First go at sort.
[metaproxy-moved-to-github.git] / src / filter_zoom.cpp
index b3ddddb..2a08f5f 100644 (file)
@@ -20,6 +20,7 @@ Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
 #include "filter_zoom.hpp"
 #include <yaz/zoom.h>
 #include <yaz/yaz-version.h>
+#include <yaz/tpath.h>
 #include <yaz/srw.h>
 #include <metaproxy/package.hpp>
 #include <metaproxy/util.hpp>
@@ -62,6 +63,7 @@ namespace metaproxy_1 {
             std::string record_encoding;
             std::string transform_xsl_fname;
             std::string urlRecipe;
+            std::string contentConnector;
             bool use_turbomarc;
             bool piggyback;
             CCL_bibset ccl_bibset;
@@ -77,6 +79,7 @@ namespace metaproxy_1 {
             std::string m_frontend_database;
             SearchablePtr sptr;
             xsltStylesheetPtr xsp;
+            std::string content_session_id;
         public:
             Backend(SearchablePtr sptr);
             ~Backend();
@@ -139,6 +142,8 @@ namespace metaproxy_1 {
             std::map<std::string,std::string> fieldmap;
             std::string xsldir;
             std::string file_path;
+            std::string content_proxy_server;
+            std::string content_tmp_file;
             CCL_bibset bibset;
             std::string element_transform;
             std::string element_raw;
@@ -386,6 +391,11 @@ yf::Zoom::SearchablePtr yf::Zoom::Impl::parse_torus_record(const xmlNode *ptr)
         {
             s->cfSubDb = mp::xml::get_text(ptr);
         }  
+        else if (!strcmp((const char *) ptr->name,
+                         "contentConnector"))
+        {
+            s->contentConnector = mp::xml::get_text(ptr);
+        }  
         else if (!strcmp((const char *) ptr->name, "udb"))
         {
             s->udb = mp::xml::get_text(ptr);
@@ -500,11 +510,10 @@ void yf::Zoom::Impl::configure_local_records(const xmlNode *ptr, bool test_only)
 void yf::Zoom::Impl::configure(const xmlNode *ptr, bool test_only,
                                const char *path)
 {
+    content_tmp_file = "/tmp/mp_content_proxy.";
     if (path && *path)
     {
         file_path = path;
-        if (path[strlen(path)-1] != '/')
-            file_path += "/";
     }
     for (ptr = ptr->children; ptr; ptr = ptr->next)
     {
@@ -554,6 +563,21 @@ void yf::Zoom::Impl::configure(const xmlNode *ptr, bool test_only,
             if (cql_field.length())
                 fieldmap[cql_field] = ccl_field;
         }
+        else if (!strcmp((const char *) ptr->name, "contentProxy"))
+        {
+            const struct _xmlAttr *attr;
+            for (attr = ptr->properties; attr; attr = attr->next)
+            {
+                if (!strcmp((const char *) attr->name, "server"))
+                    content_proxy_server = mp::xml::get_text(attr->children);
+                else if (!strcmp((const char *) attr->name, "tmp_file"))
+                    content_tmp_file = mp::xml::get_text(attr->children);
+                else
+                    throw mp::filter::FilterException(
+                        "Bad attribute " + std::string((const char *)
+                                                       attr->name));
+            }
+        }
         else
         {
             throw mp::filter::FilterException
@@ -571,7 +595,6 @@ yf::Zoom::BackendPtr yf::Zoom::Frontend::get_backend_from_databases(
     if (m_backend && m_backend->m_frontend_database == database)
         return m_backend;
 
-    const char *sru_proxy = 0;
     std::string db_args;
     std::string torus_db;
     size_t db_arg_pos = database.find(',');
@@ -627,18 +650,35 @@ yf::Zoom::BackendPtr yf::Zoom::Frontend::get_backend_from_databases(
     xsltStylesheetPtr xsp = 0;
     if (sptr->transform_xsl_fname.length())
     {
+        const char *path = 0;
+
+        if (m_p->xsldir.length())
+            path = m_p->xsldir.c_str();
+        else
+            path = m_p->file_path.c_str();
         std::string fname;
 
-        if (m_p->xsldir.length()) 
-            fname = m_p->xsldir + "/" + sptr->transform_xsl_fname;
+        char fullpath[1024];
+        char *cp = yaz_filepath_resolve(sptr->transform_xsl_fname.c_str(),
+                                        path, 0, fullpath);
+        if (cp)
+            fname.assign(cp);
         else
-            fname = m_p->file_path + sptr->transform_xsl_fname;
+        {
+            *error = YAZ_BIB1_TEMPORARY_SYSTEM_ERROR;
+            *addinfo = (char *)
+                odr_malloc(odr, 40 + sptr->transform_xsl_fname.length());
+            sprintf(*addinfo, "File could not be read: %s", 
+                    sptr->transform_xsl_fname.c_str());
+            BackendPtr b;
+            return b;
+        }
         xmlDoc *xsp_doc = xmlParseFile(fname.c_str());
         if (!xsp_doc)
         {
             *error = YAZ_BIB1_TEMPORARY_SYSTEM_ERROR;
-            *addinfo = (char *) odr_malloc(odr, 40 + strlen(fname.c_str()));
-            sprintf(*addinfo, "xmlParseFile failed. File %s", fname.c_str());
+            *addinfo = (char *) odr_malloc(odr, 40 + fname.length());
+            sprintf(*addinfo, "xmlParseFile failed. File: %s", fname.c_str());
             BackendPtr b;
             return b;
         }
@@ -659,39 +699,82 @@ yf::Zoom::BackendPtr yf::Zoom::Frontend::get_backend_from_databases(
 
     b->xsp = xsp;
     b->m_frontend_database = database;
-    std::string authentication = sptr->authentication;
-        
-    b->set_option("timeout", "40");
 
     if (sptr->query_encoding.length())
         b->set_option("rpnCharset", sptr->query_encoding.c_str());
 
+    b->set_option("timeout", "40");
+
+
+    std::string authentication = sptr->authentication;
+    std::string proxy = sptr->cfProxy;
+        
+    const char *param_user = 0;
+    const char *param_password = 0;
+    const char *param_proxy = 0;
+    if (db_args.length())
+    {
+        char **names;
+        char **values;
+        int i;
+        int no_parms = yaz_uri_to_array(db_args.c_str(),
+                                        odr, &names, &values);
+        for (i = 0; i < no_parms; i++)
+        {
+            const char *name = names[i];
+            const char *value = values[i];
+            if (!strcmp(name, "user"))
+                param_user = value;
+            else if (!strcmp(name, "password"))
+                param_password = value;
+            else if (!strcmp(name, "proxy"))
+                param_proxy = value;
+            else
+            {
+                BackendPtr notfound;
+                char *msg = (char*) odr_malloc(odr, strlen(name) + 30);
+                *error = YAZ_BIB1_TEMPORARY_SYSTEM_ERROR;
+                sprintf(msg, "Bad database argument: %s", name);
+                *addinfo = msg;
+                return notfound;
+            }
+        }
+        if (param_user && param_password)
+        {
+            authentication = std::string(param_user)
+                + "/" + std::string(param_password);
+        }
+        if (param_proxy)
+            proxy = param_proxy;
+    }
+
     if (sptr->cfAuth.length())
     {
         // A CF target
         b->set_option("user", sptr->cfAuth.c_str());
-        if (db_args.length() == 0)
+        if (!param_user && !param_password && authentication.length())
         {
-            if (authentication.length())
-            {
-                // no database (auth) args specified already.. and the
-                // Torus authentication has it.. Generate the args that CF
-                // understands..
-                size_t found = authentication.find('/');
-                if (found != std::string::npos)
-                {
-                    db_args += "user=" + mp::util::uri_encode(authentication.substr(0, found))
-                        + "&password=" + mp::util::uri_encode(authentication.substr(found+1));
-                }
-                else
-                    db_args += "user=" + mp::util::uri_encode(authentication);
-            }
-            if (sptr->cfProxy.length())
+            if (db_args.length())
+                db_args += "&";
+            // no database (auth) args specified already.. and the
+            // Torus authentication has it.. Generate the args that CF
+            // understands..
+            size_t found = authentication.find('/');
+            if (found != std::string::npos)
             {
-                if (db_args.length())
-                    db_args += "&";
-                db_args += "proxy=" + mp::util::uri_encode(sptr->cfProxy);
+                db_args += "user=" +
+                    mp::util::uri_encode(authentication.substr(0, found))
+                    + "&password=" +
+                    mp::util::uri_encode(authentication.substr(found+1));
             }
+            else
+                db_args += "user=" + mp::util::uri_encode(authentication);
+        }
+        if (!param_proxy && proxy.length())
+        {
+            if (db_args.length())
+                db_args += "&";
+            db_args += "proxy=" + mp::util::uri_encode(proxy);
         }
         if (sptr->cfSubDb.length())
         {
@@ -702,58 +785,51 @@ yf::Zoom::BackendPtr yf::Zoom::Frontend::get_backend_from_databases(
     }
     else
     {
-        // A non-CF target
-        if (db_args.length())
-        {
-            // user has specified backend authentication
-            const char *param_user = 0;
-            const char *param_password = 0;
-            char **names;
-            char **values;
-            int i;
-            int no_parms = yaz_uri_to_array(db_args.c_str(),
-                                            odr, &names, &values);
-            for (i = 0; i < no_parms; i++)
-            {
-                const char *name = names[i];
-                const char *value = values[i];
-                if (!strcmp(name, "user"))
-                    param_user = value;
-                else if (!strcmp(name, "password"))
-                    param_password = value;
-                else if (!strcmp(name, "proxy"))
-                    sru_proxy = value;
-                else
-                {
-                    BackendPtr notfound;
-                    char *msg = (char*) odr_malloc(odr, strlen(name) + 30);
-                    *error = YAZ_BIB1_TEMPORARY_SYSTEM_ERROR;
-                    sprintf(msg, "Bad database argument: %s", name);
-                    *addinfo = msg;
-                    return notfound;
-                }
-            }
-            if (param_user && param_password)
-            {
-                char *auth = (char*) odr_malloc(
-                    odr, strlen(param_user) + strlen(param_password) + 2);
-                strcpy(auth, param_user);
-                strcat(auth, "/");
-                strcat(auth, param_password);
-                b->set_option("user", auth);
-            }
-            db_args.clear(); // no arguments to be passed (non-CF)
-        }
-        else
+        db_args.clear(); // no arguments to be passed (non-CF)
+
+        if (authentication.length())
+            b->set_option("user", authentication.c_str());
+        if (proxy.length())
+            b->set_option("proxy", proxy.c_str());
+    }
+    if (b->sptr->contentConnector.length())
+    {
+        int fd;
+        
+        char *fname = (char *) xmalloc(m_p->content_tmp_file.length() + 8);
+        strcpy(fname, m_p->content_tmp_file.c_str());
+        strcat(fname, "XXXXXX");
+        fd = mkstemp(fname);
+        
+        if (fd == -1)
         {
-            // use authentication from Torus, if given
-            if (authentication.length())
-                b->set_option("user", authentication.c_str());
+            yaz_log(YLOG_WARN|YLOG_ERRNO, "create %s", fname);
+            *error = YAZ_BIB1_TEMPORARY_SYSTEM_ERROR;
+            *addinfo = (char *)  odr_malloc(odr, 40 + strlen(fname));
+            sprintf(*addinfo, "Could not create %s", fname);
+            xfree(fname);
+            BackendPtr backend_null;
+            return backend_null;
         }
+        b->content_session_id.assign(fname + (strlen(fname) - 6));
+        WRBUF w = wrbuf_alloc();
+        wrbuf_puts(w, "#content_proxy\n");
+        wrbuf_printf(w, "connector: %s\n", b->sptr->contentConnector.c_str());
+        if (authentication.length())
+            wrbuf_printf(w, "authentication: %s\n", authentication.c_str());
+        if (proxy.length())
+            wrbuf_printf(w, "proxy: %s\n", proxy.c_str());
+        if (sptr->cfAuth.length())
+            wrbuf_printf(w, "cfauth: %s\n", sptr->cfAuth.c_str());
+        if (sptr->cfProxy.length())
+            wrbuf_printf(w, "cfproxy: %s\n", sptr->cfProxy.c_str());
+
+        write(fd, wrbuf_buf(w), wrbuf_len(w));
+        close(fd);
+        yaz_log(YLOG_LOG, "file %s created\n", fname);
+        xfree(fname);
     }
-
-    if (sru_proxy)
-        b->set_option("proxy", sru_proxy);
+    
 
     std::string url;
     if (sptr->sru.length())
@@ -920,6 +996,18 @@ Z_Records *yf::Zoom::Frontend::get_records(Odr_int start,
                     xmlDoc *doc = xmlParseMemory(rec_buf, rec_len);
                     std::string res = 
                         mp::xml::url_recipe_handle(doc, b->sptr->urlRecipe);
+                    if (res.length() && b->content_session_id.length())
+                    {
+                        size_t off = res.find_first_of("://");
+                        if (off != std::string::npos)
+                        {
+                            char tmp[1024];
+                            sprintf(tmp, "%s.%s/",
+                                    b->content_session_id.c_str(),
+                                    m_p->content_proxy_server.c_str());
+                            res.insert(off + 3, tmp);
+                        }
+                    }
                     if (res.length())
                     {
                         xmlNode *ptr = xmlDocGetRootElement(doc);
@@ -1017,6 +1105,103 @@ struct cql_node *yf::Zoom::Impl::convert_cql_fields(struct cql_node *cn,
     return r;
 }
 
+static void sort_pqf_type_7(WRBUF pqf_wrbuf, const char *sru_sortkeys)
+{
+    /* sortkey layour: path,schema,ascending,caseSensitive,missingValue */
+    /* see cql_sortby_to_sortkeys of YAZ. */
+    char **sortspec;
+    int num_sortspec = 0;
+    int i;
+    NMEM nmem = nmem_create();
+    
+    if (sru_sortkeys)
+        nmem_strsplit_blank(nmem, sru_sortkeys, &sortspec, &num_sortspec);
+    if (num_sortspec > 0)
+    {
+        WRBUF w = wrbuf_alloc();
+        for (i = 0; i < num_sortspec; i++)
+        {
+            char **arg;
+            int num_arg;
+            int ascending = 1;
+            nmem_strsplitx(nmem, ",", sortspec[i], &arg, &num_arg, 0);
+            
+            if (num_arg > 2 && arg[2][0])
+                ascending = atoi(arg[2]);
+            
+            wrbuf_puts(w, "@or @attr 1=");
+            yaz_encode_pqf_term(w, arg[0], strlen(arg[0]));
+            wrbuf_printf(w, "@attr 7=%d %d ", ascending ? 1 : 2, i);
+        }
+        if (wrbuf_len(w))
+        {
+            wrbuf_puts(w, wrbuf_cstr(pqf_wrbuf));
+            wrbuf_rewind(pqf_wrbuf);
+            wrbuf_puts(pqf_wrbuf, wrbuf_cstr(w));
+        }
+        wrbuf_destroy(w);
+    }
+    nmem_destroy(nmem);
+}
+
+static void sort_via_cql(WRBUF cql_sortby, const char *sru_sortkeys)
+{
+    /* sortkey layour: path,schema,ascending,caseSensitive,missingValue */
+    /* see cql_sortby_to_sortkeys of YAZ. */
+    char **sortspec;
+    int num_sortspec = 0;
+    int i;
+    NMEM nmem = nmem_create();
+    
+    if (sru_sortkeys)
+        nmem_strsplit_blank(nmem, sru_sortkeys, &sortspec, &num_sortspec);
+    if (num_sortspec > 0)
+    {
+        WRBUF w = wrbuf_alloc();
+        for (i = 0; i < num_sortspec; i++)
+        {
+            char **arg;
+            int num_arg;
+            int ascending = 1;
+            int case_sensitive = 0;
+            const char *missing = 0;
+            nmem_strsplitx(nmem, ",", sortspec[i], &arg, &num_arg, 0);
+            
+            if (num_arg > 2 && arg[2][0])
+                ascending = atoi(arg[2]);
+            if (num_arg > 3 && arg[3][0])
+                case_sensitive = atoi(arg[3]);
+            if (num_arg > 4 && arg[4][0])
+                missing = arg[4];
+            if (i > 0)
+                wrbuf_puts(w, " ");
+            else
+                wrbuf_puts(w, " sortby ");
+            wrbuf_puts(w, arg[0]);  /* field */
+            wrbuf_puts(w, "/");
+            wrbuf_puts(w, ascending ? "ascending" : "descending");
+            if (case_sensitive)
+                wrbuf_puts(w, "/respectCase");
+            if (missing)
+            {
+                if (!strcmp(missing, "omit"))
+                    wrbuf_puts(w, "/missingOmit");
+                else if (!strcmp(missing, "abort"))
+                    wrbuf_puts(w, "/missingFail");
+                else if (!strcmp(missing, "lowValue"))
+                    wrbuf_puts(w, "/missingLow");
+                else if (!strcmp(missing, "highValue"))
+                    wrbuf_puts(w, "/missingHigh");
+            }
+        }
+        if (wrbuf_len(w))
+            wrbuf_puts(cql_sortby, wrbuf_cstr(w));
+        wrbuf_destroy(w);
+    }
+    nmem_destroy(nmem);
+}
+
+
 void yf::Zoom::Frontend::handle_search(mp::Package &package)
 {
     Z_GDU *gdu = package.request().get();
@@ -1050,6 +1235,7 @@ void yf::Zoom::Frontend::handle_search(mp::Package &package)
     Z_Query *query = sr->query;
     WRBUF ccl_wrbuf = 0;
     WRBUF pqf_wrbuf = 0;
+    std::string sru_sortkeys;
 
     if (query->which == Z_Query_type_1 || query->which == Z_Query_type_101)
     {
@@ -1104,6 +1290,13 @@ void yf::Zoom::Frontend::handle_search(mp::Package &package)
         {
             ccl_wrbuf = wrbuf_alloc();
             wrbuf_puts(ccl_wrbuf, ccl_buf);
+            
+            WRBUF sru_sortkeys_wrbuf = wrbuf_alloc();
+
+            cql_sortby_to_sortkeys(cn, wrbuf_vp_puts, sru_sortkeys_wrbuf);
+
+            sru_sortkeys.assign(wrbuf_cstr(sru_sortkeys_wrbuf));
+            wrbuf_destroy(sru_sortkeys_wrbuf);
         }
         cql_parser_destroy(cp);
         if (r)
@@ -1183,6 +1376,9 @@ void yf::Zoom::Frontend::handle_search(mp::Package &package)
             status = cql_transform_rpn2cql_wrbuf(cqlt, wrb, zquery);
             
             cql_transform_close(cqlt);
+
+            if (status == 0)
+                sort_via_cql(wrb, sru_sortkeys.c_str());
         }
         if (status == 0)
         {
@@ -1203,12 +1399,13 @@ void yf::Zoom::Frontend::handle_search(mp::Package &package)
     }
     else
     {
+        sort_pqf_type_7(pqf_wrbuf, sru_sortkeys.c_str());
+
         yaz_log(YLOG_LOG, "search PQF: %s", wrbuf_cstr(pqf_wrbuf));
         b->search_pqf(wrbuf_cstr(pqf_wrbuf), &hits, &error, &addinfo, odr);
         wrbuf_destroy(pqf_wrbuf);
     }
-    
-    
+
     const char *element_set_name = 0;
     Odr_int number_to_present = 0;
     if (!error)