Rename func termlist->get_termlist_score
[pazpar2-moved-to-github.git] / src / session.c
index 76364aa..545f0fc 100644 (file)
@@ -1,5 +1,5 @@
 /* This file is part of Pazpar2.
-   Copyright (C) 2006-2010 Index Data
+   Copyright (C) 2006-2011 Index Data
 
 Pazpar2 is free software; you can redistribute it and/or modify it under
 the terms of the GNU General Public License as published by the Free
@@ -41,6 +41,7 @@ Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
 #include <signal.h>
 #include <ctype.h>
 #include <assert.h>
+#include <math.h>
 
 #include <yaz/marcdisp.h>
 #include <yaz/comstack.h>
@@ -79,11 +80,14 @@ Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
 
 #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 = 
 {
     0,   // dump_records
-    0    // debug_mode
+    0,   // debug_mode
+    0,   // predictable sessions
 };
 
 struct client_list {
@@ -91,6 +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();
@@ -148,32 +187,56 @@ 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)
 {
-    int i;
-
-    if (!*value)
-        return;
-    for (i = 0; i < s->num_termlists; i++)
-        if (!strcmp(s->termlists[i].name, type))
-            break;
-    if (i == s->num_termlists)
+    struct conf_service *service = s->service;
+    pp2_relevance_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)))
     {
-        if (i == SESSION_MAX_TERMLISTS)
+        if (*facet_component)
         {
-            yaz_log(YLOG_FATAL, "Too many termlists");
-            return;
+            if (wrbuf_len(facet_wrbuf))
+                wrbuf_puts(facet_wrbuf, " ");
+            wrbuf_puts(facet_wrbuf, facet_component);
         }
-
-        s->termlists[i].name = nmem_strdup(s->nmem, type);
-        s->termlists[i].termlist 
-            = termlist_create(s->nmem, TERMLIST_HIGH_SCORE);
-        s->num_termlists = i + 1;
     }
-    yaz_log(YLOG_DEBUG, "Session: facets for %s: %s (%d)", type, value, count);
-
-    termlist_insert(s->termlists[i].termlist, value, count);
+    pp2_relevance_token_destroy(prt);
+    
+    if (wrbuf_len(facet_wrbuf))
+    {
+        int i;
+        for (i = 0; i < s->num_termlists; i++)
+            if (!strcmp(s->termlists[i].name, type))
+                break;
+        if (i == s->num_termlists)
+        {
+            if (i == SESSION_MAX_TERMLISTS)
+            {
+                session_log(s, YLOG_FATAL, "Too many termlists");
+                wrbuf_destroy(facet_wrbuf);
+                return;
+            }
+            
+            s->termlists[i].name = nmem_strdup(s->nmem, type);
+            s->termlists[i].termlist 
+                = termlist_create(s->nmem, TERMLIST_HIGH_SCORE);
+            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);
+#endif
+        termlist_insert(s->termlists[i].termlist, wrbuf_cstr(facet_wrbuf),
+                        count);
+    }
+    wrbuf_destroy(facet_wrbuf);
 }
 
-static xmlDoc *record_to_xml(struct session_database *sdb, const char *rec)
+static xmlDoc *record_to_xml(struct session *se,
+                             struct session_database *sdb, const char *rec)
 {
     struct database *db = sdb->database;
     xmlDoc *rdoc = 0;
@@ -182,14 +245,14 @@ static xmlDoc *record_to_xml(struct session_database *sdb, const char *rec)
 
     if (!rdoc)
     {
-        yaz_log(YLOG_FATAL, "Non-wellformed XML received from %s",
-                db->url);
+        session_log(se, YLOG_FATAL, "Non-wellformed XML received from %s",
+                    db->url);
         return 0;
     }
 
     if (global_parameters.dump_records)
     {
-        yaz_log(YLOG_LOG, "Un-normalized record from %s", db->url);
+        session_log(se, YLOG_LOG, "Un-normalized record from %s", db->url);
         log_xml_doc(rdoc);
     }
 
@@ -261,11 +324,12 @@ static void insert_settings_values(struct session_database *sdb, xmlDoc *doc,
     }
 }
 
-static xmlDoc *normalize_record(struct session_database *sdb,
+static xmlDoc *normalize_record(struct session *se,
+                                struct session_database *sdb,
                                 struct conf_service *service,
                                 const char *rec, NMEM nmem)
 {
-    xmlDoc *rdoc = record_to_xml(sdb, rec);
+    xmlDoc *rdoc = record_to_xml(se, sdb, rec);
 
     if (rdoc)
     {
@@ -275,7 +339,8 @@ static xmlDoc *normalize_record(struct session_database *sdb,
         
         if (normalize_record_transform(sdb->map, &rdoc, (const char **)parms))
         {
-            yaz_log(YLOG_WARN, "Normalize failed from %s", sdb->database->url);
+            session_log(se, YLOG_WARN, "Normalize failed from %s",
+                        sdb->database->url);
         }
         else
         {
@@ -283,8 +348,8 @@ static xmlDoc *normalize_record(struct session_database *sdb,
             
             if (global_parameters.dump_records)
             {
-                yaz_log(YLOG_LOG, "Normalized record from %s", 
-                        sdb->database->url);
+                session_log(se, YLOG_LOG, "Normalized record from %s", 
+                            sdb->database->url);
                 log_xml_doc(rdoc);
             }
         }
@@ -336,7 +401,7 @@ static int prepare_map(struct session *se, struct session_database *sdb)
 
     if (!sdb->settings)
     {
-        yaz_log(YLOG_WARN, "No settings on %s", sdb->database->url);
+        session_log(se, YLOG_WARN, "No settings on %s", sdb->database->url);
         return -1;
     }
     if ((s = session_setting_oneval(sdb, PZ_XSLT)))
@@ -362,7 +427,8 @@ static int prepare_map(struct session *se, struct session_database *sdb)
             }
             else
             {
-                yaz_log(YLOG_WARN, "No pz:requestsyntax for auto stylesheet");
+                session_log(se, YLOG_WARN,
+                            "No pz:requestsyntax for auto stylesheet");
             }
         }
         sdb->map = normalize_cache_get(se->normalize_cache,
@@ -380,7 +446,7 @@ static int prepare_session_database(struct session *se,
 {
     if (!sdb->settings)
     {
-        yaz_log(YLOG_WARN, 
+        session_log(se, YLOG_WARN, 
                 "No settings associated with %s", sdb->database->url);
         return -1;
     }
@@ -446,7 +512,8 @@ void session_alert_watch(struct session *s, int what)
         s->watchlist[what].obs = 0;
 
         session_leave(s);
-        yaz_log(YLOG_DEBUG, "session_alert_watch: %d calling function: %p", what, fun);
+        session_log(s, YLOG_DEBUG,
+                    "Alert Watch: %d calling function: %p", what, fun);
         fun(data);
     }
     else
@@ -462,6 +529,7 @@ 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;
@@ -518,12 +586,10 @@ 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++;
-    yaz_log(YLOG_DEBUG, "%p Session has %d active preferred clients.", s, 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,
@@ -536,7 +602,7 @@ enum pazpar2_error_code search(struct session *se,
     struct client_list *l;
     struct timeval tval;
 
-    yaz_log(YLOG_DEBUG, "Search");
+    session_log(se, YLOG_DEBUG, "Search");
 
     *addinfo = 0;
 
@@ -692,11 +758,10 @@ 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;
-
-    yaz_log(YLOG_DEBUG, "%p Pazpar2 session destroy", se);
+    session_log(se, YLOG_DEBUG, "Destroying");
+    session_use(-1);
     session_remove_clients(se);
 
     for (sdb = se->databases; sdb; sdb = sdb->next)
@@ -710,14 +775,35 @@ void destroy_session(struct session *se)
     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,
-                            const char *name)
+                            unsigned session_id)
 {
     int i;
     struct session *session = nmem_malloc(nmem, sizeof(*session));
 
-    yaz_log(YLOG_DEBUG, "%p New Pazpar2 session", session);
+    char tmp_str[50];
+
+    sprintf(tmp_str, "session#%u", session_id);
 
+    session->session_id = session_id;
+    session_log(session, YLOG_DEBUG, "New");
     session->service = service;
     session->relevance = 0;
     session->total_hits = 0;
@@ -738,8 +824,8 @@ 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, name);
-
+    pazpar2_mutex_create(&session->session_mutex, tmp_str);
+    session_use(1);
     return session;
 }
 
@@ -770,14 +856,16 @@ 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;
@@ -810,22 +898,25 @@ struct record_cluster *show_single_start(struct session *se, const char *id,
                                          struct record_cluster **prev_r,
                                          struct record_cluster **next_r)
 {
-    struct record_cluster *r;
+    struct record_cluster *r = 0;
 
     session_enter(se);
-    reclist_enter(se->reclist);
     *prev_r = 0;
     *next_r = 0;
-    while ((r = reclist_read_record(se->reclist)))
+    if (se->reclist)
     {
-        if (!strcmp(r->recid, id))
+        reclist_enter(se->reclist);
+        while ((r = reclist_read_record(se->reclist)))
         {
-            *next_r = reclist_read_record(se->reclist);
-            break;
+            if (!strcmp(r->recid, id))
+            {
+                *next_r = reclist_read_record(se->reclist);
+                break;
+            }
+            *prev_r = r;
         }
-        *prev_r = r;
+        reclist_leave(se->reclist);
     }
-    reclist_leave(se->reclist);
     if (!r)
         session_leave(se);
     return r;
@@ -1170,6 +1261,7 @@ 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)
@@ -1178,7 +1270,7 @@ int ingest_record(struct client *cl, const char *rec,
     int ret = 0;
     struct session_database *sdb = client_get_database(cl);
     struct conf_service *service = se->service;
-    xmlDoc *xdoc = normalize_record(sdb, service, rec, nmem);
+    xmlDoc *xdoc = normalize_record(se, sdb, service, rec, nmem);
     xmlNode *root;
     const char *mergekey_norm;
     
@@ -1189,16 +1281,16 @@ int ingest_record(struct client *cl, const char *rec,
     
     if (!check_record_filter(root, sdb))
     {
-        yaz_log(YLOG_WARN, "Filtered out record no %d from %s", record_no,
-                sdb->database->url);
+        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);
     if (!mergekey_norm)
     {
-        yaz_log(YLOG_WARN, "Got no mergekey");
+        session_log(se, YLOG_WARN, "Got no mergekey");
         xmlFreeDoc(xdoc);
         return -1;
     }
@@ -1232,11 +1324,25 @@ static int ingest_to_cluster(struct client *cl,
                                                     record,
                                                     mergekey_norm,
                                                     &se->total_merged);
+
+    const char *use_term_factor_str = session_setting_oneval(sdb, PZ_TERMLIST_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);
+    if (use_term_factor) {
+        int maxrecs = client_get_maxrecs(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 (%d / %d)", term_factor, MAX(hits, maxrecs), MAX(1, maxrecs));
+    }
+
     if (!cluster)
         return -1;
     if (global_parameters.dump_records)
-        yaz_log(YLOG_LOG, "Cluster id %s from %s (#%d)", cluster->recid,
-                sdb->database->url, record_no);
+        session_log(se, YLOG_LOG, "Cluster id %s from %s (#%d)", cluster->recid,
+                    sdb->database->url, record_no);
     relevance_newrec(se->relevance, cluster);
     
     // now parsing XML record and adding data to cluster or record metadata
@@ -1272,7 +1378,7 @@ static int ingest_to_cluster(struct client *cl,
             {
                 if (se->number_of_warnings_unknown_metadata == 0)
                 {
-                    yaz_log(YLOG_WARN, 
+                    session_log(se, YLOG_WARN, 
                             "Ignoring unknown metadata element: %s", type);
                 }
                 se->number_of_warnings_unknown_metadata++;
@@ -1291,8 +1397,8 @@ static int ingest_to_cluster(struct client *cl,
                                           ser_md->type, n->properties);
             if (!rec_md)
             {
-                yaz_log(YLOG_WARN, "bad metadata data '%s' for element '%s'",
-                        value, type);
+                session_log(se, YLOG_WARN, "bad metadata data '%s' "
+                            "for element '%s'", value, type);
                 continue;
             }
             wheretoput = &record->metadata[md_field_id];
@@ -1352,17 +1458,11 @@ static int ingest_to_cluster(struct client *cl,
                         if (!sort_str)
                         {
                             sort_str = rec_md->data.text.disp;
-                            yaz_log(YLOG_WARN, 
+                            session_log(se, YLOG_WARN, 
                                     "Could not make sortkey. Bug #1858");
                         }
                         cluster->sortkeys[sk_field_id]->text.sort = 
                             nmem_strdup(se->nmem, sort_str);
-#if 0
-                        yaz_log(YLOG_LOG, "text disp=%s",
-                                cluster->sortkeys[sk_field_id]->text.disp);
-                        yaz_log(YLOG_LOG, "text sort=%s",
-                                cluster->sortkeys[sk_field_id]->text.sort);
-#endif
                         pp2_relevance_token_destroy(prt);
                     }
                 }
@@ -1402,20 +1502,20 @@ 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];
                     sprintf(year, "%d", rec_md->data.number.max);
-                    add_facet(se, (char *) type, year, 1);
+
+                    add_facet(se, (char *) type, year, term_factor);
                     if (rec_md->data.number.max != rec_md->data.number.min)
                     {
                         sprintf(year, "%d", rec_md->data.number.min);
-                        add_facet(se, (char *) type, year, 1);
+                        add_facet(se, (char *) type, year, term_factor);
                     }
                 }
                 else
-                    add_facet(se, (char *) type, (char *) value, 1);
+                    add_facet(se, (char *) type, (char *) value, term_factor);
             }
 
             // cleaning up
@@ -1426,7 +1526,7 @@ static int ingest_to_cluster(struct client *cl,
         else
         {
             if (se->number_of_warnings_unknown_elements == 0)
-                yaz_log(YLOG_WARN,
+                session_log(se, YLOG_WARN,
                         "Unexpected element in internal record: %s", n->name);
             se->number_of_warnings_unknown_elements++;
         }
@@ -1442,6 +1542,18 @@ static int ingest_to_cluster(struct client *cl,
     return 0;
 }
 
+void session_log(struct session *s, int level, const char *fmt, ...)
+{
+    char buf[1024];
+    va_list ap;
+    va_start(ap, fmt);
+
+    yaz_vsnprintf(buf, sizeof(buf)-30, fmt, ap);
+    yaz_log(level, "Session (%u): %s", s->session_id, buf);
+
+    va_end(ap);
+}
+
 /*
  * Local variables:
  * c-basic-offset: 4