Return 1 on research. 0 on no research
[pazpar2-moved-to-github.git] / src / client.c
index 56f927d..618dfcf 100644 (file)
@@ -18,7 +18,7 @@ Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
 */
 
 /** \file client.c
-    \brief Z39.50 client 
+    \brief Z39.50 client
 */
 
 #if HAVE_CONFIG_H
@@ -52,6 +52,7 @@ Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
 #include <yaz/snprintf.h>
 #include <yaz/rpn2cql.h>
 #include <yaz/rpn2solr.h>
+#include <yaz/gettimeofday.h>
 
 #define USE_TIMING 0
 #if USE_TIMING
@@ -124,6 +125,9 @@ struct client {
     int ref_count;
     char *id;
     facet_limits_t facet_limits;
+    int same_search;
+    char *sort_strategy;
+    char *sort_criteria;
 };
 
 struct suggestions {
@@ -286,7 +290,7 @@ int client_show_raw_begin(struct client *cl, int position,
 
         if (!cl->connection)
             return -1;
-    
+
 
         rr = xmalloc(sizeof(*rr));
         rr->position = position;
@@ -306,13 +310,13 @@ int client_show_raw_begin(struct client *cl, int position,
 
         assert(nativesyntax);
         rr->nativesyntax = xstrdup(nativesyntax);
-            
+
         rr->next = 0;
-        
+
         for (rrp = &cl->show_raw; *rrp; rrp = &(*rrp)->next)
             ;
         *rrp = rr;
-        
+
         if (cl->state == Client_Failed)
         {
             client_show_raw_error(cl, "client failed");
@@ -533,7 +537,7 @@ void client_search_response(struct client *cl)
     ZOOM_resultset resultset = cl->resultset;
 
     const char *error, *addinfo = 0;
-    
+
     if (ZOOM_connection_error(link, &error, &addinfo))
     {
         cl->hits = 0;
@@ -593,7 +597,7 @@ static void client_record_ingest(struct client *cl)
             NMEM nmem = nmem_create();
             const char *xmlrec;
             char type[80];
-            
+
             const char *s = session_setting_oneval(sdb, PZ_NATIVESYNTAX);
             if (nativesyntax_to_type(s, type, rec))
                 yaz_log(YLOG_WARN, "Failed to determine record type");
@@ -615,8 +619,7 @@ static void client_record_ingest(struct client *cl)
     }
     else
     {
-        yaz_log(YLOG_WARN, "Expected record, but got NULL, offset=%d",
-                offset);
+        yaz_log(YLOG_WARN, "Expected record, but got NULL, offset=%d", offset);
     }
 }
 
@@ -657,7 +660,7 @@ void client_record_response(struct client *cl)
     }
 }
 
-void client_reingest(struct client *cl)
+int client_reingest(struct client *cl)
 {
     int i = cl->startrecs;
     int to = cl->record_offset;
@@ -666,6 +669,7 @@ void client_reingest(struct client *cl)
     cl->record_offset = i;
     for (; i < to; i++)
         client_record_ingest(cl);
+    return 0;
 }
 
 static void client_set_facets_request(struct client *cl, ZOOM_connection link)
@@ -673,7 +677,7 @@ static void client_set_facets_request(struct client *cl, ZOOM_connection link)
     struct session_database *sdb = client_get_database(cl);
 
     WRBUF w = wrbuf_alloc();
-    
+
     struct setting *s;
 
     for (s = sdb->settings[PZ_FACETMAP]; s; s = s->next)
@@ -717,7 +721,7 @@ static const char *get_strategy_plus_sort(struct client *l, const char *field)
     struct setting *s;
 
     const char *strategy_plus_sort = 0;
-    
+
     for (s = sdb->settings[PZ_SORTMAP]; s; s = s->next)
     {
         char *p = strchr(s->name + 3, ':');
@@ -736,11 +740,37 @@ static const char *get_strategy_plus_sort(struct client *l, const char *field)
     return strategy_plus_sort;
 }
 
-void client_start_search(struct client *cl)
+int client_parse_init(struct client *cl, int same_search)
+{
+    cl->same_search = same_search;
+    return 0;
+}
+
+/*
+ * TODO consider how to extend the range
+ * */
+int client_parse_range(struct client *cl, const char *startrecs, const char *maxrecs)
+{
+    if (maxrecs && atoi(maxrecs) != cl->maxrecs)
+    {
+        cl->same_search = 0;
+        cl->maxrecs = atoi(maxrecs);
+    }
+
+    if (startrecs && atoi(startrecs) != cl->startrecs)
+    {
+        cl->same_search = 0;
+        cl->startrecs = atoi(startrecs);
+    }
+
+    return 0;
+}
+
+int client_start_search(struct client *cl)
 {
     struct session_database *sdb = client_get_database(cl);
-    struct connection *co = client_get_connection(cl);
-    ZOOM_connection link = connection_get_link(co);
+    struct connection *co = 0;
+    ZOOM_connection link = 0;
     struct session *se = client_get_session(cl);
     ZOOM_resultset rs;
     const char *opt_piggyback   = session_setting_oneval(sdb, PZ_PIGGYBACK);
@@ -753,15 +783,43 @@ void client_start_search(struct client *cl)
     const char *opt_preferred   = session_setting_oneval(sdb, PZ_PREFERRED);
     const char *extra_args      = session_setting_oneval(sdb, PZ_EXTRA_ARGS);
     const char *opt_present_chunk = session_setting_oneval(sdb, PZ_PRESENT_CHUNK);
-    ZOOM_query q;
+    ZOOM_query query;
     char maxrecs_str[24], startrecs_str[24], present_chunk_str[24];
+    struct timeval tval;
     int present_chunk = 20; // Default chunk size
+    int rc_prep_connection;
+
+
+    yaz_gettimeofday(&tval);
+    tval.tv_sec += 5;
+
     if (opt_present_chunk && strcmp(opt_present_chunk,"")) {
         present_chunk = atoi(opt_present_chunk);
         yaz_log(YLOG_DEBUG, "Present chunk set to %d", present_chunk);
     }
+    rc_prep_connection =
+        client_prep_connection(cl, se->service->z3950_operation_timeout,
+                               se->service->z3950_session_timeout,
+                               se->service->server->iochan_man,
+                               &tval);
+    /* Nothing has changed and we already have a result */
+    if (cl->same_search == 1 && rc_prep_connection == 2)
+    {
+        session_log(se, YLOG_LOG, "client %s REUSE result", client_get_id(cl));
+        return client_reingest(cl);
+    }
+    else if (!rc_prep_connection)
+    {
+        session_log(se, YLOG_LOG, "client %s FAILED to search: No connection.", client_get_id(cl));
+        return -1;
+    }
+    co = client_get_connection(cl);
+    assert(cl);
+    link = connection_get_link(co);
     assert(link);
 
+    session_log(se, YLOG_LOG, "client %s NEW search", client_get_id(cl));
+
     cl->diagnostic = 0;
     cl->filtered = 0;
 
@@ -814,63 +872,36 @@ void client_start_search(struct client *cl)
     /* facets definition is in PQF */
     client_set_facets_request(cl, link);
 
-    q = ZOOM_query_create();
+    query = ZOOM_query_create();
     if (cl->cqlquery)
     {
-        yaz_log(YLOG_LOG, "Search %s CQL: %s", client_get_id(cl),
-                cl->cqlquery);
-        ZOOM_query_cql(q, cl->cqlquery);
+        yaz_log(YLOG_LOG, "Client %s: Search CQL: %s", client_get_id(cl), cl->cqlquery);
+        ZOOM_query_cql(query, cl->cqlquery);
         if (*opt_sort)
-            ZOOM_query_sortby(q, opt_sort);
+            ZOOM_query_sortby(query, opt_sort);
     }
     else
     {
-        yaz_log(YLOG_LOG, "Search %s PQF: %s", client_get_id(cl), cl->pquery);
-        
-        ZOOM_query_prefix(q, cl->pquery);
+        yaz_log(YLOG_LOG, "Client %s: Search PQF: %s", client_get_id(cl), cl->pquery);
+
+        ZOOM_query_prefix(query, cl->pquery);
     }
-    if (se->sorted_results)
-    {   /* first entry is current sorting ! */
-        const char *sort_strategy_and_spec =
-            get_strategy_plus_sort(cl, se->sorted_results->field);
-        int increasing = se->sorted_results->increasing;
-        if (sort_strategy_and_spec && strlen(sort_strategy_and_spec) < 40)
-        {
-            char spec[50], *p;
-            strcpy(spec, sort_strategy_and_spec);
-            p = strchr(spec, ':');
-            if (p)
-            {
-                *p++ = '\0'; /* cut the string in two */
-                while (*p == ' ')
-                    p++;
-                if (increasing)
-                    strcat(p, " <");
-                else
-                    strcat(p, " >");
-                yaz_log(YLOG_LOG, "applying %s %s", spec, p);
-                ZOOM_query_sortby2(q, spec, p);
-            }
-        }
-        else
-        {
-            /* no native sorting.. If this is not the first search, then
-               skip it entirely */
-            if (se->sorted_results->next)
-            {
-                ZOOM_query_destroy(q);
-                return;
-            }
-        }
+    if (cl->sort_strategy && cl->sort_criteria) {
+        yaz_log(YLOG_LOG, "Client %s: Setting ZOOM sort strategy and criteria: %s %s",
+                client_get_id(cl), cl->sort_strategy, cl->sort_criteria);
+        ZOOM_query_sortby2(query, cl->sort_strategy, cl->sort_criteria);
     }
+
+    yaz_log(YLOG_DEBUG,"Client %s: Starting search", client_get_id(cl));
     client_set_state(cl, Client_Working);
     cl->hits = 0;
     cl->record_offset = 0;
-    rs = ZOOM_connection_search(link, q);
-    ZOOM_query_destroy(q);
+    rs = ZOOM_connection_search(link, query);
+    ZOOM_query_destroy(query);
     ZOOM_resultset_destroy(cl->resultset);
     cl->resultset = rs;
     connection_continue(co);
+    return 0;
 }
 
 struct client *client_create(const char *id)
@@ -897,10 +928,12 @@ struct client *client_create(const char *id)
     cl->preferred = 0;
     cl->ref_count = 1;
     cl->facet_limits = 0;
+    cl->sort_strategy = 0;
+    cl->sort_criteria = 0;
     assert(id);
     cl->id = xstrdup(id);
     client_use(1);
-    
+
     return cl;
 }
 
@@ -992,15 +1025,33 @@ static CCL_bibset prepare_cclmap(struct client *cl, CCL_bibset base_bibset)
         res = ccl_qual_mk();
     for (s = sdb->settings[PZ_CCLMAP]; s; s = s->next)
     {
+        const char *addinfo = 0;
         char *p = strchr(s->name + 3, ':');
         if (!p)
         {
-            yaz_log(YLOG_WARN, "Malformed cclmap name: %s", s->name);
+            WRBUF w = wrbuf_alloc();
+            wrbuf_printf(w, "Malformed cclmap. name=%s", s->name);
+            yaz_log(YLOG_WARN, "%s: %s", client_get_id(cl), wrbuf_cstr(w));
+            client_set_diagnostic(cl, ZOOM_ERROR_CCL_CONFIG, wrbuf_cstr(w));
+            client_set_state_nb(cl, Client_Error);
             ccl_qual_rm(&res);
+            wrbuf_destroy(w);
             return 0;
         }
         p++;
-        ccl_qual_fitem(res, s->value, p);
+        if (ccl_qual_fitem2(res, s->value, p, &addinfo))
+        {
+            WRBUF w = wrbuf_alloc();
+
+            wrbuf_printf(w, "Malformed cclmap. name=%s: value=%s (%s)",
+                         s->name, p, addinfo);
+            yaz_log(YLOG_WARN, "%s: %s", client_get_id(cl), wrbuf_cstr(w));
+            client_set_diagnostic(cl, ZOOM_ERROR_CCL_CONFIG, wrbuf_cstr(w));
+            client_set_state_nb(cl, Client_Error);
+            ccl_qual_rm(&res);
+            wrbuf_destroy(w);
+            return 0;
+        }
     }
     return res;
 }
@@ -1020,7 +1071,7 @@ static char *make_cqlquery(struct client *cl, Z_RPNQuery *zquery)
     else
     {
         r = xstrdup(wrbuf_cstr(wrb));
-    }     
+    }
     wrbuf_destroy(wrb);
     cql_transform_close(cqlt);
     return r;
@@ -1034,7 +1085,7 @@ static char *make_solrquery(struct client *cl, Z_RPNQuery *zquery)
     char *r = 0;
     WRBUF wrb = wrbuf_alloc();
     int status;
-    
+
     if ((status = solr_transform_rpn2solr_wrbuf(sqlt, wrb, zquery)))
     {
         yaz_log(YLOG_WARN, "Failed to generate SOLR query, code=%d", status);
@@ -1058,7 +1109,7 @@ const char *client_get_facet_limit_local(struct client *cl,
     for (; (name = facet_limits_get(cl->facet_limits, *l, &value)); (*l)++)
     {
         struct setting *s = 0;
-        
+
         for (s = sdb->settings[PZ_LIMITMAP]; s; s = s->next)
         {
             const char *p = strchr(s->name + 3, ':');
@@ -1068,7 +1119,7 @@ const char *client_get_facet_limit_local(struct client *cl,
                 const char *cp = s->value + 6;
                 while (*cp == ' ')
                     cp++;
-                    
+
                 nmem_strsplit_escape2(nmem, "|", value, values,
                                       num, 1, '\\', 1);
                 (*l)++;
@@ -1081,17 +1132,12 @@ const char *client_get_facet_limit_local(struct client *cl,
 
 static int apply_limit(struct session_database *sdb,
                        facet_limits_t facet_limits,
-                       WRBUF w_pqf, WRBUF w_ccl,
-                       CCL_bibset ccl_map)
+                       WRBUF w_pqf, CCL_bibset ccl_map)
 {
     int ret = 0;
     int i = 0;
     const char *name;
     const char *value;
-    const char **and_op_names = ccl_qual_search_special(ccl_map, "and");
-    const char *and_op = and_op_names ? and_op_names[0] : "and";
-    const char **or_op_names = ccl_qual_search_special(ccl_map, "or");
-    const char *or_op = or_op_names ? or_op_names[0] : "or";
 
     NMEM nmem_tmp = nmem_create();
     for (i = 0; (name = facet_limits_get(facet_limits, i, &value)); i++)
@@ -1126,20 +1172,34 @@ static int apply_limit(struct session_database *sdb,
                 else if (!strncmp(s->value, "ccl:", 4))
                 {
                     const char *ccl = s->value + 4;
-                    
-                    wrbuf_printf(w_ccl, " %s (", and_op);
-
+                    WRBUF ccl_w = wrbuf_alloc();
                     for (i = 0; i < num; i++)
                     {
-                        if (i)
-                            wrbuf_printf(w_ccl, " %s ", or_op);
-                        wrbuf_puts(w_ccl, ccl);
-                        wrbuf_puts(w_ccl, "=\"");
-                        wrbuf_puts(w_ccl, values[i]);
-                        wrbuf_puts(w_ccl, "\"");
+                        int cerror, cpos;
+                        struct ccl_rpn_node *cn;
+
+                        wrbuf_rewind(ccl_w);
+                        wrbuf_puts(ccl_w, ccl);
+                        wrbuf_puts(ccl_w, "=\"");
+                        wrbuf_puts(ccl_w, values[i]);
+                        wrbuf_puts(ccl_w, "\"");
+
+                        cn = ccl_find_str(ccl_map, wrbuf_cstr(ccl_w),
+                                          &cerror, &cpos);
+                        if (cn)
+                        {
+                            if (i == 0)
+                                wrbuf_printf(w_pqf, "@and ");
+
+                            /* or multiple values.. could be bad if last CCL
+                               parse fails, but this is unlikely to happen */
+                            if (i < num - 1)
+                                wrbuf_printf(w_pqf, "@or ");
+                            ccl_pquery(w_pqf, cn);
+                            ccl_rpn_delete(cn);
+                        }
                     }
-                    wrbuf_puts(w_ccl, ")");
-
+                    wrbuf_destroy(ccl_w);
                 }
                 else if (!strncmp(s->value, "local:", 6)) {
                     /* no operation */
@@ -1162,15 +1222,15 @@ static int apply_limit(struct session_database *sdb,
     nmem_destroy(nmem_tmp);
     return ret;
 }
-                        
+
 // Parse the query given the settings specific to this client
-// return 0 if query is OK but different from before
-// return 1 if query is OK but same as before
+// client variable same_search is set as below as well as returned:
+// 0 if query is OK but different from before
+// 1 if query is OK but same as before
 // return -1 on query error
 // return -2 on limit error
 int client_parse_query(struct client *cl, const char *query,
                        facet_limits_t facet_limits,
-                       const char *startrecs, const char *maxrecs,
                        CCL_bibset bibset)
 {
     struct session *se = client_get_session(cl);
@@ -1188,19 +1248,7 @@ int client_parse_query(struct client *cl, const char *query,
     Z_RPNQuery *zquery;
 
     if (!ccl_map)
-        return -1;
-
-    if (maxrecs && atoi(maxrecs) != cl->maxrecs)
-    {
-        ret_value = 0;
-        cl->maxrecs = atoi(maxrecs);
-    }
-
-    if (startrecs && atoi(startrecs) != cl->startrecs)
-    {
-        ret_value = 0;
-        cl->startrecs = atoi(startrecs);
-    }
+        return -3;
 
     w_ccl = wrbuf_alloc();
     wrbuf_puts(w_ccl, query);
@@ -1212,7 +1260,7 @@ int client_parse_query(struct client *cl, const char *query,
         wrbuf_puts(w_pqf, " ");
     }
 
-    if (apply_limit(sdb, facet_limits, w_pqf, w_ccl, ccl_map))
+    if (apply_limit(sdb, facet_limits, w_pqf, ccl_map))
     {
         ccl_qual_rm(&ccl_map);
         return -2;
@@ -1221,15 +1269,15 @@ int client_parse_query(struct client *cl, const char *query,
     facet_limits_destroy(cl->facet_limits);
     cl->facet_limits = facet_limits_dup(facet_limits);
 
-    yaz_log(YLOG_LOG, "CCL query: %s", wrbuf_cstr(w_ccl));
+    yaz_log(YLOG_LOG, "Client %s: CCL query: %s limit: %s", client_get_id(cl), wrbuf_cstr(w_ccl), wrbuf_cstr(w_pqf));
     cn = ccl_find_str(ccl_map, wrbuf_cstr(w_ccl), &cerror, &cpos);
     ccl_qual_rm(&ccl_map);
     if (!cn)
     {
         client_set_state(cl, Client_Error);
-        session_log(se, YLOG_WARN, "Failed to parse CCL query '%s' for %s",
-                    wrbuf_cstr(w_ccl),
-                    client_get_id(cl));
+        session_log(se, YLOG_WARN, "Client %s: Failed to parse CCL query '%s'",
+                    client_get_id(cl),
+                    wrbuf_cstr(w_ccl));
         wrbuf_destroy(w_ccl);
         wrbuf_destroy(w_pqf);
         return -1;
@@ -1257,31 +1305,38 @@ int client_parse_query(struct client *cl, const char *query,
         }
     }
 
+    /* Compares query and limit with old one. If different we need to research */
     if (!cl->pquery || strcmp(cl->pquery, wrbuf_cstr(w_pqf)))
     {
+        if (cl->pquery)
+            session_log(se, YLOG_LOG, "Client %s: Re-search due query/limit change: %s to %s", 
+                        client_get_id(cl), cl->pquery, wrbuf_cstr(w_pqf));
         xfree(cl->pquery);
         cl->pquery = xstrdup(wrbuf_cstr(w_pqf));
+        // return value is no longer used.
         ret_value = 0;
+        // Need to (re)search
+        cl->same_search= 0;
     }
     wrbuf_destroy(w_pqf);
-    
+
     xfree(cl->cqlquery);
     cl->cqlquery = 0;
 
-    odr_out = odr_createmem(ODR_ENCODE);    
+    odr_out = odr_createmem(ODR_ENCODE);
     zquery = p_query_rpn(odr_out, cl->pquery);
     if (!zquery)
     {
 
-        session_log(se, YLOG_WARN, "Invalid PQF query for %s: %s",
+        session_log(se, YLOG_WARN, "Invalid PQF query for Client %s: %s",
                     client_get_id(cl), cl->pquery);
         ret_value = -1;
     }
     else
     {
-        session_log(se, YLOG_LOG, "PQF for %s: %s",
+        session_log(se, YLOG_LOG, "PQF for Client %s: %s",
                     client_get_id(cl), cl->pquery);
-        
+
         /* Support for PQF on SRU targets. */
         if (strcmp(query_syntax, "pqf") != 0 && *sru)
         {
@@ -1291,6 +1346,9 @@ int client_parse_query(struct client *cl, const char *query,
                 cl->cqlquery = make_cqlquery(cl, zquery);
             if (!cl->cqlquery)
                 ret_value = -1;
+            else
+                session_log(se, YLOG_LOG, "Client %s native query: %s (%s)",
+                            client_get_id(cl), cl->cqlquery, sru);
         }
     }
     odr_destroy(odr_out);
@@ -1299,12 +1357,66 @@ int client_parse_query(struct client *cl, const char *query,
     if (!se->relevance)
     {
         // Initialize relevance structure with query terms
-        se->relevance = relevance_create_ccl(se->service->charsets, cn);
+        se->relevance = relevance_create_ccl(se->service->charsets, cn,
+                                             se->service->rank_cluster,
+                                             se->service->rank_follow,
+                                             se->service->rank_lead,
+                                             se->service->rank_length);
     }
     ccl_rpn_delete(cn);
     return ret_value;
 }
 
+int client_parse_sort(struct client *cl, struct reclist_sortparms *sp)
+{
+    struct session *se = client_get_session(cl);
+    if (sp)
+    {   /* first entry is current sorting ! */
+        const char *sort_strategy_and_spec =
+            get_strategy_plus_sort(cl, se->sorted_results->name);
+
+        int increasing = se->sorted_results->increasing;
+        // int type = se->sorted_results->type;
+        if (sort_strategy_and_spec && strlen(sort_strategy_and_spec) < 40)
+        {
+            char strategy[50], *p;
+            strcpy(strategy, sort_strategy_and_spec);
+            p = strchr(strategy, ':');
+            if (p)
+            {
+                // Split the string in two
+                *p++ = 0;
+                while (*p == ' ')
+                    p++;
+                if (increasing)
+                    strcat(p, " <");
+                else
+                    strcat(p, " >");
+                yaz_log(YLOG_LOG, "Client %s: applying sorting %s %s", client_get_id(cl), strategy, p);
+                if (!cl->sort_strategy || strcmp(cl->sort_strategy, strategy))
+                    cl->same_search = 0;
+                if (!cl->sort_criteria || strcmp(cl->sort_criteria, p))
+                    cl->same_search = 0;
+                if (cl->same_search == 0) {
+                    cl->sort_strategy  = nmem_strdup(se->nmem, strategy);
+                    cl->sort_criteria = nmem_strdup(se->nmem, p);
+                }
+            }
+            else {
+                yaz_log(YLOG_LOG, "Client %s: Invalid sort strategy and spec found %s", client_get_id(cl), sort_strategy_and_spec);
+                cl->sort_strategy  = 0;
+                cl->sort_criteria = 0;
+            }
+        } else {
+            yaz_log(YLOG_LOG, "Client %s: No sort strategy and spec found.", client_get_id(cl));
+            cl->sort_strategy  = 0;
+            cl->sort_criteria = 0;
+        }
+
+    }
+    return !cl->same_search;
+}
+
 void client_set_session(struct client *cl, struct session *se)
 {
     cl->session = se;
@@ -1429,11 +1541,11 @@ struct suggestions* client_suggestions_create(const char* suggestions_string)
     int i;
     NMEM nmem;
     struct suggestions *suggestions;
-    if (suggestions_string == 0)
+    if (suggestions_string == 0 || suggestions_string[0] == 0 )
         return 0;
     nmem = nmem_create();
     suggestions = nmem_malloc(nmem, sizeof(*suggestions));
-    yaz_log(YLOG_DEBUG, "client target suggestions: %s", suggestions_string);
+    yaz_log(YLOG_DEBUG, "client target suggestions: %s.", suggestions_string);
 
     suggestions->nmem = nmem;
     suggestions->num = 0;
@@ -1465,6 +1577,11 @@ static void client_suggestions_destroy(struct client *cl)
     nmem_destroy(nmem);
 }
 
+int client_test_sort_order(struct client *cl, struct reclist_sortparms *sp)
+{
+    //TODO implement correctly.
+    return 1;
+}
 /*
  * Local variables:
  * c-basic-offset: 4