xmlFreeDoc(xdoc); return ret; } @@ -1604,11 +1621,11 @@ static int check_limit_local(struct client *cl, int md_field_id; char **values = 0; int i, num_v = 0; - + const char *name = client_get_facet_limit_local(cl, sdb, &l, nmem_tmp, &num_v, &values); if (!name) break; - + md_field_id = conf_service_metadata_field_id(service, name); if (md_field_id < 0) { @@ -1617,22 +1634,22 @@ 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) { - if (ser_md->type == Metadata_type_year + if (ser_md->type == Metadata_type_year || ser_md->type == Metadata_type_date) { int y = atoi(values[i]); - if (y >= rec_md->data.number.min + if (y >= rec_md->data.number.min && y <= rec_md->data.number.max) break; } 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. @@ -1657,7 +1674,7 @@ static int check_limit_local(struct client *cl, nmem_destroy(nmem_tmp); return skip_record; } - + static int ingest_to_cluster(struct client *cl, xmlDoc *xdoc, xmlNode *root, @@ -1672,7 +1689,7 @@ static int ingest_to_cluster(struct client *cl, int term_factor = 1; struct record_cluster *cluster; struct session_database *sdb = client_get_database(cl); - struct record *record = record_create(se->nmem, + struct record *record = record_create(se->nmem, service->num_metadata, service->num_sortkeys, cl, record_no); @@ -1684,7 +1701,7 @@ static int ingest_to_cluster(struct client *cl, if (value) xmlFree(value); type = value = 0; - + if (n->type != XML_ELEMENT_NODE) continue; if (!strcmp((const char *) n->name, "metadata")) @@ -1693,26 +1710,26 @@ static int ingest_to_cluster(struct client *cl, struct record_metadata **wheretoput = 0; struct record_metadata *rec_md = 0; int md_field_id = -1; - + type = xmlGetProp(n, (xmlChar *) "type"); value = xmlNodeListGetString(xdoc, n->children, 1); - + if (!type || !value || !*value) continue; - - md_field_id + + md_field_id = conf_service_metadata_field_id(service, (const char *) type); if (md_field_id < 0) { if (se->number_of_warnings_unknown_metadata == 0) { - session_log(se, YLOG_WARN, + session_log(se, YLOG_WARN, "Ignoring unknown metadata element: %s", type); } se->number_of_warnings_unknown_metadata++; continue; } - + ser_md = &service->metadata[md_field_id]; // non-merged metadata @@ -1765,7 +1782,7 @@ static int ingest_to_cluster(struct client *cl, relevance_newrec(se->relevance, cluster); - + // now parsing XML record and adding data to cluster or record metadata for (n = root->children; n; n = n->next) { @@ -1775,7 +1792,7 @@ static int ingest_to_cluster(struct client *cl, if (value) xmlFree(value); type = value = 0; - + if (n->type != XML_ELEMENT_NODE) continue; if (!strcmp((const char *) n->name, "metadata")) @@ -1786,20 +1803,22 @@ static int ingest_to_cluster(struct client *cl, struct record_metadata *rec_md = 0; int md_field_id = -1; int sk_field_id = -1; - + const char *rank; + xmlChar *xml_rank; + type = xmlGetProp(n, (xmlChar *) "type"); value = xmlNodeListGetString(xdoc, n->children, 1); - + if (!type || !value || !*value) continue; - - md_field_id + + md_field_id = conf_service_metadata_field_id(service, (const char *) type); if (md_field_id < 0) continue; - + ser_md = &service->metadata[md_field_id]; - + if (ser_md->sortkey_offset >= 0) { sk_field_id = ser_md->sortkey_offset; @@ -1812,6 +1831,9 @@ static int ingest_to_cluster(struct client *cl, if (!rec_md) continue; + xml_rank = xmlGetProp(n, (xmlChar *) "rank"); + rank = xml_rank ? (const char *) xml_rank : ser_md->rank; + wheretoput = &cluster->metadata[md_field_id]; // and polulate with data: @@ -1820,7 +1842,7 @@ static int ingest_to_cluster(struct client *cl, { while (*wheretoput) { - if (!strcmp((const char *) (*wheretoput)->data.text.disp, + if (!strcmp((const char *) (*wheretoput)->data.text.disp, rec_md->data.text.disp)) break; wheretoput = &(*wheretoput)->next; @@ -1830,22 +1852,22 @@ static int ingest_to_cluster(struct client *cl, } else if (ser_md->merge == Metadata_merge_longest) { - if (!*wheretoput - || strlen(rec_md->data.text.disp) + if (!*wheretoput + || strlen(rec_md->data.text.disp) > strlen((*wheretoput)->data.text.disp)) { *wheretoput = rec_md; if (ser_sk) { const char *sort_str = 0; - int skip_article = + int skip_article = ser_sk->type == Metadata_sortkey_skiparticle; if (!cluster->sortkeys[sk_field_id]) - cluster->sortkeys[sk_field_id] = - nmem_malloc(se->nmem, + cluster->sortkeys[sk_field_id] = + nmem_malloc(se->nmem, sizeof(union data_types)); - + prt = pp2_charset_token_create(service->charsets, "sort"); @@ -1853,18 +1875,18 @@ static int ingest_to_cluster(struct client *cl, skip_article); pp2_charset_token_next(prt); - + sort_str = pp2_get_sort(prt); - - cluster->sortkeys[sk_field_id]->text.disp = + + cluster->sortkeys[sk_field_id]->text.disp = rec_md->data.text.disp; if (!sort_str) { sort_str = rec_md->data.text.disp; - session_log(se, YLOG_WARN, + session_log(se, YLOG_WARN, "Could not make sortkey. Bug #1858"); } - cluster->sortkeys[sk_field_id]->text.sort = + cluster->sortkeys[sk_field_id]->text.sort = nmem_strdup(se->nmem, sort_str); pp2_charset_token_destroy(prt); } @@ -1882,7 +1904,7 @@ static int ingest_to_cluster(struct client *cl, { *wheretoput = rec_md; if (ser_sk) - cluster->sortkeys[sk_field_id] + cluster->sortkeys[sk_field_id] = &rec_md->data; } else @@ -1896,12 +1918,12 @@ static int ingest_to_cluster(struct client *cl, } } - - // ranking of _all_ fields enabled ... - if (ser_md->rank) - relevance_countwords(se->relevance, cluster, - (char *) value, ser_md->rank, - ser_md->name); + // ranking of _all_ fields enabled ... + if (rank) + { + relevance_countwords(se->relevance, cluster, + (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)) @@ -1923,6 +1945,8 @@ static int ingest_to_cluster(struct client *cl, } // cleaning up + if (xml_rank) + xmlFree(xml_rank); xmlFree(type); xmlFree(value); type = value = 0;