Fix text/cdata nodes collection
[pazpar2-moved-to-github.git] / src / session.c
index ec2f270..fbcd94e 100644 (file)
@@ -57,7 +57,7 @@ Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
 #include <yaz/oid_db.h>
 #include <yaz/snprintf.h>
 
-#define USE_TIMING 0
+#define USE_TIMING 1
 #if USE_TIMING
 #include <yaz/timing.h>
 #endif
@@ -181,7 +181,8 @@ static void session_normalize_facet(struct session *s, const char *type,
     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'",
+        session_log(s, YLOG_FATAL,
+                    "Unknown ICU chain '%s' for facet of type '%s'",
                 icu_chain_id, type);
         wrbuf_destroy(facet_wrbuf);
         wrbuf_destroy(display_wrbuf);
@@ -653,11 +654,12 @@ static void session_sort_unlocked(struct session *se, struct reclist_sortparms *
     int type  = sp->type;
     int clients_research = 0;
 
-    yaz_log(YLOG_LOG, "session_sort field=%s increasing=%d type=%d", field, increasing, type);
+    session_log(se, YLOG_DEBUG, "session_sort field=%s increasing=%d type=%d",
+                field, increasing, type);
     /* see if we already have sorted for this criteria */
     for (sr = se->sorted_results; sr; sr = sr->next)
     {
-        if (!reclist_sortparms_cmp(sr,sp))
+        if (!reclist_sortparms_cmp(sr, sp))
             break;
     }
     if (sr)
@@ -681,7 +683,9 @@ static void session_sort_unlocked(struct session *se, struct reclist_sortparms *
         clients_research += client_parse_sort(cl, sp);
     }
     if (clients_research) {
-        yaz_log(YLOG_DEBUG, "Reset results due to %d clients researching", clients_research);
+        session_log(se, YLOG_DEBUG,
+                    "Reset results due to %d clients researching",
+                    clients_research);
         session_clear_set(se, sp);
     }
     else {
@@ -706,7 +710,8 @@ static void session_sort_unlocked(struct session *se, struct reclist_sortparms *
             client_start_search(cl);
         }
         else {
-            yaz_log(YLOG_DEBUG, "Client %s: No re-start/ingest in show. Wrong client state: %d",
+            session_log(se, YLOG_DEBUG,
+                        "Client %s: No re-start/ingest in show. Wrong client state: %d",
                         client_get_id(cl), client_get_state(cl));
         }
 
@@ -734,7 +739,6 @@ enum pazpar2_error_code session_search(struct session *se,
     int no_failed_query = 0;
     int no_failed_limit = 0;
     struct client_list *l, *l0;
-    int same_sort_order = 0;
 
     session_log(se, YLOG_DEBUG, "Search");
 
@@ -749,10 +753,6 @@ enum pazpar2_error_code session_search(struct session *se,
     session_enter(se, "session_search");
     se->settings_modified = 0;
 
-    if (se->sorted_results) {
-        if (!reclist_sortparms_cmp(se->sorted_results, sp))
-            same_sort_order = 1;
-    }
     session_clear_set(se, sp);
     relevance_destroy(&se->relevance);
 
@@ -1234,76 +1234,139 @@ void show_single_stop(struct session *se, struct record_cluster *rec)
 }
 
 
+int session_fetch_more(struct session *se)
+{
+    struct client_list *l;
+    int ret = 0;
+
+    for (l = se->clients_active; l; l = l->next)
+    {
+        struct client *cl = l->client;
+        if (client_get_state(cl) == Client_Idle)
+        {
+            if (client_fetch_more(cl))
+            {
+                session_log(se, YLOG_LOG, "%s: more to fetch",
+                            client_get_id(cl));
+                ret = 1;
+            }
+            else
+            {
+                session_log(se, YLOG_LOG, "%s: no more to fetch",
+                            client_get_id(cl));
+            }
+        }
+        else
+        {
+            session_log(se, YLOG_LOG, "%s: no fetch due to state=%s",
+                        client_get_id(cl), client_get_state_str(cl));
+        }
+
+    }
+    return ret;
+}
+
 struct record_cluster **show_range_start(struct session *se,
                                          struct reclist_sortparms *sp,
-                                         int start, int *num, int *total, Odr_int *sumhits, Odr_int *approx_hits)
+                                         int start, int *num, int *total,
+                                         Odr_int *sumhits, Odr_int *approx_hits,
+                                         void (*show_records_ready)(void *data),
+                                         struct http_channel *chan)
 {
-    struct record_cluster **recs;
+    struct record_cluster **recs = 0;
     struct reclist_sortparms *spp;
+    struct client_list *l;
     int i;
 #if USE_TIMING
     yaz_timing_t t = yaz_timing_create();
 #endif
     session_enter(se, "show_range_start");
-    recs = nmem_malloc(se->nmem, *num * sizeof(struct record_cluster *));
-    if (!se->relevance)
+    *sumhits = 0;
+    *approx_hits = 0;
+    *total = 0;
+    reclist_limit(se->reclist, se);
+    if (se->relevance)
     {
-        *num = 0;
-        *total = 0;
-        *sumhits = 0;
-        *approx_hits = 0;
-        recs = 0;
-    }
-    else
-    {
-        struct client_list *l;
-
-        reclist_limit(se->reclist, se);
-
         for (spp = sp; spp; spp = spp->next)
             if (spp->type == Metadata_sortkey_relevance)
             {
                 relevance_prepare_read(se->relevance, se->reclist);
                 break;
             }
-        reclist_sort(se->reclist, sp);
-
-        reclist_enter(se->reclist);
-        *total = reclist_get_num_records(se->reclist);
-
-        *sumhits = 0;
-        *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))
-            {
-                *num = 0;
-                recs = 0;
-                break;
-            }
+    }
+    reclist_sort(se->reclist, sp);
+
+    reclist_enter(se->reclist);
+    *total = reclist_get_num_records(se->reclist);
 
-        for (i = 0; i < *num; i++)
+    for (l = se->clients_active; l; l = l->next)
+        client_update_show_stat(l->client, 0);
+
+    for (i = 0; i < start; i++)
+    {
+        struct record_cluster *r = reclist_read_record(se->reclist);
+        if (!r)
         {
-            struct record_cluster *r = reclist_read_record(se->reclist);
-            if (!r)
-            {
-                *num = i;
-                break;
-            }
+            *num = 0;
+            break;
+        }
+        else
+        {
+            struct record *rec = r->records;
+            for (;rec; rec = rec->next)
+                client_update_show_stat(rec->client, 1);
+        }
+    }
+    if (*num > 0)
+        recs =
+            nmem_malloc(se->nmem, *num * sizeof(struct record_cluster *));
+    for (i = 0; i < *num; i++)
+    {
+        struct record_cluster *r = reclist_read_record(se->reclist);
+        if (!r)
+        {
+            *num = i;
+            break;
+        }
+        else
+        {
+            struct record *rec = r->records;
+            for (;rec; rec = rec->next)
+                client_update_show_stat(rec->client, 1);
             recs[i] = r;
         }
-        reclist_leave(se->reclist);
     }
+    reclist_leave(se->reclist);
 #if USE_TIMING
     yaz_timing_stop(t);
-    yaz_log(YLOG_LOG, "show %6.5f %3.2f %3.2f",
+    session_log(se, 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);
 #endif
+
+    if (!session_fetch_more(se))
+        session_log(se, YLOG_LOG, "can not fetch more");
+    else
+    {
+        show_range_stop(se, recs);
+        session_log(se, YLOG_LOG, "fetching more in progress");
+        if (session_set_watch(se, SESSION_WATCH_SHOW,
+                              show_records_ready, chan, chan))
+        {
+            session_log(se, YLOG_WARN, "Ignoring show block");
+            session_enter(se, "show_range_start");
+        }
+        else
+        {
+            session_log(se, YLOG_LOG, "session watch OK");
+            return 0;
+        }
+    }
     return recs;
 }
 
@@ -1352,10 +1415,11 @@ static struct record_metadata *record_metadata_init(
     {
         if (attr->children && attr->children->content)
         {
-            if (strcmp((const char *) attr->name, "type"))
-            {  /* skip the "type" attribute.. Its value is already part of
-                  the element in output (md-%s) and so repeating it here
-                  is redundant */
+            if (strcmp((const char *) attr->name, "type")
+                && strcmp((const char *) attr->name, "empty"))
+            {  /* skip the "type" + "empty" attribute..
+                  The "Type" is already part of the element in output
+                  (md-%s) and so repeating it here is redundant */
                 *attrp = nmem_malloc(nmem, sizeof(**attrp));
                 (*attrp)->name =
                     nmem_strdup(nmem, (const char *) attr->name);
@@ -1807,10 +1871,17 @@ static int ingest_to_cluster(struct client *cl,
 
             type = xmlGetProp(n, (xmlChar *) "type");
             value = xmlNodeListGetString(xdoc, n->children, 1);
-
-            if (!type || !value || !*value)
+            if (!type)
                 continue;
-
+            if (!value || !*value)
+            {
+                xmlChar *empty = xmlGetProp(n, (xmlChar *) "empty");
+                if (!empty)
+                    continue;
+                if (value)
+                    xmlFree(value);
+                value = empty;
+            }
             md_field_id
                 = conf_service_metadata_field_id(service, (const char *) type);
             if (md_field_id < 0)
@@ -1855,7 +1926,7 @@ static int ingest_to_cluster(struct client *cl,
     cluster = reclist_insert(se->reclist, service, record,
                              mergekey_norm, &se->total_merged);
     if (!cluster)
-        return -1;
+        return 0; // complete match with existing record
 
     {
         const char *use_term_factor_str =
@@ -1866,7 +1937,8 @@ 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 (%d / %d)", term_factor, MAX(hits, maxrecs), MAX(1, maxrecs));
+            session_log(se, YLOG_DEBUG, "Using term factor: %d (%d / %d)",
+                        term_factor, MAX(hits, maxrecs), MAX(1, maxrecs));
         }
     }