X-Git-Url: http://git.indexdata.com/?a=blobdiff_plain;f=src%2Fclient.c;h=56f927daa054ea7d9c93b04713e2a60aebac1a1d;hb=6b86af5db49a421480705a262ce56be54d12d1ab;hp=d8c239dc93e05823d1d90dacbb8719bc0471d9d7;hpb=7e12d1c3bb9e12d6786fc2e16178ee918b440e5a;p=pazpar2-moved-to-github.git

diff --git a/src/client.c b/src/client.c
index d8c239d..56f927d 100644
--- a/src/client.c
+++ b/src/client.c
@@ -1,5 +1,5 @@
 /* This file is part of Pazpar2.
-   Copyright (C) 2006-2011 Index Data
+   Copyright (C) 2006-2012 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
@@ -111,6 +111,7 @@ struct client {
     char *addinfo; // diagnostic info for most resent error
     Odr_int hits;
     int record_offset;
+    int filtered; // When using local:, this will count the number of filtered records.
     int maxrecs;
     int startrecs;
     int diagnostic;
@@ -411,16 +412,12 @@ static int nativesyntax_to_type(const char *s, char *type,
             const char *cp = strchr(s, ';');
             yaz_snprintf(type, 80, "xml; charset=%s", cp ? cp+1 : "marc-8s");
         }
-        else if (!strncmp(s, "xml", 3))
-        {
-            strcpy(type, "xml");
-        }
         else if (!strncmp(s, "txml", 4))
         {
             const char *cp = strchr(s, ';');
             yaz_snprintf(type, 80, "txml; charset=%s", cp ? cp+1 : "marc-8s");
         }
-        else
+        else /* pass verbatim to ZOOM - including "xml" */
             strcpy(type, s);
         return 0;
     }
@@ -546,11 +543,10 @@ void client_search_response(struct client *cl)
     }
     else
     {
-        yaz_log(YLOG_DEBUG, "client_search_response: hits "
-                ODR_INT_PRINTF, cl->hits);
         client_report_facets(cl, resultset);
         cl->record_offset = cl->startrecs;
         cl->hits = ZOOM_resultset_size(resultset);
+        yaz_log(YLOG_DEBUG, "client_search_response: hits " ODR_INT_PRINTF, cl->hits);
         if (cl->suggestions)
             client_suggestions_destroy(cl);
         cl->suggestions = client_suggestions_create(ZOOM_resultset_option_get(resultset, "suggestions"));
@@ -589,8 +585,7 @@ static void client_record_ingest(struct client *cl)
         else if (ZOOM_record_error(rec, &msg, &addinfo, 0))
         {
             yaz_log(YLOG_WARN, "Record error %s (%s): %s (rec #%d)",
-                    msg, addinfo, client_get_id(cl),
-                    cl->record_offset);
+                    msg, addinfo, client_get_id(cl), cl->record_offset);
         }
         else
         {
@@ -609,8 +604,11 @@ static void client_record_ingest(struct client *cl)
             else
             {
                 /* OK = 0, -1 = failure, -2 = Filtered */
-                if (ingest_record(cl, xmlrec, cl->record_offset, nmem) == -1)
+                int rc = ingest_record(cl, xmlrec, cl->record_offset, nmem);
+                if (rc == -1)
                     yaz_log(YLOG_WARN, "Failed to ingest from %s", client_get_id(cl));
+                if (rc == -2)
+                    cl->filtered += 1;
             }
             nmem_destroy(nmem);
         }
@@ -663,6 +661,7 @@ void client_reingest(struct client *cl)
 {
     int i = cl->startrecs;
     int to = cl->record_offset;
+    cl->filtered = 0;
 
     cl->record_offset = i;
     for (; i < to; i++)
@@ -692,7 +691,7 @@ static void client_set_facets_request(struct client *cl, ZOOM_connection link)
                 wrbuf_puts(w, ",");
         }
     }
-    yaz_log(YLOG_LOG, "using facets str: %s", wrbuf_cstr(w));
+    yaz_log(YLOG_DEBUG, "using facets str: %s", wrbuf_cstr(w));
     ZOOM_connection_option_set(link, "facets",
                                wrbuf_len(w) ? wrbuf_cstr(w) : 0);
     wrbuf_destroy(w);
@@ -753,12 +752,18 @@ void client_start_search(struct client *cl)
     const char *opt_sort        = session_setting_oneval(sdb, PZ_SORT);
     const char *opt_preferred   = session_setting_oneval(sdb, PZ_PREFERRED);
     const char *extra_args      = session_setting_oneval(sdb, PZ_EXTRA_ARGS);
-    char maxrecs_str[24], startrecs_str[24];
+    const char *opt_present_chunk = session_setting_oneval(sdb, PZ_PRESENT_CHUNK);
     ZOOM_query q;
-
+    char maxrecs_str[24], startrecs_str[24], present_chunk_str[24];
+    int present_chunk = 20; // Default chunk size
+    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);
+    }
     assert(link);
 
     cl->diagnostic = 0;
+    cl->filtered = 0;
 
     if (extra_args && *extra_args)
         ZOOM_connection_option_set(link, "extraArgs", extra_args);
@@ -769,7 +774,6 @@ void client_start_search(struct client *cl)
             yaz_log(YLOG_LOG, "Target %s has preferred status: %d",
                     client_get_id(cl), cl->preferred);
     }
-    client_set_state(cl, Client_Working);
 
     if (*opt_piggyback)
         ZOOM_connection_option_set(link, "piggyback", opt_piggyback);
@@ -793,11 +797,16 @@ void client_start_search(struct client *cl)
     sprintf(maxrecs_str, "%d", cl->maxrecs);
     ZOOM_connection_option_set(link, "count", maxrecs_str);
 
-    if (cl->maxrecs > 20)
-        ZOOM_connection_option_set(link, "presentChunk", "20");
-    else
+    /* A present_chunk less than 1 will disable chunking. */
+    if (present_chunk > 0 && cl->maxrecs > present_chunk) {
+        sprintf(present_chunk_str, "%d", present_chunk);
+        ZOOM_connection_option_set(link, "presentChunk", present_chunk_str);
+        yaz_log(YLOG_DEBUG, "Present chunk set to %s", present_chunk_str);
+    }
+    else {
         ZOOM_connection_option_set(link, "presentChunk", maxrecs_str);
-
+        yaz_log(YLOG_DEBUG, "Present chunk set to %s (maxrecs)", maxrecs_str);
+    }
     sprintf(startrecs_str, "%d", cl->startrecs);
     ZOOM_connection_option_set(link, "start", startrecs_str);
 
@@ -850,11 +859,11 @@ void client_start_search(struct client *cl)
             if (se->sorted_results->next)
             {
                 ZOOM_query_destroy(q);
-                client_set_state_nb(cl, Client_Idle);
                 return;
             }
         }
     }
+    client_set_state(cl, Client_Working);
     cl->hits = 0;
     cl->record_offset = 0;
     rs = ZOOM_connection_search(link, q);
@@ -877,6 +886,7 @@ struct client *client_create(const char *id)
     cl->session = 0;
     cl->hits = 0;
     cl->record_offset = 0;
+    cl->filtered = 0;
     cl->diagnostic = 0;
     cl->state = Client_Disconnected;
     cl->show_raw = 0;
@@ -968,7 +978,7 @@ void client_disconnect(struct client *cl)
 
 
 // Initialize CCL map for a target
-static CCL_bibset prepare_cclmap(struct client *cl)
+static CCL_bibset prepare_cclmap(struct client *cl, CCL_bibset base_bibset)
 {
     struct session_database *sdb = client_get_database(cl);
     struct setting *s;
@@ -976,7 +986,10 @@ static CCL_bibset prepare_cclmap(struct client *cl)
 
     if (!sdb->settings)
         return 0;
-    res = ccl_qual_mk();
+    if (base_bibset)
+        res = ccl_qual_dup(base_bibset);
+    else
+        res = ccl_qual_mk();
     for (s = sdb->settings[PZ_CCLMAP]; s; s = s->next)
     {
         char *p = strchr(s->name + 3, ':');
@@ -993,60 +1006,44 @@ static CCL_bibset prepare_cclmap(struct client *cl)
 }
 
 // returns a xmalloced CQL query corresponding to the pquery in client
-static char *make_cqlquery(struct client *cl)
+static char *make_cqlquery(struct client *cl, Z_RPNQuery *zquery)
 {
     cql_transform_t cqlt = cql_transform_create();
-    Z_RPNQuery *zquery;
-    char *r;
+    char *r = 0;
     WRBUF wrb = wrbuf_alloc();
     int status;
-    ODR odr_out = odr_createmem(ODR_ENCODE);
 
-    zquery = p_query_rpn(odr_out, cl->pquery);
-    yaz_log(YLOG_LOG, "PQF: %s", cl->pquery);
     if ((status = cql_transform_rpn2cql_wrbuf(cqlt, wrb, zquery)))
     {
         yaz_log(YLOG_WARN, "Failed to generate CQL query, code=%d", status);
-        r = 0;
     }
     else
     {
         r = xstrdup(wrbuf_cstr(wrb));
     }     
     wrbuf_destroy(wrb);
-    odr_destroy(odr_out);
     cql_transform_close(cqlt);
     return r;
 }
 
 // returns a xmalloced SOLR query corresponding to the pquery in client
 // TODO Could prob. be merge with the similar make_cqlquery
-static char *make_solrquery(struct client *cl)
+static char *make_solrquery(struct client *cl, Z_RPNQuery *zquery)
 {
     solr_transform_t sqlt = solr_transform_create();
-    Z_RPNQuery *zquery;
-    char *r;
+    char *r = 0;
     WRBUF wrb = wrbuf_alloc();
     int status;
-    ODR odr_out = odr_createmem(ODR_ENCODE);
-
-    zquery = p_query_rpn(odr_out, cl->pquery);
-    if (zquery == 0) {
-        yaz_log(YLOG_WARN, "Failed to generate RPN from PQF: %s", cl->pquery);
-        return 0;
-    }
-    yaz_log(YLOG_LOG, "PQF: %s", cl->pquery);
+    
     if ((status = solr_transform_rpn2solr_wrbuf(sqlt, wrb, zquery)))
     {
-        yaz_log(YLOG_WARN, "Failed to generate SOLR query from PQF %s, code=%d", cl->pquery, status);
-        r = 0;
+        yaz_log(YLOG_WARN, "Failed to generate SOLR query, code=%d", status);
     }
     else
     {
         r = xstrdup(wrbuf_cstr(wrb));
     }
     wrbuf_destroy(wrb);
-    odr_destroy(odr_out);
     solr_transform_close(sqlt);
     return r;
 }
@@ -1084,17 +1081,23 @@ 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)
+                       WRBUF w_pqf, WRBUF w_ccl,
+                       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++)
     {
         struct setting *s = 0;
-        
+        nmem_reset(nmem_tmp);
         for (s = sdb->settings[PZ_LIMITMAP]; s; s = s->next)
         {
             const char *p = strchr(s->name + 3, ':');
@@ -1123,13 +1126,13 @@ static int apply_limit(struct session_database *sdb,
                 else if (!strncmp(s->value, "ccl:", 4))
                 {
                     const char *ccl = s->value + 4;
-
-                    wrbuf_puts(w_ccl, " and (");
+                    
+                    wrbuf_printf(w_ccl, " %s (", and_op);
 
                     for (i = 0; i < num; i++)
                     {
                         if (i)
-                            wrbuf_puts(w_ccl, " or ");
+                            wrbuf_printf(w_ccl, " %s ", or_op);
                         wrbuf_puts(w_ccl, ccl);
                         wrbuf_puts(w_ccl, "=\"");
                         wrbuf_puts(w_ccl, values[i]);
@@ -1150,7 +1153,6 @@ static int apply_limit(struct session_database *sdb,
                 break;
             }
         }
-        nmem_reset(nmem_tmp);
         if (!s)
         {
             yaz_log(YLOG_WARN, "Target %s: limit %s used, but no limitmap defined",
@@ -1162,26 +1164,32 @@ static int apply_limit(struct session_database *sdb,
 }
                         
 // 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
+// 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)
+                       const char *startrecs, const char *maxrecs,
+                       CCL_bibset bibset)
 {
     struct session *se = client_get_session(cl);
     struct session_database *sdb = client_get_database(cl);
     struct ccl_rpn_node *cn;
     int cerror, cpos;
-    CCL_bibset ccl_map = prepare_cclmap(cl);
+    ODR odr_out;
+    CCL_bibset ccl_map = prepare_cclmap(cl, bibset);
     const char *sru = session_setting_oneval(sdb, PZ_SRU);
     const char *pqf_prefix = session_setting_oneval(sdb, PZ_PQF_PREFIX);
     const char *pqf_strftime = session_setting_oneval(sdb, PZ_PQF_STRFTIME);
     const char *query_syntax = session_setting_oneval(sdb, PZ_QUERY_SYNTAX);
     WRBUF w_ccl, w_pqf;
     int ret_value = 1;
+    Z_RPNQuery *zquery;
 
     if (!ccl_map)
         return -1;
 
-
     if (maxrecs && atoi(maxrecs) != cl->maxrecs)
     {
         ret_value = 0;
@@ -1204,8 +1212,11 @@ int client_parse_query(struct client *cl, const char *query,
         wrbuf_puts(w_pqf, " ");
     }
 
-    if (apply_limit(sdb, facet_limits, w_pqf, w_ccl))
+    if (apply_limit(sdb, facet_limits, w_pqf, w_ccl, ccl_map))
+    {
+        ccl_qual_rm(&ccl_map);
         return -2;
+    }
 
     facet_limits_destroy(cl->facet_limits);
     cl->facet_limits = facet_limits_dup(facet_limits);
@@ -1245,6 +1256,7 @@ int client_parse_query(struct client *cl, const char *query,
                 wrbuf_putc(w_pqf, cp[0]);
         }
     }
+
     if (!cl->pquery || strcmp(cl->pquery, wrbuf_cstr(w_pqf)))
     {
         xfree(cl->pquery);
@@ -1252,36 +1264,43 @@ int client_parse_query(struct client *cl, const char *query,
         ret_value = 0;
     }
     wrbuf_destroy(w_pqf);
-
-    yaz_log(YLOG_LOG, "PQF query: %s", cl->pquery);
-
+    
     xfree(cl->cqlquery);
+    cl->cqlquery = 0;
 
-    /* Support for PQF on SRU targets. */
-    /* TODO Refactor */
-    yaz_log(YLOG_DEBUG, "Query syntax: %s", query_syntax);
-    if (strcmp(query_syntax, "pqf") != 0 && *sru)
+    odr_out = odr_createmem(ODR_ENCODE);    
+    zquery = p_query_rpn(odr_out, cl->pquery);
+    if (!zquery)
     {
-        if (!strcmp(sru, "solr")) {
-            if (!(cl->cqlquery = make_solrquery(cl)))
-                return -1;
-        }
-        else {
-            if (!(cl->cqlquery = make_cqlquery(cl)))
-                return -1;
-        }
+
+        session_log(se, YLOG_WARN, "Invalid PQF query for %s: %s",
+                    client_get_id(cl), cl->pquery);
+        ret_value = -1;
     }
     else
-        cl->cqlquery = 0;
+    {
+        session_log(se, YLOG_LOG, "PQF for %s: %s",
+                    client_get_id(cl), cl->pquery);
+        
+        /* Support for PQF on SRU targets. */
+        if (strcmp(query_syntax, "pqf") != 0 && *sru)
+        {
+            if (!strcmp(sru, "solr"))
+                cl->cqlquery = make_solrquery(cl, zquery);
+            else
+                cl->cqlquery = make_cqlquery(cl, zquery);
+            if (!cl->cqlquery)
+                ret_value = -1;
+        }
+    }
+    odr_destroy(odr_out);
 
     /* TODO FIX Not thread safe */
     if (!se->relevance)
     {
         // Initialize relevance structure with query terms
-        se->relevance = relevance_create_ccl(
-            se->service->charsets, se->nmem, cn);
+        se->relevance = relevance_create_ccl(se->service->charsets, cn);
     }
-
     ccl_rpn_delete(cn);
     return ret_value;
 }
@@ -1316,11 +1335,26 @@ Odr_int client_get_hits(struct client *cl)
     return cl->hits;
 }
 
+Odr_int client_get_approximation(struct client *cl)
+{
+    if (cl->record_offset > 0) {
+        Odr_int approx = ((10 * cl->hits * (cl->record_offset - cl->filtered)) / cl->record_offset + 5) /10;
+        yaz_log(YLOG_DEBUG, "%s: Approx: %lld * %d / %d = %lld ", client_get_id(cl), cl->hits, cl->record_offset - cl->filtered, cl->record_offset, approx);
+        return approx;
+    }
+    return cl->hits;
+}
+
 int client_get_num_records(struct client *cl)
 {
     return cl->record_offset;
 }
 
+int client_get_num_records_filtered(struct client *cl)
+{
+    return cl->filtered;
+}
+
 void client_set_diagnostic(struct client *cl, int diagnostic,
                            const char *addinfo)
 {