X-Git-Url: http://git.indexdata.com/?a=blobdiff_plain;f=src%2Fsession.c;h=3567ab62086ab874d2adc38675b984bc3ef24575;hb=856d06997c1494ed13ddad8eed3531d0852290c9;hp=f1f742a57ab5026c7019a034d4882f2efc60ee65;hpb=3606fd6346ccfe3508aa7cd366b849e389f1b861;p=pazpar2-moved-to-github.git

diff --git a/src/session.c b/src/session.c
index f1f742a..3567ab6 100644
--- a/src/session.c
+++ b/src/session.c
@@ -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 {
@@ -97,6 +98,7 @@ struct client_list {
 /* 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)
 {
@@ -105,16 +107,30 @@ static int session_use(int delta)
         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);
+    yaz_log(YLOG_DEBUG, "%s sessions=%d", delta == 0 ? "" : (delta > 0 ? "INC" : "DEC"), no_sessions);
     return sessions;
 }
 
-int sessions_count(void) {
+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();
@@ -143,53 +159,50 @@ static void session_leave(struct session *s)
     yaz_mutex_leave(s->session_mutex);
 }
 
-// Recursively traverse query structure to extract terms.
-void pull_terms(NMEM nmem, struct ccl_rpn_node *n, char **termlist, int *num)
-{
-    char **words;
-    int numwords;
-    int i;
-
-    switch (n->kind)
-    {
-    case CCL_RPN_AND:
-    case CCL_RPN_OR:
-    case CCL_RPN_NOT:
-    case CCL_RPN_PROX:
-        pull_terms(nmem, n->u.p[0], termlist, num);
-        pull_terms(nmem, n->u.p[1], termlist, num);
-        break;
-    case CCL_RPN_TERM:
-        nmem_strsplit(nmem, " ", n->u.t.term, &words, &numwords);
-        for (i = 0; i < numwords; i++)
-            termlist[(*num)++] = words[i];
-        break;
-    default: // NOOP
-        break;
-    }
-}
-
-
 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;
@@ -202,6 +215,7 @@ 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;
             }
             
@@ -211,13 +225,14 @@ void add_facet(struct session *s, const char *type, const char *value, int count
             s->num_termlists = i + 1;
         }
         
-#if
+#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);
+        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,
@@ -473,7 +488,7 @@ 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)
@@ -506,11 +521,18 @@ void session_alert_watch(struct session *s, int what)
 }
 
 //callback for grep_databases
-static void select_targets_callback(void *context, struct session_database *db)
+static void select_targets_callback(struct session *se,
+                                    struct session_database *db)
 {
-    struct session *se = (struct session*) context;
     struct client *cl = client_create();
     struct client_list *l;
+    const char *url = session_setting_oneval(db, PZ_URL);
+    
+    if (!url || !*url)
+        url = db->database->url;
+
+    resolve_database(se->service, db->database, url);
+
     client_set_database(cl, db);
 
     client_set_session(cl, se);
@@ -575,12 +597,11 @@ int session_is_preferred_clients_ready(struct session *s)
     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;
@@ -588,6 +609,7 @@ 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");
 
@@ -614,6 +636,13 @@ 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;
@@ -624,7 +653,7 @@ 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
         {
@@ -636,6 +665,7 @@ enum pazpar2_error_code search(struct session *se,
                 client_start_search(cl);
         }
     }
+    facet_limits_destroy(facet_limits);
     session_leave(se);
     if (no_working == 0)
     {
@@ -694,8 +724,6 @@ static struct session_database *load_session_database(struct session *se,
 {
     struct database *db = new_database(id, se->session_nmem);
 
-    resolve_database(se->service, db);
-
     session_init_databases_fun((void*) se, db);
 
     // New sdb is head of se->databases list
@@ -745,8 +773,7 @@ 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, "Destroying");
     session_use(-1);
@@ -760,7 +787,12 @@ 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) {
@@ -797,7 +829,6 @@ 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++)
     {
@@ -811,13 +842,13 @@ struct session *new_session(NMEM nmem, struct conf_service *service,
     return session;
 }
 
-struct hitsbytarget *hitsbytarget(struct session *se, int *count, NMEM nmem)
+static struct hitsbytarget *hitsbytarget_nb(struct session *se,
+                                            int *count, NMEM nmem)
 {
     struct hitsbytarget *res = 0;
     struct client_list *l;
     size_t sz = 0;
 
-    session_enter(se);
     for (l = se->clients; l; l = l->next)
         sz++;
 
@@ -838,14 +869,24 @@ 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 hitsbytarget *get_hitsbytarget(struct session *se, int *count, NMEM nmem)
+{
+    struct hitsbytarget *p;
+    session_enter(se);
+    p = hitsbytarget_nb(se, count, 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;
@@ -861,6 +902,118 @@ struct termlist_score **termlist(struct session *se, const char *name, int *num)
     return tl;
 }
 
+// Compares two hitsbytarget nodes by hitcount
+static int cmp_ht(const void *p1, const void *p2)
+{
+    const struct hitsbytarget *h1 = p1;
+    const struct hitsbytarget *h2 = p2;
+    return h2->hits - h1->hits;
+}
+
+static int targets_termlist_nb(WRBUF wrbuf, struct session *se, int num,
+                               NMEM nmem)
+{
+    struct hitsbytarget *ht;
+    int count, i;
+
+    ht = hitsbytarget_nb(se, &count, nmem);
+    qsort(ht, count, sizeof(struct hitsbytarget), cmp_ht);
+    for (i = 0; i < count && i < num && ht[i].hits > 0; i++)
+    {
+
+        // do only print terms which have display names
+    
+        wrbuf_puts(wrbuf, "<term>\n");
+
+        wrbuf_puts(wrbuf, "<id>");
+        wrbuf_xmlputs(wrbuf, ht[i].id);
+        wrbuf_puts(wrbuf, "</id>\n");
+        
+        wrbuf_puts(wrbuf, "<name>");
+        if (!ht[i].name || !ht[i].name[0])
+            wrbuf_xmlputs(wrbuf, "NO TARGET NAME");
+        else
+            wrbuf_xmlputs(wrbuf, ht[i].name);
+        wrbuf_puts(wrbuf, "</name>\n");
+        
+        wrbuf_printf(wrbuf, "<frequency>" ODR_INT_PRINTF "</frequency>\n",
+                     ht[i].hits);
+        
+        wrbuf_puts(wrbuf, "<state>");
+        wrbuf_xmlputs(wrbuf, ht[i].state);
+        wrbuf_puts(wrbuf, "</state>\n");
+        
+        wrbuf_printf(wrbuf, "<diagnostic>%d</diagnostic>\n", 
+                     ht[i].diagnostic);
+        wrbuf_puts(wrbuf, "</term>\n");
+    }
+    return count;
+}
+
+void perform_termlist(struct http_channel *c, struct session *se,
+                      const char *name, int num)
+{
+    int i, j;
+    NMEM nmem_tmp = nmem_create();
+    char **names;
+    int num_names = 0;
+
+    if (name)
+        nmem_strsplit(nmem_tmp, ",", name, &names, &num_names);
+
+    session_enter(se);
+
+    for (j = 0; j < num_names; j++)
+    {
+        const char *tname;
+        for (i = 0; i < se->num_termlists; i++)
+        {
+            tname = se->termlists[i].name;
+            if (num_names > 0 && !strcmp(names[j], tname))
+            {
+                struct termlist_score **p = 0;
+                int len;
+                p = termlist_highscore(se->termlists[i].termlist, &len);
+                if (p)
+                {
+                    int i;
+                    wrbuf_puts(c->wrbuf, "<list name=\"");
+                    wrbuf_xmlputs(c->wrbuf, tname);
+                    wrbuf_puts(c->wrbuf, "\">\n");
+                    for (i = 0; i < len && i < num; i++)
+                    {
+                        // prevent sending empty term elements
+                        if (!p[i]->display_term || !p[i]->display_term[0])
+                            continue;
+                        
+                        wrbuf_puts(c->wrbuf, "<term>");
+                        wrbuf_puts(c->wrbuf, "<name>");
+                        wrbuf_xmlputs(c->wrbuf, p[i]->display_term);
+                        wrbuf_puts(c->wrbuf, "</name>");
+                        
+                        wrbuf_printf(c->wrbuf, 
+                                     "<frequency>%d</frequency>", 
+                                     p[i]->frequency);
+                        wrbuf_puts(c->wrbuf, "</term>\n");
+                    }
+                    wrbuf_puts(c->wrbuf, "</list>\n");
+                }
+            }
+        }
+        tname = "xtargets";
+        if (num_names > 0 && !strcmp(names[j], tname))
+        {
+            wrbuf_puts(c->wrbuf, "<list name=\"");
+            wrbuf_xmlputs(c->wrbuf, tname);
+            wrbuf_puts(c->wrbuf, "\">\n");
+            targets_termlist_nb(c->wrbuf, se, num, c->nmem);
+            wrbuf_puts(c->wrbuf, "</list>\n");
+        }
+    }
+    session_leave(se);
+    nmem_destroy(nmem_tmp);
+}
+
 #ifdef MISSING_HEADERS
 void report_nmem_stats(void)
 {
@@ -1077,15 +1230,15 @@ 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)
                         {
@@ -1094,7 +1247,7 @@ 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++;
                 }
             }
@@ -1116,11 +1269,11 @@ 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)
             {
@@ -1129,7 +1282,7 @@ 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
@@ -1241,6 +1394,7 @@ 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)
@@ -1260,10 +1414,10 @@ 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);
@@ -1327,7 +1481,7 @@ 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)
@@ -1394,16 +1548,15 @@ 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)
             {
@@ -1423,12 +1576,13 @@ 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);
                          
@@ -1442,13 +1596,14 @@ 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)
@@ -1481,7 +1636,6 @@ 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];