zoom: Do not include ZOOM C's fake surrogate diag recs MP-547
[metaproxy-moved-to-github.git] / src / filter_zoom.cpp
index 0905765..db3719a 100644 (file)
@@ -1,5 +1,5 @@
 /* This file is part of Metaproxy.
-   Copyright (C) 2005-2012 Index Data
+   Copyright (C) Index Data
 
 Metaproxy 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
@@ -51,6 +51,7 @@ Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
 #include <yaz/sortspec.h>
 #include <yaz/tokenizer.h>
 #include <yaz/zoom.h>
+#include <yaz/otherinfo.h>
 
 namespace mp = metaproxy_1;
 namespace yf = mp::filter;
@@ -60,6 +61,7 @@ namespace metaproxy_1 {
         class Zoom::Searchable : boost::noncopyable {
           public:
             std::string authentication;
+            std::string authenticationMode;
             std::string cfAuth;
             std::string cfProxy;
             std::string cfSubDB;
@@ -107,10 +109,11 @@ namespace metaproxy_1 {
             void connect(std::string zurl, int *error, char **addinfo,
                          ODR odr);
             void search(ZOOM_query q, Odr_int *hits,
-                        int *error, char **addinfo, ODR odr);
+                        int *error, char **addinfo, Z_FacetList **fl, ODR odr);
             void present(Odr_int start, Odr_int number, ZOOM_record *recs,
                          int *error, char **addinfo, ODR odr);
             void set_option(const char *name, const char *value);
+            void set_option(const char *name, const char *value, size_t l);
             void set_option(const char *name, std::string value);
             const char *get_option(const char *name);
             void get_zoom_error(int *error, char **addinfo, ODR odr);
@@ -125,7 +128,6 @@ namespace metaproxy_1 {
             BackendPtr m_backend;
             void handle_package(mp::Package &package);
             void handle_search(mp::Package &package);
-
             void auth(mp::Package &package, Z_InitRequest *req,
                       int *error, char **addinfo, ODR odr);
 
@@ -370,15 +372,84 @@ void yf::Zoom::Backend::connect(std::string zurl,
 }
 
 void yf::Zoom::Backend::search(ZOOM_query q, Odr_int *hits,
-                               int *error, char **addinfo, ODR odr)
+                               int *error, char **addinfo, Z_FacetList **flp,
+                               ODR odr)
 {
     ZOOM_resultset_destroy(m_resultset);
+    m_resultset = 0;
+    if (*flp)
+    {
+        WRBUF w = wrbuf_alloc();
+        yaz_facet_list_to_wrbuf(w, *flp);
+        set_option("facets", wrbuf_cstr(w));
+        wrbuf_destroy(w);
+    }
     m_resultset = ZOOM_connection_search(m_connection, q);
     get_zoom_error(error, addinfo, odr);
     if (*error == 0)
         *hits = ZOOM_resultset_size(m_resultset);
     else
         *hits = 0;
+    *flp = 0;
+    size_t num_facets = ZOOM_resultset_facets_size(m_resultset);
+    if (num_facets > 0)
+    {
+        size_t i;
+        Z_FacetList *fl = (Z_FacetList *) odr_malloc(odr, sizeof(*fl));
+        fl->elements = (Z_FacetField **)
+            odr_malloc(odr, num_facets * sizeof(*fl->elements));
+        for (i = 0; i < num_facets; i++)
+        {
+            ZOOM_facet_field ff =
+                ZOOM_resultset_get_facet_field_by_index(m_resultset, i);
+            if (!ff)
+                break;
+            Z_AttributeList *al = (Z_AttributeList *)
+                odr_malloc(odr, sizeof(*al));
+            al->num_attributes = 1;
+            al->attributes = (Z_AttributeElement **)
+                odr_malloc(odr, sizeof(*al->attributes));
+            Z_AttributeElement *ae = al->attributes[0] = (Z_AttributeElement *)
+                odr_malloc(odr, sizeof(**al->attributes));
+            ae->attributeSet = 0;
+            ae->attributeType = odr_intdup(odr, 1);
+            ae->which = Z_AttributeValue_complex;
+            ae->value.complex = (Z_ComplexAttribute *)
+                odr_malloc(odr, sizeof(*ae->value.complex));
+            ae->value.complex->num_list = 1;
+            ae->value.complex->list = (Z_StringOrNumeric **)
+                odr_malloc(odr, sizeof(**ae->value.complex->list));
+            ae->value.complex->list[0] = (Z_StringOrNumeric *)
+                odr_malloc(odr, sizeof(**ae->value.complex->list));
+            ae->value.complex->list[0]->which = Z_StringOrNumeric_string;
+            ae->value.complex->list[0]->u.string =
+                odr_strdup(odr, ZOOM_facet_field_name(ff));
+            ae->value.complex->num_semanticAction = 0;
+            ae->value.complex->semanticAction = 0;
+
+            int num_terms = ZOOM_facet_field_term_count(ff);
+            fl->elements[i] = (Z_FacetField *)
+                odr_malloc(odr, sizeof(Z_FacetField));
+            fl->elements[i]->attributes = al;
+            fl->elements[i]->num_terms = num_terms;
+            fl->elements[i]->terms = (Z_FacetTerm **)
+                odr_malloc(odr, num_terms * sizeof(Z_FacetTerm *));
+            int j;
+            for (j = 0; j < num_terms; j++)
+            {
+                int freq;
+                const char *a_term = ZOOM_facet_field_get_term(ff, j, &freq);
+                Z_FacetTerm *ft = (Z_FacetTerm *) odr_malloc(odr, sizeof(*ft));
+                ft->term = z_Term_create(odr, Z_Term_general, a_term,
+                                         strlen(a_term));
+                ft->count = odr_intdup(odr, freq);
+
+                fl->elements[i]->terms[j] = ft;
+            }
+        }
+        fl->num = i;
+        *flp = fl;
+    }
 }
 
 void yf::Zoom::Backend::present(Odr_int start, Odr_int number,
@@ -389,6 +460,12 @@ void yf::Zoom::Backend::present(Odr_int start, Odr_int number,
     get_zoom_error(error, addinfo, odr);
 }
 
+
+void yf::Zoom::Backend::set_option(const char *name, const char *value, size_t l)
+{
+    ZOOM_connection_option_setl(m_connection, name, value, l);
+}
+
 void yf::Zoom::Backend::set_option(const char *name, const char *value)
 {
     ZOOM_connection_option_set(m_connection, name, value);
@@ -508,6 +585,11 @@ yf::Zoom::SearchablePtr yf::Zoom::Impl::parse_torus_record(const xmlNode *ptr)
             s->authentication = mp::xml::get_text(ptr);
         }
         else if (!strcmp((const char *) ptr->name,
+                         "authenticationMode"))
+        {
+            s->authenticationMode = mp::xml::get_text(ptr);
+        }
+        else if (!strcmp((const char *) ptr->name,
                          "cfAuth"))
         {
             s->cfAuth = mp::xml::get_text(ptr);
@@ -1175,15 +1257,17 @@ yf::Zoom::BackendPtr yf::Zoom::Frontend::get_backend_from_databases(
         sptr = it->second;
     else if (torus_url.length() > 0)
     {
+        std::string torus_addinfo;
         std::string torus_query = "udb==" + torus_db;
         xmlDoc *doc = mp::get_searchable(package,torus_url, torus_db,
                                          torus_query,
-                                         realm, m_p->proxy);
+                                         realm, m_p->proxy,
+                                         torus_addinfo);
         if (!doc)
         {
             *error = YAZ_BIB1_UNSPECIFIED_ERROR;
-            *addinfo = odr_strdup(odr, "Torus server unavailable or "
-                                  "incorrectly configured");
+            if (torus_addinfo.length())
+                *addinfo = odr_strdup(odr, torus_addinfo.c_str());
             BackendPtr b;
             return b;
         }
@@ -1345,8 +1429,7 @@ yf::Zoom::BackendPtr yf::Zoom::Frontend::get_backend_from_databases(
     if (sptr->query_encoding.length())
         b->set_option("rpnCharset", sptr->query_encoding);
 
-    if (sptr->extraArgs.length())
-        b->set_option("extraArgs", sptr->extraArgs);
+    std::string extraArgs = sptr->extraArgs;
 
     b->set_option("timeout", m_p->zoom_timeout.c_str());
 
@@ -1399,6 +1482,9 @@ yf::Zoom::BackendPtr yf::Zoom::Frontend::get_backend_from_databases(
             out_names[no_out_args] = "subdatabase";
             out_values[no_out_args++] = odr_strdup(odr, sptr->cfSubDB.c_str());
         }
+        if (!param_nocproxy && b->sptr->contentConnector.length())
+            param_nocproxy = "1";
+
         if (param_nocproxy)
         {
             out_names[no_out_args] = "nocproxy";
@@ -1407,25 +1493,42 @@ yf::Zoom::BackendPtr yf::Zoom::Frontend::get_backend_from_databases(
     }
     else
     {
-        size_t found = authentication.find('/');
-
-        if (sptr->sru.length() && found != std::string::npos)
+        const char *auth = authentication.c_str();
+        const char *cp1 = strchr(auth, ' ');
+        if (!cp1 && sptr->sru.length())
+            cp1 =  strchr(auth, '/');
+        if (!cp1)
         {
-            b->set_option("user", authentication.substr(0, found));
-            b->set_option("password", authentication.substr(found+1));
+            /* Z39.50 user/password style, or no password for SRU */
+            b->set_option("user", auth);
         }
         else
-            b->set_option("user", authentication);
+        {
+            /* now consider group as well */
+            const char *cp2 = strchr(cp1 + 1, ' ');
 
+            b->set_option("user", auth, cp1 - auth);
+            if (!cp2)
+                b->set_option("password", cp1 + 1);
+            else
+            {
+                b->set_option("group", cp1 + 1, cp2 - cp1 - 1);
+                b->set_option("password", cp2 + 1);
+            }
+        }
+        if (sptr->authenticationMode.length())
+            b->set_option("authenticationMode", sptr->authenticationMode);
         if (proxy.length())
             b->set_option("proxy", proxy);
     }
+    if (extraArgs.length())
+        b->set_option("extraArgs", extraArgs);
+
     std::string url(sptr->target);
     if (sptr->sru.length())
     {
-        yaz_log(YLOG_LOG, "Got URL %s", url.c_str());
         b->set_option("sru", sptr->sru);
-        if (url.find_first_of("://") == std::string::npos)
+        if (url.find("://") == std::string::npos)
             url = "http://" + url;
         if (sptr->sru_version.length())
             b->set_option("sru_version", sptr->sru_version);
@@ -1629,8 +1732,19 @@ Z_Records *yf::Zoom::Frontend::get_records(mp::Package &package,
     if (!*error)
     {
         for (i = 0; i < number_to_present; i++)
+        {
             if (!recs[i])
                 break;
+
+            const char *addinfo;
+            int sur_error = ZOOM_record_error(recs[i], 0 /* msg */,
+                                              &addinfo, 0 /* diagset */);
+            if (sur_error ==
+                YAZ_BIB1_SYSTEM_ERROR_IN_PRESENTING_RECORDS && addinfo &&
+                !strcmp(addinfo,
+                        "ZOOM C generated. Present phase and no records"))
+                break;
+        }
     }
     if (i > 0)
     {  // only return records if no error and at least one record
@@ -1661,7 +1775,7 @@ Z_Records *yf::Zoom::Frontend::get_records(mp::Package &package,
         npl->num_records = i;
         npl->records = (Z_NamePlusRecord **)
             odr_malloc(odr, i * sizeof(*npl->records));
-        for (i = 0; i < number_to_present; i++)
+        for (i = 0; i < npl->num_records; i++)
         {
             Z_NamePlusRecord *npr = 0;
             const char *addinfo;
@@ -1912,10 +2026,12 @@ yf::Zoom::BackendPtr yf::Zoom::Frontend::explain_search(mp::Package &package,
     else if (query->which == Z_Query_type_104 &&
         query->u.type_104->which == Z_External_CQL)
     {
+        std::string torus_addinfo;
         std::string torus_query(query->u.type_104->u.cql);
         xmlDoc *doc = mp::get_searchable(package, torus_url, "",
                                          torus_query,
-                                         realm, m_p->proxy);
+                                         realm, m_p->proxy,
+                                         torus_addinfo);
         if (m_p->explain_xsp)
         {
             xmlDoc *rec_res =  xsltApplyStylesheet(m_p->explain_xsp, doc, 0);
@@ -1926,8 +2042,8 @@ yf::Zoom::BackendPtr yf::Zoom::Frontend::explain_search(mp::Package &package,
         if (!doc)
         {
             *error = YAZ_BIB1_UNSPECIFIED_ERROR;
-            *addinfo = odr_strdup(odr, "Torus server unavailable or "
-                                  "incorrectly configured");
+            if (torus_addinfo.length())
+                *addinfo = odr_strdup(odr, torus_addinfo.c_str());
         }
         else
         {
@@ -2266,48 +2382,47 @@ next_proxy:
     ZOOM_query q = ZOOM_query_create();
     ZOOM_query_sortby2(q, b->sptr->sortStrategy.c_str(), sortkeys.c_str());
 
+    Z_FacetList *fl = 0;
+
+    // Facets for request.. And later for reponse
+    if (!fl)
+        fl = yaz_oi_get_facetlist(&sr->otherInfo);
+    if (!fl)
+        fl = yaz_oi_get_facetlist(&sr->additionalSearchInfo);
+
     if (b->get_option("sru"))
     {
-        int status = 0;
         Z_RPNQuery *zquery;
         zquery = p_query_rpn(odr, wrbuf_cstr(pqf_wrbuf));
-        mp::wrbuf wrb;
+        mp::wrbuf wrb_cql;
+        mp::wrbuf wrb_addinfo;
 
         if (!strcmp(b->get_option("sru"), "solr"))
-        {
-            solr_transform_t cqlt = solr_transform_create();
-
-            status = solr_transform_rpn2solr_wrbuf(cqlt, wrb, zquery);
-
-            solr_transform_close(cqlt);
-        }
+            error = solr_transform_rpn2solr_stream_r(b->cqlt, wrb_addinfo,
+                                                     wrbuf_vp_puts, wrb_cql,
+                                                     zquery);
         else
+            error = cql_transform_rpn2cql_stream_r(b->cqlt, wrb_addinfo,
+                                                   wrbuf_vp_puts, wrb_cql,
+                                                   zquery);
+        if (error)
         {
-            status = cql_transform_rpn2cql_wrbuf(b->cqlt, wrb, zquery);
-        }
-        if (status == 0)
-        {
-            ZOOM_query_cql(q, wrbuf_cstr(wrb));
-            package.log("zoom", YLOG_LOG, "CQL: %s", wrbuf_cstr(wrb));
-            b->search(q, &hits, &error, &addinfo, odr);
-        }
-        ZOOM_query_destroy(q);
-
-        if (status)
-        {
-            error = YAZ_BIB1_MALFORMED_QUERY;
-            const char *addinfo = "can not convert from RPN to CQL/SOLR";
-            log_diagnostic(package, error, addinfo);
-            apdu_res = odr.create_searchResponse(apdu_req, error, addinfo);
+            log_diagnostic(package, error, wrb_addinfo.c_str_null());
+            apdu_res = odr.create_searchResponse(apdu_req, error,
+                                                 wrb_addinfo.c_str_null());
             package.response() = apdu_res;
             return;
         }
+        ZOOM_query_cql(q, wrb_cql.c_str());
+        package.log("zoom", YLOG_LOG, "search CQL: %s", wrb_cql.c_str());
+        b->search(q, &hits, &error, &addinfo, &fl, odr);
+        ZOOM_query_destroy(q);
     }
     else
     {
-        ZOOM_query_prefix(q, wrbuf_cstr(pqf_wrbuf));
-        package.log("zoom", YLOG_LOG, "search PQF: %s", wrbuf_cstr(pqf_wrbuf));
-        b->search(q, &hits, &error, &addinfo, odr);
+        ZOOM_query_prefix(q, pqf_wrbuf.c_str());
+        package.log("zoom", YLOG_LOG, "search PQF: %s", pqf_wrbuf.c_str());
+        b->search(q, &hits, &error, &addinfo, &fl, odr);
         ZOOM_query_destroy(q);
     }
 
@@ -2339,6 +2454,9 @@ next_proxy:
             odr_intdup(odr, number_of_records_returned);
     }
     apdu_res->u.searchResponse->resultCount = odr_intdup(odr, hits);
+    if (fl)
+        yaz_oi_set_facetlist(&apdu_res->u.searchResponse->additionalSearchInfo,
+                             odr, fl);
     package.response() = apdu_res;
 }
 
@@ -2503,8 +2621,13 @@ void yf::Zoom::Frontend::auth(mp::Package &package, Z_InitRequest *req,
         }
     }
 
-    std::string ip = package.origin().get_address();
-    yaz_log(YLOG_LOG, "IP=%s", ip.c_str());
+    Z_OtherInformation **oi = &req->otherInfo;
+    const char *ip =
+        yaz_oi_get_string_oid(oi, yaz_oid_userinfo_client_ip, 1, 0);
+    if (!ip)
+        ip = package.origin().get_address().c_str();
+
+    yaz_log(YLOG_LOG, "IP=%s", ip);
 
     std::string torus_query;
     int failure_code;
@@ -2517,27 +2640,24 @@ void yf::Zoom::Frontend::auth(mp::Package &package, Z_InitRequest *req,
     }
     else
     {
-        const char *ip_cstr = ip.c_str();
-        const char *cp = strchr(ip_cstr, ':');
-        if (cp)
-            ip_cstr = cp + 1;
-
         torus_query = "ip encloses/net.ipaddress \"";
-        torus_query += escape_cql_term(std::string(ip_cstr));
+        torus_query += escape_cql_term(std::string(ip));
         torus_query += "\"";
         failure_code = YAZ_BIB1_INIT_AC_BLOCKED_NETWORK_ADDRESS;
     }
 
     std::string dummy_db;
     std::string dummy_realm;
+    std::string torus_addinfo;
     xmlDoc *doc = mp::get_searchable(package, m_p->torus_auth_url, dummy_db,
-                                     torus_query, dummy_realm, m_p->proxy);
+                                     torus_query, dummy_realm, m_p->proxy,
+                                     torus_addinfo);
     if (!doc)
     {
         // something fundamental broken in lookup.
         *error = YAZ_BIB1_TEMPORARY_SYSTEM_ERROR;
-        *addinfo = odr_strdup(odr, "zoom: torus server unavailable or "
-                              "incorrectly configured.");
+        if (torus_addinfo.length())
+            *addinfo = odr_strdup(odr, torus_addinfo.c_str());
         return;
     }
     const xmlNode *ptr = xmlDocGetRootElement(doc);