Working fetch more, but settings/tests needs updating
[pazpar2-moved-to-github.git] / src / client.c
index 3627cb9..d3265c3 100644 (file)
@@ -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
@@ -33,6 +33,9 @@ Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
 #if HAVE_UNISTD_H
 #include <unistd.h>
 #endif
+#ifdef WIN32
+#include <windows.h>
+#endif
 #include <signal.h>
 #include <assert.h>
 
@@ -116,6 +119,7 @@ struct client {
     int maxrecs;
     int startrecs;
     int diagnostic;
+    char *message;
     int preferred;
     struct suggestions *suggestions;
     enum client_state state;
@@ -579,17 +583,18 @@ static void client_record_ingest(struct client *cl)
     const char *msg, *addinfo;
     ZOOM_record rec = 0;
     ZOOM_resultset resultset = cl->resultset;
-    int offset = cl->record_offset;
-    if ((rec = ZOOM_resultset_record_immediate(resultset, offset)))
+    struct session *se = client_get_session(cl);
+
+    if ((rec = ZOOM_resultset_record_immediate(resultset, cl->record_offset)))
     {
-        cl->record_offset++;
+        int offset = ++cl->record_offset;
         if (cl->session == 0) {
             /* no operation */
         }
         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);
+            session_log(se, YLOG_WARN, "Record error %s (%s): %s #%d",
+                        msg, addinfo, client_get_id(cl), offset);
         }
         else
         {
@@ -600,17 +605,33 @@ static void client_record_ingest(struct client *cl)
 
             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");
+                session_log(se, YLOG_WARN, "Failed to determine record type");
             xmlrec = ZOOM_record_get(rec, type, NULL);
             if (!xmlrec)
-                yaz_log(YLOG_WARN, "ZOOM_record_get failed from %s",
-                        client_get_id(cl));
+            {
+                const char *rec_syn =  ZOOM_record_get(rec, "syntax", NULL);
+                session_log(se, YLOG_WARN, "ZOOM_record_get failed from %s #%d",
+                            client_get_id(cl), offset);
+                session_log(se, YLOG_LOG, "pz:nativesyntax=%s . "
+                            "ZOOM record type=%s . Actual record syntax=%s",
+                            s ? s : "null", type,
+                            rec_syn ? rec_syn : "null");
+            }
             else
             {
                 /* OK = 0, -1 = failure, -2 = Filtered */
                 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));
+                {
+                    const char *rec_syn =  ZOOM_record_get(rec, "syntax", NULL);
+                    session_log(se, YLOG_WARN,
+                                "Failed to ingest record from %s #%d",
+                                client_get_id(cl), offset);
+                    session_log(se, YLOG_LOG, "pz:nativesyntax=%s . "
+                                "ZOOM record type=%s . Actual record syntax=%s",
+                                s ? s : "null", type,
+                                rec_syn ? rec_syn : "null");
+                }
                 if (rc == -2)
                     cl->filtered += 1;
             }
@@ -619,7 +640,8 @@ static void client_record_ingest(struct client *cl)
     }
     else
     {
-        yaz_log(YLOG_WARN, "Expected record, but got NULL, offset=%d", offset);
+        session_log(se, YLOG_WARN, "Got NULL record from %s #%d",
+                    client_get_id(cl), cl->record_offset);
     }
 }
 
@@ -740,6 +762,35 @@ static const char *get_strategy_plus_sort(struct client *l, const char *field)
     return strategy_plus_sort;
 }
 
+int client_fetch_more(struct client *cl)
+{
+    struct session_database *sdb = client_get_database(cl);
+    const char *str;
+    int extra = cl->hits - cl->record_offset;
+
+    if (extra > 0)
+    {
+        ZOOM_resultset set = cl->resultset;
+        struct connection *co = client_get_connection(cl);
+        int max_extra = 10;
+
+        if (extra > max_extra)
+            extra = max_extra;
+
+        str = session_setting_oneval(sdb, PZ_REQUESTSYNTAX);
+        ZOOM_resultset_option_set(set, "preferredRecordSyntax", str);
+        str = session_setting_oneval(sdb, PZ_ELEMENTS);
+        if (str && *str)
+            ZOOM_resultset_option_set(set, "elementSetName", str);
+
+        ZOOM_resultset_records(set, 0, cl->record_offset, extra);
+        client_set_state(cl, Client_Working);
+        connection_continue(co);
+        return 1;
+    }
+    return 0;
+}
+
 int client_parse_init(struct client *cl, int same_search)
 {
     cl->same_search = same_search;
@@ -912,6 +963,7 @@ struct client *client_create(const char *id)
     cl->pquery = 0;
     cl->cqlquery = 0;
     cl->addinfo = 0;
+    cl->message = 0;
     cl->database = 0;
     cl->connection = 0;
     cl->session = 0;
@@ -968,7 +1020,11 @@ int client_destroy(struct client *c)
             c->cqlquery = 0;
             xfree(c->addinfo);
             c->addinfo = 0;
+            xfree(c->message);
+            c->message = 0;
             xfree(c->id);
+            xfree(c->sort_strategy);
+            xfree(c->sort_criteria);
             assert(!c->connection);
             facet_limits_destroy(c->facet_limits);
 
@@ -1032,7 +1088,9 @@ static CCL_bibset prepare_cclmap(struct client *cl, CCL_bibset base_bibset)
             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_diagnostic(cl, ZOOM_ERROR_CCL_CONFIG,
+                                  ZOOM_diag_str(ZOOM_ERROR_CCL_CONFIG),
+                                  wrbuf_cstr(w));
             client_set_state_nb(cl, Client_Error);
             ccl_qual_rm(&res);
             wrbuf_destroy(w);
@@ -1046,7 +1104,9 @@ static CCL_bibset prepare_cclmap(struct client *cl, CCL_bibset base_bibset)
             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_diagnostic(cl, ZOOM_ERROR_CCL_CONFIG,
+                                  ZOOM_diag_str(ZOOM_ERROR_CCL_CONFIG),
+                                  wrbuf_cstr(w));
             client_set_state_nb(cl, Client_Error);
             ccl_qual_rm(&res);
             wrbuf_destroy(w);
@@ -1113,17 +1173,28 @@ const char *client_get_facet_limit_local(struct client *cl,
         for (s = sdb->settings[PZ_LIMITMAP]; s; s = s->next)
         {
             const char *p = strchr(s->name + 3, ':');
-            if (p && !strcmp(p + 1, name) && s->value &&
-                !strncmp(s->value, "local:", 6))
+            if (p && !strcmp(p + 1, name) && s->value)
             {
-                const char *cp = s->value + 6;
-                while (*cp == ' ')
-                    cp++;
-
-                nmem_strsplit_escape2(nmem, "|", value, values,
-                                      num, 1, '\\', 1);
-                (*l)++;
-                return *cp ? cp : name;
+                int j, cnum;
+                char **cvalues;
+                nmem_strsplit_escape2(nmem, ",", s->value, &cvalues,
+                                      &cnum, 1, '\\', 1);
+                for (j = 0; j < cnum; j++)
+                {
+                    const char *cvalue = cvalues[j];
+                    while (*cvalue == ' ')
+                        cvalue++;
+                    if (!strncmp(cvalue, "local:", 6))
+                    {
+                        const char *cp = cvalue + 6;
+                        while (*cp == ' ')
+                            cp++;
+                        nmem_strsplit_escape2(nmem, "|", value, values,
+                                              num, 1, '\\', 1);
+                        (*l)++;
+                        return *cp ? cp : name;
+                    }
+                }
             }
         }
     }
@@ -1132,7 +1203,8 @@ 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, CCL_bibset ccl_map)
+                       WRBUF w_pqf, CCL_bibset ccl_map,
+                       struct conf_service *service)
 {
     int ret = 0;
     int i = 0;
@@ -1144,6 +1216,7 @@ static int apply_limit(struct session_database *sdb,
     {
         struct setting *s = 0;
         nmem_reset(nmem_tmp);
+        /* name="pz:limitmap:author" value="rpn:@attr 1=4|local:other" */
         for (s = sdb->settings[PZ_LIMITMAP]; s; s = s->next)
         {
             const char *p = strchr(s->name + 3, ':');
@@ -1151,72 +1224,96 @@ static int apply_limit(struct session_database *sdb,
             {
                 char **values = 0;
                 int i, num = 0;
+                char **cvalues = 0;
+                int j, cnum = 0;
                 nmem_strsplit_escape2(nmem_tmp, "|", value, &values,
                                       &num, 1, '\\', 1);
 
-                if (!strncmp(s->value, "rpn:", 4))
-                {
-                    const char *pqf = s->value + 4;
+                nmem_strsplit_escape2(nmem_tmp, ",", s->value, &cvalues,
+                                      &cnum, 1, '\\', 1);
 
-                    wrbuf_puts(w_pqf, "@and ");
-                    wrbuf_puts(w_pqf, pqf);
-                    wrbuf_puts(w_pqf, " ");
-                    for (i = 0; i < num; i++)
-                    {
-                        if (i < num - 1)
-                            wrbuf_puts(w_pqf, "@or ");
-                        yaz_encode_pqf_term(w_pqf, values[i],
-                                            strlen(values[i]));
-                    }
-                }
-                else if (!strncmp(s->value, "ccl:", 4))
+                for (j = 0; ret == 0 && j < cnum; j++)
                 {
-                    const char *ccl = s->value + 4;
-                    WRBUF ccl_w = wrbuf_alloc();
-                    for (i = 0; i < num; i++)
+                    const char *cvalue = cvalues[j];
+                    while (*cvalue == ' ')
+                        cvalue++;
+                    if (!strncmp(cvalue, "rpn:", 4))
                     {
-                        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)
+                        const char *pqf = cvalue + 4;
+                        wrbuf_puts(w_pqf, "@and ");
+                        wrbuf_puts(w_pqf, pqf);
+                        wrbuf_puts(w_pqf, " ");
+                        for (i = 0; i < num; i++)
                         {
-                            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_pqf, "@or ");
+                            yaz_encode_pqf_term(w_pqf, values[i],
+                                                strlen(values[i]));
                         }
                     }
-                    wrbuf_destroy(ccl_w);
-                }
-                else if (!strncmp(s->value, "local:", 6)) {
-                    /* no operation */
-                }
-                else
-                {
-                    yaz_log(YLOG_WARN, "Target %s: Bad limitmap '%s'",
-                            sdb->database->id, s->value);
-                    ret = -1; /* bad limitmap */
+                    else if (!strncmp(cvalue, "ccl:", 4))
+                    {
+                        const char *ccl = cvalue + 4;
+                        WRBUF ccl_w = wrbuf_alloc();
+                        for (i = 0; i < num; i++)
+                        {
+                            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_destroy(ccl_w);
+                    }
+                    else if (!strncmp(cvalue, "local:", 6)) {
+                        /* no operation */
+                    }
+                    else
+                    {
+                        yaz_log(YLOG_WARN, "Target %s: Bad limitmap '%s'",
+                                sdb->database->id, cvalue);
+                        ret = -1; /* bad limitmap */
+                    }
+                    break;
                 }
-                break;
             }
         }
         if (!s)
         {
-            yaz_log(YLOG_WARN, "Target %s: limit %s used, but no limitmap defined",
-                    (sdb->database ? sdb->database->id : "<no id>"), name);
+            int i;
+            for (i = 0; i < service->num_metadata; i++)
+            {
+                struct conf_metadata *md = service->metadata + i;
+                if (!strcmp(md->name, name) && md->limitcluster)
+                {
+                    yaz_log(YLOG_LOG, "limitcluster in use for %s",
+                            md->name);
+                    break;
+                }
+            }
+            if (i == service->num_metadata)
+            {
+                yaz_log(YLOG_WARN, "Target %s: limit %s used, but no limitmap defined",
+                        (sdb->database ? sdb->database->id : "<no id>"), name);
+            }
         }
     }
     nmem_destroy(nmem_tmp);
@@ -1224,20 +1321,21 @@ 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
+// 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,
-                       CCL_bibset bibset)
+                       facet_limits_t facet_limits)
 {
     struct session *se = client_get_session(cl);
+    struct conf_service *service = se->service;
     struct session_database *sdb = client_get_database(cl);
     struct ccl_rpn_node *cn;
     int cerror, cpos;
     ODR odr_out;
-    CCL_bibset ccl_map = prepare_cclmap(cl, bibset);
+    CCL_bibset ccl_map = prepare_cclmap(cl, service->ccl_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);
@@ -1259,7 +1357,7 @@ int client_parse_query(struct client *cl, const char *query,
         wrbuf_puts(w_pqf, " ");
     }
 
-    if (apply_limit(sdb, facet_limits, w_pqf, ccl_map))
+    if (apply_limit(sdb, facet_limits, w_pqf, ccl_map, service))
     {
         ccl_qual_rm(&ccl_map);
         return -2;
@@ -1268,7 +1366,7 @@ 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, "Client %s: CCL query: %s", client_get_id(cl), 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)
@@ -1307,9 +1405,15 @@ 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);
 
@@ -1339,6 +1443,9 @@ 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);
@@ -1359,14 +1466,11 @@ int client_parse_query(struct client *cl, const char *query,
 
 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;
+            get_strategy_plus_sort(cl, sp->name);
+        int increasing = sp->increasing;
         if (sort_strategy_and_spec && strlen(sort_strategy_and_spec) < 40)
         {
             char strategy[50], *p;
@@ -1388,23 +1492,29 @@ int client_parse_sort(struct client *cl, struct reclist_sortparms *sp)
                 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);
+                    xfree(cl->sort_strategy);
+                    cl->sort_strategy = xstrdup(strategy);
+                    xfree(cl->sort_criteria);
+                    cl->sort_criteria = xstrdup(p);
                 }
             }
             else {
                 yaz_log(YLOG_LOG, "Client %s: Invalid sort strategy and spec found %s", client_get_id(cl), sort_strategy_and_spec);
+                xfree(cl->sort_strategy);
                 cl->sort_strategy  = 0;
+                xfree(cl->sort_criteria);
                 cl->sort_criteria = 0;
             }
         } else {
-            yaz_log(YLOG_LOG, "Client %s: No sort strategy and spec found.", client_get_id(cl));
+            yaz_log(YLOG_DEBUG, "Client %s: No sort strategy and spec found.", client_get_id(cl));
+            xfree(cl->sort_strategy);
             cl->sort_strategy  = 0;
+            xfree(cl->sort_criteria);
             cl->sort_criteria = 0;
         }
 
     }
-    return cl->same_search;
+    return !cl->same_search;
 }
 
 void client_set_session(struct client *cl, struct session *se)
@@ -1458,17 +1568,22 @@ int client_get_num_records_filtered(struct client *cl)
 }
 
 void client_set_diagnostic(struct client *cl, int diagnostic,
-                           const char *addinfo)
+                           const char *message, const char *addinfo)
 {
     cl->diagnostic = diagnostic;
+    xfree(cl->message);
+    cl->message = xstrdup(message);
     xfree(cl->addinfo);
     cl->addinfo = 0;
     if (addinfo)
         cl->addinfo = xstrdup(addinfo);
 }
 
-int client_get_diagnostic(struct client *cl, const char **addinfo)
+int client_get_diagnostic(struct client *cl, const char **message,
+                          const char **addinfo)
 {
+    if (message)
+        *message = cl->message;
     if (addinfo)
         *addinfo = cl->addinfo;
     return cl->diagnostic;