Remove a few log stms
[pazpar2-moved-to-github.git] / src / session.c
index 04807a8..9875515 100644 (file)
@@ -86,7 +86,8 @@ Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
 struct parameters global_parameters = 
 {
     0,   // dump_records
-    0    // debug_mode
+    0,   // debug_mode
+    0,   // predictable sessions
 };
 
 struct client_list {
@@ -94,6 +95,41 @@ struct client_list {
     struct client_list *next;
 };
 
+/* session counting (1) , disable client counting (0) */
+static YAZ_MUTEX g_session_mutex = 0;
+static int no_sessions = 0;
+static int no_session_total = 0;
+
+static int session_use(int delta)
+{
+    int sessions;
+    if (!g_session_mutex)
+        yaz_mutex_create(&g_session_mutex);
+    yaz_mutex_enter(g_session_mutex);
+    no_sessions += delta;
+    if (delta > 0)
+        no_session_total += delta;
+    sessions = no_sessions;
+    yaz_mutex_leave(g_session_mutex);
+    yaz_log(YLOG_DEBUG, "%s sesions=%d", delta == 0 ? "" : (delta > 0 ? "INC" : "DEC"), no_sessions);
+    return sessions;
+}
+
+int sessions_count(void) {
+    return session_use(0);
+}
+
+int  session_count_total(void) {
+    int total = 0;
+    if (!g_session_mutex)
+        return 0;
+    yaz_mutex_enter(g_session_mutex);
+    total = no_session_total;
+    yaz_mutex_leave(g_session_mutex);
+    return total;
+}
+
+
 static void log_xml_doc(xmlDoc *doc)
 {
     FILE *lf = yaz_log_file();
@@ -152,23 +188,47 @@ void pull_terms(NMEM nmem, struct ccl_rpn_node *n, char **termlist, int *num)
 void add_facet(struct session *s, const char *type, const char *value, int count)
 {
     struct conf_service *service = s->service;
-    pp2_relevance_token_t prt;
+    pp2_charset_token_t prt;
     const char *facet_component;
     WRBUF facet_wrbuf = wrbuf_alloc();
-    prt = pp2_relevance_tokenize(service->facet_pct);
-    
-    pp2_relevance_first(prt, value, 0);
-    while ((facet_component = pp2_relevance_token_next(prt)))
+    WRBUF display_wrbuf = wrbuf_alloc();
+    int i;
+    const char *icu_chain_id = 0;
+
+    for (i = 0; i < service->num_metadata; i++)
+        if (!strcmp((service->metadata + i)->name, type))
+            icu_chain_id = (service->metadata + i)->facetrule;
+    if (!icu_chain_id)
+        icu_chain_id = "facet";
+    prt = pp2_charset_token_create(service->charsets, icu_chain_id);
+    if (!prt)
+    {
+        yaz_log(YLOG_FATAL, "Unknown ICU chain '%s' for facet of type '%s'",
+                icu_chain_id, type);
+        wrbuf_destroy(facet_wrbuf);
+        wrbuf_destroy(display_wrbuf);
+        return;
+    }
+    pp2_charset_token_first(prt, value, 0);
+    while ((facet_component = pp2_charset_token_next(prt)))
     {
+        const char *display_component;
         if (*facet_component)
         {
             if (wrbuf_len(facet_wrbuf))
                 wrbuf_puts(facet_wrbuf, " ");
             wrbuf_puts(facet_wrbuf, facet_component);
         }
+        display_component = pp2_get_display(prt);
+        if (display_component)
+        {
+            if (wrbuf_len(display_wrbuf))
+                wrbuf_puts(display_wrbuf, " ");
+            wrbuf_puts(display_wrbuf, display_component);
+        }
     }
-    pp2_relevance_token_destroy(prt);
-    
+    pp2_charset_token_destroy(prt);
     if (wrbuf_len(facet_wrbuf))
     {
         int i;
@@ -181,6 +241,7 @@ void add_facet(struct session *s, const char *type, const char *value, int count
             {
                 session_log(s, YLOG_FATAL, "Too many termlists");
                 wrbuf_destroy(facet_wrbuf);
+                wrbuf_destroy(display_wrbuf);
                 return;
             }
             
@@ -190,12 +251,14 @@ void add_facet(struct session *s, const char *type, const char *value, int count
             s->num_termlists = i + 1;
         }
         
-        session_log(s, YLOG_DEBUG, "Session: facets for %s: %s norm:%s (%d)",
-                    type, value, wrbuf_cstr(facet_wrbuf), count);
-        termlist_insert(s->termlists[i].termlist, wrbuf_cstr(facet_wrbuf),
-                        count);
+#if 0
+        session_log(s, YLOG_DEBUG, "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);
     }
     wrbuf_destroy(facet_wrbuf);
+    wrbuf_destroy(display_wrbuf);
 }
 
 static xmlDoc *record_to_xml(struct session *se,
@@ -451,7 +514,7 @@ int session_set_watch(struct session *s, int what,
         ret = 0;
     }
     session_leave(s);
-    return 0;
+    return ret;
 }
 
 void session_alert_watch(struct session *s, int what)
@@ -476,7 +539,7 @@ void session_alert_watch(struct session *s, int what)
 
         session_leave(s);
         session_log(s, YLOG_DEBUG,
-                    "session_alert_watch: %d calling function: %p", what, fun);
+                    "Alert Watch: %d calling function: %p", what, fun);
         fun(data);
     }
     else
@@ -492,6 +555,7 @@ static void select_targets_callback(void *context, struct session_database *db)
     client_set_database(cl, db);
 
     client_set_session(cl, se);
+
     l = xmalloc(sizeof(*l));
     l->client = cl;
     l->next = se->clients;
@@ -548,16 +612,15 @@ int session_is_preferred_clients_ready(struct session *s)
     for (l = s->clients; l; l = l->next)
         if (client_is_active_preferred(l->client))
             res++;
-    session_log(s, YLOG_DEBUG, "Session has %d active preferred clients.", res);
+    session_log(s, YLOG_DEBUG, "Has %d active preferred clients.", res);
     return res == 0;
 }
 
-
-
 enum pazpar2_error_code search(struct session *se,
                                const char *query,
                                const char *startrecs, const char *maxrecs,
                                const char *filter,
+                               const char *limit,
                                const char **addinfo)
 {
     int live_channels = 0;
@@ -565,6 +628,7 @@ enum pazpar2_error_code search(struct session *se,
     int no_failed = 0;
     struct client_list *l;
     struct timeval tval;
+    facet_limits_t facet_limits;
 
     session_log(se, YLOG_DEBUG, "Search");
 
@@ -591,6 +655,13 @@ enum pazpar2_error_code search(struct session *se,
     
     tval.tv_sec += 5;
 
+    facet_limits = facet_limits_create(limit);
+    if (!facet_limits)
+    {
+        *addinfo = "limit";
+        session_leave(se);
+        return PAZPAR2_MALFORMED_PARAMETER_VALUE;
+    }
     for (l = se->clients; l; l = l->next)
     {
         struct client *cl = l->client;
@@ -601,7 +672,7 @@ enum pazpar2_error_code search(struct session *se,
             client_set_startrecs(cl, atoi(startrecs));
         if (prepare_session_database(se, client_get_database(cl)) < 0)
             ;
-        else if (client_parse_query(cl, query) < 0)
+        else if (client_parse_query(cl, query, facet_limits) < 0)
             no_failed++;
         else
         {
@@ -613,6 +684,7 @@ enum pazpar2_error_code search(struct session *se,
                 client_start_search(cl);
         }
     }
+    facet_limits_destroy(facet_limits);
     session_leave(se);
     if (no_working == 0)
     {
@@ -722,11 +794,10 @@ void session_apply_setting(struct session *se, char *dbname, char *setting,
     }
 }
 
-void destroy_session(struct session *se)
-{
+void session_destroy(struct session *se) {
     struct session_database *sdb;
-
-    session_log(se, YLOG_DEBUG, "Pazpar2 session destroy");
+    session_log(se, YLOG_DEBUG, "Destroying");
+    session_use(-1);
     session_remove_clients(se);
 
     for (sdb = se->databases; sdb; sdb = sdb->next)
@@ -737,9 +808,25 @@ void destroy_session(struct session *se)
     nmem_destroy(se->nmem);
     service_destroy(se->service);
     yaz_mutex_destroy(&se->session_mutex);
-    wrbuf_destroy(se->wrbuf);
 }
 
+/* Depreciated: use session_destroy */
+void destroy_session(struct session *se)
+{
+    session_destroy(se);
+}
+
+size_t session_get_memory_status(struct session *session) {
+    size_t session_nmem;
+    if (session == 0)
+        return 0;
+    session_enter(session);
+    session_nmem = nmem_total(session->nmem);
+    session_leave(session);
+    return session_nmem;
+}
+
+
 struct session *new_session(NMEM nmem, struct conf_service *service,
                             unsigned session_id)
 {
@@ -751,7 +838,7 @@ struct session *new_session(NMEM nmem, struct conf_service *service,
     sprintf(tmp_str, "session#%u", session_id);
 
     session->session_id = session_id;
-    session_log(session, YLOG_DEBUG, "New session");
+    session_log(session, YLOG_DEBUG, "New");
     session->service = service;
     session->relevance = 0;
     session->total_hits = 0;
@@ -763,7 +850,6 @@ struct session *new_session(NMEM nmem, struct conf_service *service,
     session->clients = 0;
     session->session_nmem = nmem;
     session->nmem = nmem_create();
-    session->wrbuf = wrbuf_alloc();
     session->databases = 0;
     for (i = 0; i <= SESSION_WATCH_MAX; i++)
     {
@@ -773,7 +859,7 @@ struct session *new_session(NMEM nmem, struct conf_service *service,
     session->normalize_cache = normalize_cache_create();
     session->session_mutex = 0;
     pazpar2_mutex_create(&session->session_mutex, tmp_str);
-
+    session_use(1);
     return session;
 }
 
@@ -804,14 +890,16 @@ struct hitsbytarget *hitsbytarget(struct session *se, int *count, NMEM nmem)
         res[*count].state = client_get_state_str(cl);
         res[*count].connected  = client_get_connection(cl) ? 1 : 0;
         session_settings_dump(se, client_get_database(cl), w);
-        res[*count].settings_xml = w;
+        res[*count].settings_xml = nmem_strdup(nmem, wrbuf_cstr(w));
+        wrbuf_destroy(w);
         (*count)++;
     }
     session_leave(se);
     return res;
 }
 
-struct termlist_score **termlist(struct session *se, const char *name, int *num)
+struct termlist_score **get_termlist_score(struct session *se,
+                                           const char *name, int *num)
 {
     int i;
     struct termlist_score **tl = 0;
@@ -1043,15 +1131,15 @@ static int get_mergekey_from_doc(xmlDoc *doc, xmlNode *root, const char *name,
                 if (value)
                 {
                     const char *norm_str;
-                    pp2_relevance_token_t prt =
-                        pp2_relevance_tokenize(service->mergekey_pct);
+                    pp2_charset_token_t prt =
+                        pp2_charset_token_create(service->charsets, "mergekey");
                     
-                    pp2_relevance_first(prt, (const char *) value, 0);
+                    pp2_charset_token_first(prt, (const char *) value, 0);
                     if (wrbuf_len(norm_wr) > 0)
                         wrbuf_puts(norm_wr, " ");
                     wrbuf_puts(norm_wr, name);
                     while ((norm_str =
-                            pp2_relevance_token_next(prt)))
+                            pp2_charset_token_next(prt)))
                     {
                         if (*norm_str)
                         {
@@ -1060,7 +1148,7 @@ static int get_mergekey_from_doc(xmlDoc *doc, xmlNode *root, const char *name,
                         }
                     }
                     xmlFree(value);
-                    pp2_relevance_token_destroy(prt);
+                    pp2_charset_token_destroy(prt);
                     no_found++;
                 }
             }
@@ -1082,11 +1170,11 @@ static const char *get_mergekey(xmlDoc *doc, struct client *cl, int record_no,
     if (mergekey)
     {
         const char *norm_str;
-        pp2_relevance_token_t prt =
-            pp2_relevance_tokenize(service->mergekey_pct);
+        pp2_charset_token_t prt =
+            pp2_charset_token_create(service->charsets, "mergekey");
 
-        pp2_relevance_first(prt, (const char *) mergekey, 0);
-        while ((norm_str = pp2_relevance_token_next(prt)))
+        pp2_charset_token_first(prt, (const char *) mergekey, 0);
+        while ((norm_str = pp2_charset_token_next(prt)))
         {
             if (*norm_str)
             {
@@ -1095,7 +1183,7 @@ static const char *get_mergekey(xmlDoc *doc, struct client *cl, int record_no,
                 wrbuf_puts(norm_wr, norm_str);
             }
         }
-        pp2_relevance_token_destroy(prt);
+        pp2_charset_token_destroy(prt);
         xmlFree(mergekey);
     }
     else
@@ -1207,6 +1295,7 @@ static int ingest_to_cluster(struct client *cl,
     \param nmem working NMEM
     \retval 0 OK
     \retval -1 failure
+    \retval -2 Filtered
 */
 int ingest_record(struct client *cl, const char *rec,
                   int record_no, NMEM nmem)
@@ -1226,10 +1315,10 @@ int ingest_record(struct client *cl, const char *rec,
     
     if (!check_record_filter(root, sdb))
     {
-        session_log(se, YLOG_WARN, "Filtered out record no %d from %s",
+        session_log(se, YLOG_LOG, "Filtered out record no %d from %s",
                     record_no, sdb->database->url);
         xmlFreeDoc(xdoc);
-        return -1;
+        return -2;
     }
     
     mergekey_norm = get_mergekey(xdoc, cl, record_no, service, nmem);
@@ -1271,8 +1360,7 @@ static int ingest_to_cluster(struct client *cl,
                                                     &se->total_merged);
 
     const char *use_term_factor_str = session_setting_oneval(sdb, PZ_TERMLIST_TERM_FACTOR);
-    int use_term_factor = 1;
-    // HACK: default to use term factor.
+    int use_term_factor = 0;
     int term_factor = 1; 
     if (use_term_factor_str && use_term_factor_str[0] != 0)
        use_term_factor =  atoi(use_term_factor_str);
@@ -1281,7 +1369,7 @@ static int ingest_to_cluster(struct client *cl,
         int hits = (int) client_get_hits(cl);
         term_factor = MAX(hits, maxrecs) /  MAX(1, maxrecs);
         assert(term_factor >= 1);
-        yaz_log(YLOG_DEBUG, "Using term factor %d ", term_factor); 
+        yaz_log(YLOG_DEBUG, "Using term factor: %d (%d / %d)", term_factor, MAX(hits, maxrecs), MAX(1, maxrecs));
     }
 
     if (!cluster)
@@ -1294,7 +1382,7 @@ static int ingest_to_cluster(struct client *cl,
     // now parsing XML record and adding data to cluster or record metadata
     for (n = root->children; n; n = n->next)
     {
-        pp2_relevance_token_t prt;
+        pp2_charset_token_t prt;
         if (type)
             xmlFree(type);
         if (value)
@@ -1361,16 +1449,15 @@ static int ingest_to_cluster(struct client *cl,
             // assign cluster or record based on merge action
             if (ser_md->merge == Metadata_merge_unique)
             {
-                struct record_metadata *mnode;
-                for (mnode = *wheretoput; mnode; mnode = mnode->next)
-                    if (!strcmp((const char *) mnode->data.text.disp, 
+                while (*wheretoput)
+                {
+                    if (!strcmp((const char *) (*wheretoput)->data.text.disp, 
                                 rec_md->data.text.disp))
                         break;
-                if (!mnode)
-                {
-                    rec_md->next = *wheretoput;
-                    *wheretoput = rec_md;
+                    wheretoput = &(*wheretoput)->next;
                 }
+                if (!*wheretoput)
+                    *wheretoput = rec_md;
             }
             else if (ser_md->merge == Metadata_merge_longest)
             {
@@ -1390,12 +1477,13 @@ static int ingest_to_cluster(struct client *cl,
                                 nmem_malloc(se->nmem, 
                                             sizeof(union data_types));
                          
-                        prt = pp2_relevance_tokenize(service->sort_pct);
+                        prt =
+                            pp2_charset_token_create(service->charsets, "sort");
 
-                        pp2_relevance_first(prt, rec_md->data.text.disp,
-                                            skip_article);
+                        pp2_charset_token_first(prt, rec_md->data.text.disp,
+                                                skip_article);
 
-                        pp2_relevance_token_next(prt);
+                        pp2_charset_token_next(prt);
                          
                         sort_str = pp2_get_sort(prt);
                          
@@ -1409,13 +1497,14 @@ static int ingest_to_cluster(struct client *cl,
                         }
                         cluster->sortkeys[sk_field_id]->text.sort = 
                             nmem_strdup(se->nmem, sort_str);
-                        pp2_relevance_token_destroy(prt);
+                        pp2_charset_token_destroy(prt);
                     }
                 }
             }
             else if (ser_md->merge == Metadata_merge_all)
             {
-                rec_md->next = *wheretoput;
+                while (*wheretoput)
+                    wheretoput = &(*wheretoput)->next;
                 *wheretoput = rec_md;
             }
             else if (ser_md->merge == Metadata_merge_range)
@@ -1448,7 +1537,6 @@ static int ingest_to_cluster(struct client *cl,
             // construct facets ... unless the client already has reported them
             if (ser_md->termlist && !client_has_facet(cl, (char *) type))
             {
-
                 if (ser_md->type == Metadata_type_year)
                 {
                     char year[64];
@@ -1496,7 +1584,7 @@ void session_log(struct session *s, int level, const char *fmt, ...)
     va_start(ap, fmt);
 
     yaz_vsnprintf(buf, sizeof(buf)-30, fmt, ap);
-    yaz_log(level, "%u %s", s->session_id, buf);
+    yaz_log(level, "Session (%u): %s", s->session_id, buf);
 
     va_end(ap);
 }