X-Git-Url: http://lists.indexdata.com/cgi-bin?a=blobdiff_plain;f=src%2Fsession.c;h=5e39968d6a496fd6153e3ddcf9ae1a3a045168da;hb=8e14ee4be4caa036057d7d7a13cce8ecd06f2fc2;hp=f3e71fe92e13ca58aafac0142853fe975950785a;hpb=0a70ba6da785b09a4bb1a2a0db5576c157a5dae7;p=pazpar2-moved-to-github.git

diff --git a/src/session.c b/src/session.c
index f3e71fe..5e39968 100644
--- a/src/session.c
+++ b/src/session.c
@@ -1,5 +1,5 @@
 /* This file is part of Pazpar2.
-   Copyright (C) 2006-2012 Index Data
+   Copyright (C) 2006-2013 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
@@ -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);
@@ -623,7 +624,6 @@ int session_is_preferred_clients_ready(struct session *s)
 static void session_clear_set(struct session *se, struct reclist_sortparms *sp)
 {
     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));
@@ -653,11 +653,12 @@ 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 +682,9 @@ 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 +709,8 @@ 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));
         }
 
@@ -734,8 +738,6 @@ enum pazpar2_error_code session_search(struct session *se,
     int no_failed_query = 0;
     int no_failed_limit = 0;
     struct client_list *l, *l0;
-    facet_limits_t facet_limits;
-    int same_sort_order = 0;
 
     session_log(se, YLOG_DEBUG, "Search");
 
@@ -750,10 +752,6 @@ 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);
 
@@ -764,8 +762,9 @@ enum pazpar2_error_code session_search(struct session *se,
         return PAZPAR2_NO_TARGETS;
     }
 
-    facet_limits = facet_limits_create(limit);
-    if (!facet_limits)
+    facet_limits_destroy(se->facet_limits);
+    se->facet_limits = facet_limits_create(limit);
+    if (!se->facet_limits)
     {
         *addinfo = "limit";
         session_leave(se, "session_search");
@@ -784,7 +783,7 @@ enum pazpar2_error_code session_search(struct session *se,
         if (prepare_map(se, client_get_database(cl)) < 0)
             continue;
 
-        parse_ret = client_parse_query(cl, query, facet_limits, se->service->ccl_bibset);
+        parse_ret = client_parse_query(cl, query, se->facet_limits);
         if (parse_ret == -1)
             no_failed_query++;
         else if (parse_ret == -2)
@@ -799,7 +798,6 @@ enum pazpar2_error_code session_search(struct session *se,
             no_working++;
         }
     }
-    facet_limits_destroy(facet_limits);
     session_reset_active_clients(se, l0);
 
     if (no_working == 0)
@@ -931,6 +929,7 @@ void session_destroy(struct session *se)
         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));
+    facet_limits_destroy(se->facet_limits);
     nmem_destroy(se->nmem);
     service_destroy(se->service);
     yaz_mutex_destroy(&se->session_mutex);
@@ -965,7 +964,7 @@ struct session *new_session(NMEM nmem, struct conf_service *service,
     session->number_of_warnings_unknown_elements = 0;
     session->number_of_warnings_unknown_metadata = 0;
     session->num_termlists = 0;
-    session->reclist = 0;
+    session->reclist = reclist_create(nmem);
     session->clients_active = 0;
     session->clients_cached = 0;
     session->settings_modified = 0;
@@ -973,6 +972,7 @@ struct session *new_session(NMEM nmem, struct conf_service *service,
     session->nmem = nmem_create();
     session->databases = 0;
     session->sorted_results = 0;
+    session->facet_limits = 0;
 
     for (i = 0; i <= SESSION_WATCH_MAX; i++)
     {
@@ -1206,20 +1206,19 @@ struct record_cluster *show_single_start(struct session *se, const char *id,
     session_enter(se, "show_single_start");
     *prev_r = 0;
     *next_r = 0;
-    if (se->reclist)
+    reclist_limit(se->reclist, se);
+
+    reclist_enter(se->reclist);
+    while ((r = reclist_read_record(se->reclist)))
     {
-        reclist_enter(se->reclist);
-        while ((r = reclist_read_record(se->reclist)))
+        if (!strcmp(r->recid, id))
         {
-            if (!strcmp(r->recid, id))
-            {
-                *next_r = reclist_read_record(se->reclist);
-                break;
-            }
-            *prev_r = r;
+            *next_r = reclist_read_record(se->reclist);
+            break;
         }
-        reclist_leave(se->reclist);
+        *prev_r = r;
     }
+    reclist_leave(se->reclist);
     if (!r)
         session_leave(se, "show_single_start");
     return r;
@@ -1230,74 +1229,138 @@ void show_single_stop(struct session *se, struct record_cluster *rec)
     session_leave(se, "show_single_stop");
 }
 
+
+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;
-
         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 (l = se->clients_active; l; l = l->next)
+        client_update_show_stat(l->client, 0);
 
-        for (i = 0; i < *num; i++)
+    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);
+        }
+    }
+    recs = nmem_malloc(se->nmem, (*num > 0 ? *num : 1) * sizeof(*recs));
+    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;
 }
 
@@ -1346,10 +1409,11 @@ 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);
@@ -1619,14 +1683,14 @@ int ingest_record(struct client *cl, const char *rec,
     return ret;
 }
 
-static int match_metadata_local(struct record *record,
-                                struct conf_service *service,
-                                int md_field_id,
+//    struct conf_metadata *ser_md = &service->metadata[md_field_id];
+//    struct record_metadata *rec_md = record->metadata[md_field_id];
+static int match_metadata_local(struct conf_metadata *ser_md,
+                                struct record_metadata *rec_md0,
                                 char **values, int num_v)
 {
     int i;
-    struct conf_metadata *ser_md = &service->metadata[md_field_id];
-    struct record_metadata *rec_md = record->metadata[md_field_id];
+    struct record_metadata *rec_md = rec_md0;
     for (i = 0; i < num_v; )
     {
         if (rec_md)
@@ -1652,13 +1716,59 @@ static int match_metadata_local(struct record *record,
         }
         else
         {
-            rec_md = record->metadata[md_field_id];
+            rec_md = rec_md0;
             i++;
         }
     }
     return i < num_v ? 1 : 0;
 }
 
+int session_check_cluster_limit(struct session *se, struct record_cluster *rec)
+{
+    int i;
+    struct conf_service *service = se->service;
+    int ret = 1;
+    const char *name;
+    const char *value;
+    NMEM nmem_tmp = nmem_create();
+
+    for (i = 0; (name = facet_limits_get(se->facet_limits, i, &value)); i++)
+    {
+        int j;
+        for (j = 0; j < service->num_metadata; j++)
+        {
+            struct conf_metadata *md = service->metadata + j;
+            if (!strcmp(md->name, name) && md->limitcluster)
+            {
+                char **values = 0;
+                int num = 0;
+                int md_field_id =
+                    conf_service_metadata_field_id(service,
+                                                   md->limitcluster);
+
+                if (md_field_id < 0)
+                {
+                    ret = 0;
+                    break;
+                }
+
+                nmem_strsplit_escape2(nmem_tmp, "|", value, &values,
+                                      &num, 1, '\\', 1);
+
+                if (!match_metadata_local(&service->metadata[md_field_id],
+                                          rec->metadata[md_field_id],
+                                          values, num))
+                {
+                    ret = 0;
+                    break;
+                }
+            }
+        }
+    }
+    nmem_destroy(nmem_tmp);
+    return ret;
+}
+
 // Skip record on non-zero
 static int check_limit_local(struct client *cl,
                              struct record *record,
@@ -1681,17 +1791,35 @@ static int check_limit_local(struct client *cl,
         if (!name)
             break;
 
-        md_field_id = conf_service_metadata_field_id(service, name);
-        if (md_field_id < 0)
+        if (!strcmp(name, "*"))
         {
-            skip_record = 1;
-            break;
+            for (md_field_id = 0; md_field_id < service->num_metadata;
+                 md_field_id++)
+            {
+                if (match_metadata_local(
+                        &service->metadata[md_field_id],
+                        record->metadata[md_field_id],
+                        values, num_v))
+                    break;
+            }
+            if (md_field_id == service->num_metadata)
+                skip_record = 1;
         }
-        if (!match_metadata_local(record, service, md_field_id,
-                                  values, num_v))
+        else
         {
-            skip_record = 1;
-            break;
+            md_field_id = conf_service_metadata_field_id(service, name);
+            if (md_field_id < 0)
+            {
+                skip_record = 1;
+                break;
+            }
+            if (!match_metadata_local(
+                    &service->metadata[md_field_id],
+                    record->metadata[md_field_id],
+                    values, num_v))
+            {
+                skip_record = 1;
+            }
         }
     }
     nmem_destroy(nmem_tmp);
@@ -1711,6 +1839,7 @@ static int ingest_to_cluster(struct client *cl,
     struct conf_service *service = se->service;
     int term_factor = 1;
     struct record_cluster *cluster;
+    struct record_metadata **metadata0;
     struct session_database *sdb = client_get_database(cl);
     struct record *record = record_create(se->nmem,
                                           service->num_metadata,
@@ -1736,10 +1865,17 @@ 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)
@@ -1784,7 +1920,7 @@ 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 =
@@ -1795,7 +1931,8 @@ 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));
         }
     }
 
@@ -1806,6 +1943,11 @@ static int ingest_to_cluster(struct client *cl,
 
     relevance_newrec(se->relevance, cluster);
 
+    // original metadata, to check if first existence of a field
+    metadata0 = xmalloc(sizeof(*metadata0) * service->num_metadata);
+    memcpy(metadata0, cluster->metadata,
+           sizeof(*metadata0) * service->num_metadata);
+
     // now parsing XML record and adding data to cluster or record metadata
     for (n = root->children; n; n = n->next)
     {
@@ -1851,6 +1993,9 @@ static int ingest_to_cluster(struct client *cl,
             // merged metadata
             rec_md = record_metadata_init(se->nmem, (const char *) value,
                                           ser_md->type, 0);
+
+            // see if the field was not in cluster already (from beginning)
+
             if (!rec_md)
                 continue;
 
@@ -1859,9 +2004,16 @@ static int ingest_to_cluster(struct client *cl,
 
             wheretoput = &cluster->metadata[md_field_id];
 
-            // and polulate with data:
-            // assign cluster or record based on merge action
-            if (ser_md->merge == Metadata_merge_unique)
+            if (ser_md->merge == Metadata_merge_first)
+            {
+                if (!metadata0[md_field_id])
+                {
+                    while (*wheretoput)
+                        wheretoput = &(*wheretoput)->next;
+                    *wheretoput = rec_md;
+                }
+            }
+            else if (ser_md->merge == Metadata_merge_unique)
             {
                 while (*wheretoput)
                 {
@@ -1987,6 +2139,7 @@ static int ingest_to_cluster(struct client *cl,
     if (value)
         xmlFree(value);
 
+    xfree(metadata0);
     relevance_donerecord(se->relevance, cluster);
     se->total_records++;
 
@@ -2000,7 +2153,7 @@ 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, "Session %u: %s", s->session_id, buf);
+    yaz_log(level, "Session %u: %s", s ? s->session_id : 0, buf);
 
     va_end(ap);
 }