zoom: Set option schema in SRU mode
[metaproxy-moved-to-github.git] / src / filter_zoom.cpp
index 9055067..ca1c47f 100644 (file)
@@ -17,6 +17,9 @@ Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
 */
 
 #include "config.hpp"
+
+#include <stdlib.h>
+#include <sys/types.h>
 #include "filter_zoom.hpp"
 #include <yaz/zoom.h>
 #include <yaz/yaz-version.h>
@@ -49,7 +52,8 @@ namespace yf = mp::filter;
 
 namespace metaproxy_1 {
     namespace filter {
-        struct Zoom::Searchable : boost::noncopyable {
+        class Zoom::Searchable : boost::noncopyable {
+          public:
             std::string authentication;
             std::string cfAuth;
             std::string cfProxy;
@@ -62,6 +66,7 @@ namespace metaproxy_1 {
             std::string element_set;
             std::string record_encoding;
             std::string transform_xsl_fname;
+            std::string transform_xsl_content;
             std::string urlRecipe;
             std::string contentConnector;
             bool use_turbomarc;
@@ -92,6 +97,7 @@ namespace metaproxy_1 {
             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, std::string value);
             const char *get_option(const char *name);
             void get_zoom_error(int *error, char **addinfo, ODR odr);
         };
@@ -109,6 +115,15 @@ namespace metaproxy_1 {
                                                   int *error,
                                                   char **addinfo,
                                                   ODR odr);
+
+
+            void prepare_elements(BackendPtr b,
+                                  Odr_oid *preferredRecordSyntax,
+                                  const char *element_set_name,
+                                  bool &enable_pz2_retrieval,
+                                  bool &enable_pz2_transform,
+                                  bool &assume_marc8_charset);
+
             Z_Records *get_records(Odr_int start,
                                    Odr_int number_to_present,
                                    int *error,
@@ -144,9 +159,11 @@ namespace metaproxy_1 {
             std::string file_path;
             std::string content_proxy_server;
             std::string content_tmp_file;
+            bool apdu_log;
             CCL_bibset bibset;
             std::string element_transform;
             std::string element_raw;
+            std::string proxy;
             std::map<std::string,SearchablePtr> s_map;
         };
     }
@@ -197,7 +214,8 @@ void yf::Zoom::Backend::get_zoom_error(int *error, char **addinfo,
 {
     const char *msg = 0;
     const char *zoom_addinfo = 0;
-    *error = ZOOM_connection_error(m_connection, &msg, &zoom_addinfo);
+    const char *dset = 0;
+    *error = ZOOM_connection_error_x(m_connection, &msg, &zoom_addinfo, &dset);
     if (*error)
     {
         if (*error >= ZOOM_ERROR_CONNECT)
@@ -217,6 +235,8 @@ void yf::Zoom::Backend::get_zoom_error(int *error, char **addinfo,
         }
         else
         {
+            if (dset && !strcmp(dset, "info:srw/diagnostic/1"))
+                *error = yaz_diag_srw_to_bib1(*error);
             *addinfo = (char *) odr_malloc(
                 odr, 20 + (zoom_addinfo ? strlen(zoom_addinfo) : 0));
             **addinfo = '\0';
@@ -280,6 +300,11 @@ void yf::Zoom::Backend::set_option(const char *name, const char *value)
         ZOOM_resultset_option_set(m_resultset, name, value);
 }
 
+void yf::Zoom::Backend::set_option(const char *name, std::string value)
+{
+    set_option(name, value.c_str());
+}
+
 const char *yf::Zoom::Backend::get_option(const char *name)
 {
     return ZOOM_connection_option_get(m_connection, name);
@@ -351,9 +376,12 @@ void yf::Zoom::Impl::release_frontend(mp::Package &package)
     }
 }
 
-yf::Zoom::Impl::Impl() : element_transform("pz2") , element_raw("raw")
+yf::Zoom::Impl::Impl() :
+    apdu_log(false), element_transform("pz2") , element_raw("raw")
 {
     bibset = ccl_qual_mk();
+
+    srand((unsigned int) time(0));
 }
 
 yf::Zoom::Impl::~Impl()
@@ -439,6 +467,11 @@ yf::Zoom::SearchablePtr yf::Zoom::Impl::parse_torus_record(const xmlNode *ptr)
             s->transform_xsl_fname = mp::xml::get_text(ptr);
         }
         else if (!strcmp((const char *) ptr->name,
+                         "literalTransform"))
+        {
+            s->transform_xsl_content = mp::xml::get_text(ptr);
+        }
+        else if (!strcmp((const char *) ptr->name,
                          "urlRecipe"))
         {
             s->urlRecipe = mp::xml::get_text(ptr);
@@ -510,7 +543,7 @@ 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.";
+    content_tmp_file = "/tmp/cf.XXXXXX.p";
     if (path && *path)
     {
         file_path = path;
@@ -532,6 +565,8 @@ void yf::Zoom::Impl::configure(const xmlNode *ptr, bool test_only,
                     element_transform = mp::xml::get_text(attr->children);
                 else if (!strcmp((const char *) attr->name, "element_raw"))
                     element_raw = mp::xml::get_text(attr->children);
+                else if (!strcmp((const char *) attr->name, "proxy"))
+                    proxy = mp::xml::get_text(attr->children);
                 else
                     throw mp::filter::FilterException(
                         "Bad attribute " + std::string((const char *)
@@ -578,6 +613,19 @@ void yf::Zoom::Impl::configure(const xmlNode *ptr, bool test_only,
                                                        attr->name));
             }
         }
+        else if (!strcmp((const char *) ptr->name, "log"))
+        { 
+            const struct _xmlAttr *attr;
+            for (attr = ptr->properties; attr; attr = attr->next)
+            {
+                if (!strcmp((const char *) attr->name, "apdu"))
+                    apdu_log = mp::xml::get_bool(attr->children, false);
+                else
+                    throw mp::filter::FilterException(
+                        "Bad attribute " + std::string((const char *)
+                                                       attr->name));
+            }
+        }
         else
         {
             throw mp::filter::FilterException
@@ -614,7 +662,7 @@ yf::Zoom::BackendPtr yf::Zoom::Frontend::get_backend_from_databases(
         sptr = it->second;
     else
     {
-        xmlDoc *doc = mp::get_searchable(m_p->torus_url, torus_db);
+        xmlDoc *doc = mp::get_searchable(m_p->torus_url, torus_db, m_p->proxy);
         if (!doc)
         {
             *error = YAZ_BIB1_DATABASE_DOES_NOT_EXIST;
@@ -631,8 +679,17 @@ yf::Zoom::BackendPtr yf::Zoom::Frontend::get_backend_from_databases(
                 if (ptr->type == XML_ELEMENT_NODE
                     && !strcmp((const char *) ptr->name, "record"))
                 {
+                    if (sptr)
+                    {
+                        *error = YAZ_BIB1_UNSPECIFIED_ERROR;
+                        *addinfo = (char*) odr_malloc(odr, 40 + database.length()),
+                        sprintf(*addinfo, "multiple records for udb=%s",
+                                 database.c_str());
+                        xmlFreeDoc(doc);
+                        BackendPtr b;
+                        return b;
+                    }
                     sptr = m_p->parse_torus_record(ptr);
-                    break;
                 }
             }
         }
@@ -648,7 +705,29 @@ yf::Zoom::BackendPtr yf::Zoom::Frontend::get_backend_from_databases(
     }
         
     xsltStylesheetPtr xsp = 0;
-    if (sptr->transform_xsl_fname.length())
+    if (sptr->transform_xsl_content.length())
+    {
+        xmlDoc *xsp_doc = xmlParseMemory(sptr->transform_xsl_content.c_str(),
+                                         sptr->transform_xsl_content.length());
+        if (!xsp_doc)
+        {
+            *error = YAZ_BIB1_TEMPORARY_SYSTEM_ERROR;
+            *addinfo = (char *) odr_malloc(odr, 40);
+            sprintf(*addinfo, "xmlParseMemory failed");
+            BackendPtr b;
+            return b;
+        }
+        xsp = xsltParseStylesheetDoc(xsp_doc);
+        if (!xsp)
+        {
+            *error = YAZ_BIB1_DATABASE_DOES_NOT_EXIST;
+            *addinfo = odr_strdup(odr, "xsltParseStylesheetDoc failed");
+            BackendPtr b;
+            xmlFreeDoc(xsp_doc);
+            return b;
+        }
+    }
+    else if (sptr->transform_xsl_fname.length())
     {
         const char *path = 0;
 
@@ -701,10 +780,17 @@ yf::Zoom::BackendPtr yf::Zoom::Frontend::get_backend_from_databases(
     b->m_frontend_database = database;
 
     if (sptr->query_encoding.length())
-        b->set_option("rpnCharset", sptr->query_encoding.c_str());
+        b->set_option("rpnCharset", sptr->query_encoding);
 
     b->set_option("timeout", "40");
+    
+    if (m_p->apdu_log) 
+        b->set_option("apdulog", "1");
 
+    if (sptr->piggyback)
+        b->set_option("count", "1"); /* some SRU servers INSIST on getting
+                                        maximumRecords > 0 */
+    b->set_option("piggyback", sptr->piggyback ? "1" : "0");
 
     std::string authentication = sptr->authentication;
     std::string proxy = sptr->cfProxy;
@@ -741,10 +827,11 @@ yf::Zoom::BackendPtr yf::Zoom::Frontend::get_backend_from_databases(
                 return notfound;
             }
         }
-        if (param_user && param_password)
+        if (param_user)
         {
-            authentication = std::string(param_user)
-                + "/" + std::string(param_password);
+            authentication = std::string(param_user);
+            if (param_password)
+                authentication += "/" + std::string(param_password);
         }
         if (param_proxy)
             proxy = param_proxy;
@@ -753,7 +840,7 @@ yf::Zoom::BackendPtr yf::Zoom::Frontend::get_backend_from_databases(
     if (sptr->cfAuth.length())
     {
         // A CF target
-        b->set_option("user", sptr->cfAuth.c_str());
+        b->set_option("user", sptr->cfAuth);
         if (!param_user && !param_password && authentication.length())
         {
             if (db_args.length())
@@ -789,21 +876,35 @@ yf::Zoom::BackendPtr yf::Zoom::Frontend::get_backend_from_databases(
     {
         db_args.clear(); // no arguments to be passed (non-CF)
 
-        if (authentication.length())
-            b->set_option("user", authentication.c_str());
+        size_t found = authentication.find('/');
+        
+        if (sptr->sru.length() && found != std::string::npos)
+        {
+            b->set_option("user", authentication.substr(0, found));
+            b->set_option("password", authentication.substr(found+1));
+        }
+        else
+            b->set_option("user", authentication);
+
         if (proxy.length())
-            b->set_option("proxy", proxy.c_str());
+            b->set_option("proxy", proxy);
     }
     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)
+        char *xx = strstr(fname, "XXXXXX");
+        if (!xx)
+        {
+            xx = fname + strlen(fname);
+            strcat(fname, "XXXXXX");
+        }
+        char tmp_char = xx[6];
+        sprintf(xx, "%06d", ((unsigned) rand()) % 1000000);
+        xx[6] = tmp_char;
+
+        FILE *file = fopen(fname, "w");
+        if (!file)
         {
             yaz_log(YLOG_WARN|YLOG_ERRNO, "create %s", fname);
             *error = YAZ_BIB1_TEMPORARY_SYSTEM_ERROR;
@@ -813,7 +914,7 @@ yf::Zoom::BackendPtr yf::Zoom::Frontend::get_backend_from_databases(
             BackendPtr backend_null;
             return backend_null;
         }
-        b->content_session_id.assign(fname + (strlen(fname) - 6));
+        b->content_session_id.assign(xx, 6);
         WRBUF w = wrbuf_alloc();
         wrbuf_puts(w, "#content_proxy\n");
         wrbuf_printf(w, "connector: %s\n", b->sptr->contentConnector.c_str());
@@ -821,23 +922,20 @@ yf::Zoom::BackendPtr yf::Zoom::Frontend::get_backend_from_databases(
             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);
+        fwrite(wrbuf_buf(w), 1, wrbuf_len(w), file);
+        fclose(file);
         yaz_log(YLOG_LOG, "file %s created\n", fname);
         xfree(fname);
     }
-    
 
     std::string url;
     if (sptr->sru.length())
     {
         url = "http://" + sptr->target;
-        b->set_option("sru", sptr->sru.c_str());
+        b->set_option("sru", sptr->sru);
     }
     else
     {
@@ -854,31 +952,14 @@ yf::Zoom::BackendPtr yf::Zoom::Frontend::get_backend_from_databases(
     return b;
 }
 
-Z_Records *yf::Zoom::Frontend::get_records(Odr_int start,
-                                           Odr_int number_to_present,
-                                           int *error,
-                                           char **addinfo,
-                                           Odr_int *number_of_records_returned,
-                                           ODR odr,
-                                           BackendPtr b,
-                                           Odr_oid *preferredRecordSyntax,
-                                           const char *element_set_name)
-{
-    *number_of_records_returned = 0;
-    Z_Records *records = 0;
-    bool enable_pz2_retrieval = false; // whether target profile is used
-    bool enable_pz2_transform = false; // whether XSLT is used as well
-    bool assume_marc8_charset = false;
-
-    if (start < 0 || number_to_present <= 0)
-        return records;
-    
-    if (number_to_present > 10000)
-        number_to_present = 10000;
-    
-    ZOOM_record *recs = (ZOOM_record *)
-        odr_malloc(odr, number_to_present * sizeof(*recs));
+void yf::Zoom::Frontend::prepare_elements(BackendPtr b,
+                                          Odr_oid *preferredRecordSyntax,
+                                          const char *element_set_name,
+                                          bool &enable_pz2_retrieval,
+                                          bool &enable_pz2_transform,
+                                          bool &assume_marc8_charset)
 
+{
     char oid_name_str[OID_STR_MAX];
     const char *syntax_name = 0;
     
@@ -899,9 +980,10 @@ Z_Records *yf::Zoom::Frontend::get_records(Odr_int start,
     
     if (enable_pz2_retrieval)
     {
-        if (b->sptr->request_syntax.length())
+        std::string configured_request_syntax = b->sptr->request_syntax;
+        if (configured_request_syntax.length())
         {
-            syntax_name = b->sptr->request_syntax.c_str();
+            syntax_name = configured_request_syntax.c_str();
             const Odr_oid *syntax_oid = 
                 yaz_string_to_oid(yaz_oid_std(), CLASS_RECSYN, syntax_name);
             if (!oid_oidcmp(syntax_oid, yaz_oid_recsyn_usmarc)
@@ -913,6 +995,9 @@ Z_Records *yf::Zoom::Frontend::get_records(Odr_int start,
         syntax_name =
             yaz_oid_to_string_buf(preferredRecordSyntax, 0, oid_name_str);
 
+    if (b->sptr->sru.length())
+        syntax_name = "XML";
+
     b->set_option("preferredRecordSyntax", syntax_name);
 
     if (enable_pz2_retrieval)
@@ -923,10 +1008,44 @@ Z_Records *yf::Zoom::Frontend::get_records(Odr_int start,
     }
 
     b->set_option("elementSetName", element_set_name);
+    if (b->sptr->sru.length() && element_set_name)
+        b->set_option("schema", element_set_name);
+}
+
+Z_Records *yf::Zoom::Frontend::get_records(Odr_int start,
+                                           Odr_int number_to_present,
+                                           int *error,
+                                           char **addinfo,
+                                           Odr_int *number_of_records_returned,
+                                           ODR odr,
+                                           BackendPtr b,
+                                           Odr_oid *preferredRecordSyntax,
+                                           const char *element_set_name)
+{
+    *number_of_records_returned = 0;
+    Z_Records *records = 0;
+    bool enable_pz2_retrieval = false; // whether target profile is used
+    bool enable_pz2_transform = false; // whether XSLT is used as well
+    bool assume_marc8_charset = false;
+
+    prepare_elements(b, preferredRecordSyntax,
+                     element_set_name,
+                     enable_pz2_retrieval,
+                     enable_pz2_transform,
+                     assume_marc8_charset);
+
+    if (start < 0 || number_to_present <=0)
+        return records;
+    
+    if (number_to_present > 10000)
+        number_to_present = 10000;
+
+    ZOOM_record *recs = (ZOOM_record *)
+        odr_malloc(odr, (size_t) number_to_present * sizeof(*recs));
 
     b->present(start, number_to_present, recs, error, addinfo, odr);
 
-    Odr_int i = 0;
+    int i = 0;
     if (!*error)
     {
         for (i = 0; i < number_to_present; i++)
@@ -976,20 +1095,52 @@ Z_Records *yf::Zoom::Frontend::get_records(Odr_int start,
                 xmlChar *xmlrec_buf = 0;
                 const char *rec_buf = ZOOM_record_get(recs[i], rec_type_str,
                                                       &rec_len);
+                if (!rec_buf && !npr)
+                {
+                    std::string addinfo("ZOOM_record_get failed for type ");
+
+                    addinfo += rec_type_str;
+                    npr = zget_surrogateDiagRec(
+                        odr, odr_database, 
+                        YAZ_BIB1_SYSTEM_ERROR_IN_PRESENTING_RECORDS,
+                        addinfo.c_str());
+                }
+
                 if (rec_buf && b->xsp && enable_pz2_transform)
                 {
                     xmlDoc *rec_doc = xmlParseMemory(rec_buf, rec_len);
-                    if (rec_doc)
+                    if (!rec_doc)
+                    {
+                        npr = zget_surrogateDiagRec(
+                            odr, odr_database, 
+                            YAZ_BIB1_SYSTEM_ERROR_IN_PRESENTING_RECORDS,
+                            "xml parse failed for record");
+                    }
+                    else
                     { 
-                        xmlDoc *rec_res;
-                        rec_res = xsltApplyStylesheet(b->xsp, rec_doc, 0);
+                        xmlDoc *rec_res = 
+                            xsltApplyStylesheet(b->xsp, rec_doc, 0);
 
                         if (rec_res)
+                        {
                             xsltSaveResultToString(&xmlrec_buf, &rec_len,
                                                    rec_res, b->xsp);
-                        rec_buf = (const char *) xmlrec_buf;
+                            rec_buf = (const char *) xmlrec_buf;
+
+                            xmlFreeDoc(rec_res);
+                        }
+                        if (!rec_buf)
+                        {
+                            std::string addinfo;
+
+                            addinfo = "xslt apply failed for "
+                                + b->sptr->transform_xsl_fname;
+                            npr = zget_surrogateDiagRec(
+                                odr, odr_database, 
+                                YAZ_BIB1_SYSTEM_ERROR_IN_PRESENTING_RECORDS,
+                                addinfo.c_str());
+                        }
                         xmlFreeDoc(rec_doc);
-                        xmlFreeDoc(rec_res);
                     }
                 }
 
@@ -1028,20 +1179,22 @@ Z_Records *yf::Zoom::Frontend::get_records(Odr_int start,
                     }
                     xmlFreeDoc(doc);
                 }
-                if (rec_buf)
+                if (!npr)
                 {
-                    npr = (Z_NamePlusRecord *) odr_malloc(odr, sizeof(*npr));
-                    npr->databaseName = odr_database;
-                    npr->which = Z_NamePlusRecord_databaseRecord;
-                    npr->u.databaseRecord =
-                        z_ext_record_xml(odr, rec_buf, rec_len);
-                }
-                else
-                {
-                    npr = zget_surrogateDiagRec(
-                        odr, odr_database, 
-                        YAZ_BIB1_SYSTEM_ERROR_IN_PRESENTING_RECORDS,
-                        rec_type_str);
+                    if (!rec_buf)
+                        npr = zget_surrogateDiagRec(
+                            odr, odr_database, 
+                            YAZ_BIB1_SYSTEM_ERROR_IN_PRESENTING_RECORDS,
+                            rec_type_str);
+                    else
+                    {
+                        npr = (Z_NamePlusRecord *)
+                            odr_malloc(odr, sizeof(*npr));
+                        npr->databaseName = odr_database;
+                        npr->which = Z_NamePlusRecord_databaseRecord;
+                        npr->u.databaseRecord =
+                            z_ext_record_xml(odr, rec_buf, rec_len);
+                    }
                 }
                 if (xmlrec_buf)
                     xmlFree(xmlrec_buf);
@@ -1073,7 +1226,7 @@ Z_Records *yf::Zoom::Frontend::get_records(Odr_int start,
     }
     return records;
 }
-    
+
 struct cql_node *yf::Zoom::Impl::convert_cql_fields(struct cql_node *cn,
                                                     ODR odr)
 {
@@ -1203,6 +1356,13 @@ static void sort_via_cql(WRBUF cql_sortby, const char *sru_sortkeys)
     nmem_destroy(nmem);
 }
 
+#if YAZ_VERSIONL < 0x40206
+static void wrbuf_vp_puts(const char *buf, void *client_data)
+{
+    WRBUF b = (WRBUF) client_data;
+    wrbuf_puts(b, buf);
+}
+#endif
 
 void yf::Zoom::Frontend::handle_search(mp::Package &package)
 {
@@ -1233,6 +1393,14 @@ void yf::Zoom::Frontend::handle_search(mp::Package &package)
 
     b->set_option("setname", "default");
 
+    bool enable_pz2_retrieval = false;
+    bool enable_pz2_transform = false;
+    bool assume_marc8_charset = false;
+    prepare_elements(b, sr->preferredRecordSyntax, 0 /*element_set_name */,
+                     enable_pz2_retrieval,
+                     enable_pz2_transform,
+                     assume_marc8_charset);
+
     Odr_int hits = 0;
     Z_Query *query = sr->query;
     WRBUF ccl_wrbuf = 0;