virt_db: use query_encoding for search and scan
[metaproxy-moved-to-github.git] / src / filter_virt_db.cpp
index 243f8d7..c9474c8 100644 (file)
@@ -26,11 +26,13 @@ Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
 #include <boost/shared_ptr.hpp>
 
 #include <metaproxy/util.hpp>
+#include "torus.hpp"
 
 #include <yaz/zgdu.h>
 #include <yaz/otherinfo.h>
 #include <yaz/diagbib1.h>
 #include <yaz/match_glob.h>
+#include <yaz/query-charset.h>
 #include <yaz/log.h>
 
 #include <map>
@@ -52,17 +54,21 @@ namespace metaproxy_1 {
         };
         struct VirtualDB::Map {
             Map(std::string database, std::list<std::string> targets, std::string route);
+            Map(std::string database, std::string target, std::string route);
             Map();
             bool match(const std::string db) const;
             std::string m_dbpattern;
             std::list<std::string> m_targets;
             std::string m_route;
+        public:
+            std::string query_encoding;
         };
         struct VirtualDB::Backend {
             mp::Session m_backend_session;
             std::list<std::string> m_frontend_databases;
             std::list<std::string> m_targets;
             std::string m_route;
+            std::string query_encoding;
             bool m_named_result_sets;
             int m_number_of_sets;
         };
@@ -79,6 +85,7 @@ namespace metaproxy_1 {
             void search(Package &package, Z_APDU *apdu);
             void present(Package &package, Z_APDU *apdu);
             void scan(Package &package, Z_APDU *apdu);
+            int relay_apdu(Package &package, Z_APDU *apdu);
 
             void close(Package &package);
             typedef std::map<std::string,VirtualDB::Set>::iterator Sets_it;
@@ -107,6 +114,7 @@ namespace metaproxy_1 {
             
             FrontendPtr get_frontend(Package &package);
             void release_frontend(Package &package);
+            void refresh_torus();
         private:
             std::list<VirtualDB::Map>m_maps;
             typedef std::map<std::string,VirtualDB::Set>::iterator Sets_it;
@@ -114,6 +122,7 @@ namespace metaproxy_1 {
             boost::condition m_cond_session_ready;
             std::map<mp::Session, FrontendPtr> m_clients;
             bool pass_vhosts;
+            mp::Torus torus;
         };
     }
 }
@@ -174,12 +183,20 @@ yf::VirtualDB::BackendPtr yf::VirtualDB::Frontend::create_backend_from_databases
         // see if we have a route conflict.
         if (!first_route && b->m_route != map_it->m_route)
         {
-            // we have a conflict.. 
+            // we have a conflict.. routing must be same for all
             error_code =  YAZ_BIB1_COMBI_OF_SPECIFIED_DATABASES_UNSUPP;
             BackendPtr ptr;
             return ptr;
         }
         b->m_route = map_it->m_route;
+        if (!first_route && b->query_encoding != map_it->query_encoding)
+        {
+            // we have a conflict. query encoding must be same for all
+            error_code =  YAZ_BIB1_COMBI_OF_SPECIFIED_DATABASES_UNSUPP;
+            BackendPtr ptr;
+            return ptr;
+        }
+        b->query_encoding = map_it->query_encoding;
         first_route = false;
     }
     return b;
@@ -347,6 +364,20 @@ void yf::VirtualDB::Frontend::search(mp::Package &package, Z_APDU *apdu_req)
                                                 &req->databaseNames);
     }
 
+    if (b->query_encoding.length() && 
+        (req->query->which == Z_Query_type_1 ||
+         req->query->which == Z_Query_type_101))
+    {
+        yaz_iconv_t cd = yaz_iconv_open(b->query_encoding.c_str(), "UTF-8");
+        if (cd)
+        {
+            int r = yaz_query_charset_convert_rpnquery_check(
+                req->query->u.type_1, odr, cd);
+            yaz_iconv_close(cd);
+            if (r)
+                yaz_log(YLOG_LOG, "query could not be converted");
+        }
+    }
     *req->replaceIndicator = 1;
 
     search_package.request() = yazpp_1::GDU(apdu_req);
@@ -436,6 +467,72 @@ void yf::VirtualDB::Rep::release_frontend(mp::Package &package)
     }
 }
 
+void yf::VirtualDB::Rep::refresh_torus(void)
+{
+    xmlDoc *doc = torus.get_doc();
+    if (!doc)
+        return;
+
+    xmlNode *ptr1 = xmlDocGetRootElement(doc);
+    if (!ptr1)
+        return ;
+
+    for (ptr1 = ptr1->children; ptr1; ptr1 = ptr1->next)
+    {
+        if (ptr1->type != XML_ELEMENT_NODE)
+            continue;
+        if (!strcmp((const char *) ptr1->name, "record"))
+        {
+            xmlNode *ptr2 = ptr1;
+            for (ptr2 = ptr2->children; ptr2; ptr2 = ptr2->next)
+            {
+                if (ptr2->type != XML_ELEMENT_NODE)
+                    continue;
+                if (!strcmp((const char *) ptr2->name, "layer"))
+                {
+                    std::string database;
+                    std::string target;
+                    std::string route;
+                    std::string solr;
+                    std::string query_encoding;
+                    xmlNode *ptr3 = ptr2;
+                    for (ptr3 = ptr3->children; ptr3; ptr3 = ptr3->next)
+                    {
+                        if (ptr3->type != XML_ELEMENT_NODE)
+                            continue;
+                        if (!strcmp((const char *) ptr3->name, "id"))
+                        {
+                            database = mp::xml::get_text(ptr3);
+                        }
+                        else if (!strcmp((const char *) ptr3->name, "zurl"))
+                        {
+                            target = mp::xml::get_text(ptr3);
+                        }
+                        else if (!strcmp((const char *) ptr3->name, "sru"))
+                        {
+                            solr = mp::xml::get_text(ptr3);
+                        }
+                        else if (!strcmp((const char *) ptr3->name,
+                                         "queryEncoding"))
+                        {
+                            query_encoding = mp::xml::get_text(ptr3);
+                        }
+                    }
+                    if (solr.length() == 0 && 
+                        database.length() && target.length())
+                    {
+                        VirtualDB::Map vmap(
+                            mp::util::database_name_normalize(database),
+                            target, route);
+                        vmap.query_encoding = query_encoding;
+                        m_maps.push_back(vmap);
+                    }
+                }
+            }
+        }
+    }
+}
+
 yf::VirtualDB::Set::Set(BackendPtr b, std::string setname)
     :  m_backend(b), m_setname(setname)
 {
@@ -457,6 +554,14 @@ yf::VirtualDB::Map::Map(std::string database,
 {
 }
 
+yf::VirtualDB::Map::Map(std::string database, 
+                        std::string target, std::string route)
+    : m_dbpattern(database), m_route(route) 
+{
+    m_targets.push_back(target);
+}
+
+
 yf::VirtualDB::Map::Map()
 {
 }
@@ -603,6 +708,27 @@ void yf::VirtualDB::Frontend::present(mp::Package &package, Z_APDU *apdu_req)
     delete id;
 }
 
+int yf::VirtualDB::Frontend::relay_apdu(mp::Package &package, Z_APDU *apdu_req)
+{
+    int no = 0;
+    std::list<BackendPtr>::const_iterator map_it;
+    map_it = m_backend_list.begin();
+    for (; map_it != m_backend_list.end(); map_it++)
+    {
+        BackendPtr b = *map_it;
+        
+        Package relay_package(b->m_backend_session, package.origin());
+        relay_package.copy_filter(package);
+
+        relay_package.request() = yazpp_1::GDU(apdu_req);
+
+        relay_package.move(b->m_route);
+        package.response() = relay_package.response();
+        no++;
+    }
+    return no;
+}
+
 void yf::VirtualDB::Frontend::scan(mp::Package &package, Z_APDU *apdu_req)
 {
     Z_ScanRequest *req = apdu_req->u.scanRequest;
@@ -655,6 +781,17 @@ void yf::VirtualDB::Frontend::scan(mp::Package &package, Z_APDU *apdu_req)
                                                 &req->num_databaseNames,
                                                 &req->databaseNames);
     }
+
+    if (b->query_encoding.length())
+    {
+        yaz_iconv_t cd = yaz_iconv_open(b->query_encoding.c_str(), "UTF-8");
+        if (cd)
+        {
+            yaz_query_charset_convert_apt(req->termListAndStartPoint, odr, cd);
+            yaz_iconv_close(cd);
+        }
+    }
+
     scan_package.request() = yazpp_1::GDU(apdu_req);
     
     scan_package.move(b->m_route);
@@ -670,8 +807,8 @@ void yf::VirtualDB::Frontend::scan(mp::Package &package, Z_APDU *apdu_req)
 
 
 void yf::VirtualDB::add_map_db2targets(std::string db, 
-                                     std::list<std::string> targets,
-                                     std::string route)
+                                       std::list<std::string> targets,
+                                       std::string route)
 {
     m_p->m_maps.push_back(
         VirtualDB::Map(mp::util::database_name_normalize(db), targets, route));
@@ -679,13 +816,12 @@ void yf::VirtualDB::add_map_db2targets(std::string db,
 
 
 void yf::VirtualDB::add_map_db2target(std::string db, 
-                                    std::string target,
-                                    std::string route)
-{
-    std::list<std::string> targets;
-    targets.push_back(target);
+                                      std::string target,
+                                      std::string route)
 
-    add_map_db2targets(db, targets, route);
+{
+    m_p->m_maps.push_back(
+        VirtualDB::Map(mp::util::database_name_normalize(db), target, route));
 }
 
 void yf::VirtualDB::process(mp::Package &package) const
@@ -775,7 +911,15 @@ void yf::VirtualDB::process(mp::Package &package) const
         }
         else if (apdu->which == Z_APDU_close)
         {
-            package.session().close();
+            if (f->relay_apdu(package, apdu) == 0)
+            {
+                mp::odr odr;
+                
+                package.response() = odr.create_close(
+                    apdu, Z_Close_finished, "virt_db");
+                
+                package.session().close();
+            }
         }
         else
         {
@@ -791,7 +935,6 @@ void yf::VirtualDB::process(mp::Package &package) const
     m_p->release_frontend(package);
 }
 
-
 void mp::filter::VirtualDB::configure(const xmlNode * ptr, bool test_only)
 {
     for (ptr = ptr->children; ptr; ptr = ptr->next)
@@ -804,6 +947,7 @@ void mp::filter::VirtualDB::configure(const xmlNode * ptr, bool test_only)
         }
         else if (!strcmp((const char *) ptr->name, "virtual"))
         {
+            std::string query_encoding;
             std::string database;
             std::list<std::string> targets;
             xmlNode *v_node = ptr->children;
@@ -816,6 +960,8 @@ void mp::filter::VirtualDB::configure(const xmlNode * ptr, bool test_only)
                     database = mp::xml::get_text(v_node);
                 else if (mp::xml::is_element_mp(v_node, "target"))
                     targets.push_back(mp::xml::get_text(v_node));
+                else if (mp::xml::is_element_mp(v_node, "query-encoding"))
+                    targets.push_back(mp::xml::get_text(v_node));
                 else
                     throw mp::filter::FilterException
                         ("Bad element " 
@@ -824,7 +970,27 @@ void mp::filter::VirtualDB::configure(const xmlNode * ptr, bool test_only)
                             );
             }
             std::string route = mp::xml::get_route(ptr);
-            add_map_db2targets(database, targets, route);
+
+            VirtualDB::Map vmap(mp::util::database_name_normalize(database),
+                                targets, route);
+            vmap.query_encoding = query_encoding;
+            m_p->m_maps.push_back(vmap);
+        }
+        else if (!strcmp((const char *) ptr->name, "torus"))
+        {
+            std::string url;
+            const struct _xmlAttr *attr;
+            for (attr = ptr->properties; attr; attr = attr->next)
+            {
+                if (!strcmp((const char *) attr->name, "url"))
+                    url = mp::xml::get_text(attr->children);
+                else
+                    throw mp::filter::FilterException(
+                        "Bad attribute " + std::string((const char *)
+                                                       attr->name));
+            }
+            m_p->torus.read_searchables(url);
+            m_p->refresh_torus();
         }
         else
         {