X-Git-Url: http://git.indexdata.com/?a=blobdiff_plain;f=src%2Ffilter_zoom.cpp;h=b48b58f42ddf80812c681d81954e0cad005bc095;hb=1cdd720a9a3ba872aa092c80b28413d8b2579eed;hp=586328528b873c19e565973a57a40cd54eab1f3d;hpb=0c6b0bda325d897e4c1e959747317fa8e76638f6;p=metaproxy-moved-to-github.git diff --git a/src/filter_zoom.cpp b/src/filter_zoom.cpp index 5863285..b48b58f 100644 --- a/src/filter_zoom.cpp +++ b/src/filter_zoom.cpp @@ -17,12 +17,13 @@ Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA */ #include "config.hpp" + +#include +#include #include "filter_zoom.hpp" -#include -#include -#include #include #include +#include #include "torus.hpp" #include @@ -30,6 +31,10 @@ Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA #include #include + +#include +#include +#include #include #include #include @@ -41,28 +46,38 @@ Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA #include #include #include +#include +#include +#include namespace mp = metaproxy_1; 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; - std::string cfSubDb; + std::string cfSubDB; std::string udb; std::string target; std::string query_encoding; std::string sru; + std::string sru_version; std::string request_syntax; std::string element_set; std::string record_encoding; std::string transform_xsl_fname; + std::string transform_xsl_content; + std::string urlRecipe; + std::string contentConnector; + std::string sortStrategy; bool use_turbomarc; bool piggyback; CCL_bibset ccl_bibset; + std::map sortmap; Searchable(CCL_bibset base); ~Searchable(); }; @@ -70,24 +85,26 @@ namespace metaproxy_1 { friend class Impl; friend class Frontend; std::string zurl; + WRBUF m_apdu_wrbuf; ZOOM_connection m_connection; ZOOM_resultset m_resultset; std::string m_frontend_database; SearchablePtr sptr; xsltStylesheetPtr xsp; + std::string content_session_id; public: Backend(SearchablePtr sptr); ~Backend(); - void connect(std::string zurl, int *error, const char **addinfo); - void search_pqf(const char *pqf, Odr_int *hits, - int *error, const char **addinfo); - void search_cql(const char *cql, Odr_int *hits, - int *error, const char **addinfo); + 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); void present(Odr_int start, Odr_int number, ZOOM_record *recs, - int *error, const char **addinfo); + 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, const char **addinfo); + void get_zoom_error(int *error, char **addinfo, ODR odr); }; class Zoom::Frontend : boost::noncopyable { friend class Impl; @@ -99,13 +116,24 @@ namespace metaproxy_1 { void handle_package(mp::Package &package); void handle_search(mp::Package &package); void handle_present(mp::Package &package); - BackendPtr get_backend_from_databases(std::string &database, + BackendPtr get_backend_from_databases(mp::Package &package, + std::string &database, int *error, - const char **addinfo); + 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, - const char **addinfo, + char **addinfo, Odr_int *number_of_records_returned, ODR odr, BackendPtr b, Odr_oid *preferredRecordSyntax, @@ -120,7 +148,8 @@ namespace metaproxy_1 { Impl(); ~Impl(); void process(metaproxy_1::Package & package); - void configure(const xmlNode * ptr, bool test_only); + void configure(const xmlNode * ptr, bool test_only, + const char *path); private: void configure_local_records(const xmlNode * ptr, bool test_only); FrontendPtr get_frontend(mp::Package &package); @@ -131,9 +160,17 @@ namespace metaproxy_1 { boost::mutex m_mutex; boost::condition m_cond_session_ready; std::string torus_url; + std::string default_realm; std::map fieldmap; std::string xsldir; + 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 s_map; }; } @@ -149,9 +186,10 @@ yf::Zoom::~Zoom() { // must have a destructor because of boost::scoped_ptr } -void yf::Zoom::configure(const xmlNode *xmlnode, bool test_only) +void yf::Zoom::configure(const xmlNode *xmlnode, bool test_only, + const char *path) { - m_p->configure(xmlnode, test_only); + m_p->configure(xmlnode, test_only, path); } void yf::Zoom::process(mp::Package &package) const @@ -164,7 +202,9 @@ void yf::Zoom::process(mp::Package &package) const yf::Zoom::Backend::Backend(SearchablePtr ptr) : sptr(ptr) { + m_apdu_wrbuf = wrbuf_alloc(); m_connection = ZOOM_connection_create(0); + ZOOM_connection_save_apdu_wrbuf(m_connection, m_apdu_wrbuf); m_resultset = 0; xsp = 0; } @@ -178,50 +218,60 @@ yf::Zoom::Backend::~Backend() } -void yf::Zoom::Backend::get_zoom_error(int *error, const char **addinfo) +void yf::Zoom::Backend::get_zoom_error(int *error, char **addinfo, + ODR odr) { const char *msg = 0; - *error = ZOOM_connection_error(m_connection, &msg, addinfo); + const char *zoom_addinfo = 0; + const char *dset = 0; + *error = ZOOM_connection_error_x(m_connection, &msg, &zoom_addinfo, &dset); if (*error) { if (*error >= ZOOM_ERROR_CONNECT) { // turn ZOOM diagnostic into a Bib-1 2: with addinfo=zoom err msg *error = YAZ_BIB1_TEMPORARY_SYSTEM_ERROR; - if (addinfo) - *addinfo = msg; + *addinfo = (char *) odr_malloc( + odr, 20 + strlen(msg) + + (zoom_addinfo ? strlen(zoom_addinfo) : 0)); + strcpy(*addinfo, msg); + if (zoom_addinfo) + { + strcat(*addinfo, ": "); + strcat(*addinfo, zoom_addinfo); + strcat(*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'; + if (zoom_addinfo && *zoom_addinfo) + { + strcpy(*addinfo, zoom_addinfo); + strcat(*addinfo, " "); + } + strcat(*addinfo, "(backend)"); } } } void yf::Zoom::Backend::connect(std::string zurl, - int *error, const char **addinfo) + int *error, char **addinfo, + ODR odr) { ZOOM_connection_connect(m_connection, zurl.c_str(), 0); - get_zoom_error(error, addinfo); + get_zoom_error(error, addinfo, odr); } -void yf::Zoom::Backend::search_pqf(const char *pqf, Odr_int *hits, - int *error, const char **addinfo) +void yf::Zoom::Backend::search(ZOOM_query q, Odr_int *hits, + int *error, char **addinfo, ODR odr) { - m_resultset = ZOOM_connection_search_pqf(m_connection, pqf); - get_zoom_error(error, addinfo); - if (*error == 0) - *hits = ZOOM_resultset_size(m_resultset); - else - *hits = 0; -} - -void yf::Zoom::Backend::search_cql(const char *cql, Odr_int *hits, - int *error, const char **addinfo) -{ - ZOOM_query q = ZOOM_query_create(); - - ZOOM_query_cql(q, cql); - m_resultset = ZOOM_connection_search(m_connection, q); - ZOOM_query_destroy(q); - get_zoom_error(error, addinfo); + get_zoom_error(error, addinfo, odr); if (*error == 0) *hits = ZOOM_resultset_size(m_resultset); else @@ -230,10 +280,10 @@ void yf::Zoom::Backend::search_cql(const char *cql, Odr_int *hits, void yf::Zoom::Backend::present(Odr_int start, Odr_int number, ZOOM_record *recs, - int *error, const char **addinfo) + int *error, char **addinfo, ODR odr) { ZOOM_resultset_records(m_resultset, recs, start, number); - get_zoom_error(error, addinfo); + get_zoom_error(error, addinfo, odr); } void yf::Zoom::Backend::set_option(const char *name, const char *value) @@ -243,6 +293,11 @@ 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); @@ -252,6 +307,7 @@ yf::Zoom::Searchable::Searchable(CCL_bibset base) { piggyback = true; use_turbomarc = true; + sortStrategy = "embed"; ccl_bibset = ccl_qual_dup(base); } @@ -314,9 +370,12 @@ void yf::Zoom::Impl::release_frontend(mp::Package &package) } } -yf::Zoom::Impl::Impl() +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() @@ -350,9 +409,14 @@ yf::Zoom::SearchablePtr yf::Zoom::Impl::parse_torus_record(const xmlNode *ptr) s->cfProxy = mp::xml::get_text(ptr); } else if (!strcmp((const char *) ptr->name, - "cfSubDb")) + "cfSubDB")) + { + s->cfSubDB = mp::xml::get_text(ptr); + } + else if (!strcmp((const char *) ptr->name, + "contentConnector")) { - s->cfSubDb = mp::xml::get_text(ptr); + s->contentConnector = mp::xml::get_text(ptr); } else if (!strcmp((const char *) ptr->name, "udb")) { @@ -366,6 +430,10 @@ yf::Zoom::SearchablePtr yf::Zoom::Impl::parse_torus_record(const xmlNode *ptr) { s->sru = mp::xml::get_text(ptr); } + else if (!strcmp((const char *) ptr->name, "SRUVersion")) + { + s->sru_version = mp::xml::get_text(ptr); + } else if (!strcmp((const char *) ptr->name, "queryEncoding")) { @@ -397,6 +465,16 @@ 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); + } + else if (!strcmp((const char *) ptr->name, "useTurboMarc")) { ; // useTurboMarc is ignored @@ -408,6 +486,17 @@ yf::Zoom::SearchablePtr yf::Zoom::Impl::parse_torus_record(const xmlNode *ptr) ccl_qual_fitem(s->ccl_bibset, value.c_str(), (const char *) ptr->name + 7); } + else if (!strncmp((const char *) ptr->name, + "sortmap_", 8)) + { + std::string value = mp::xml::get_text(ptr); + s->sortmap[(const char *) ptr->name + 8] = value; + } + else if (!strcmp((const char *) ptr->name, + "sortStrategy")) + { + s->sortStrategy = mp::xml::get_text(ptr); + } } return s; } @@ -460,8 +549,14 @@ void yf::Zoom::Impl::configure_local_records(const xmlNode *ptr, bool test_only) } } -void yf::Zoom::Impl::configure(const xmlNode *ptr, bool test_only) +void yf::Zoom::Impl::configure(const xmlNode *ptr, bool test_only, + const char *path) { + content_tmp_file = "/tmp/cf.XXXXXX.p"; + if (path && *path) + { + file_path = path; + } for (ptr = ptr->children; ptr; ptr = ptr->next) { if (ptr->type != XML_ELEMENT_NODE) @@ -473,8 +568,16 @@ void yf::Zoom::Impl::configure(const xmlNode *ptr, bool test_only) { if (!strcmp((const char *) attr->name, "url")) torus_url = mp::xml::get_text(attr->children); + else if (!strcmp((const char *) attr->name, "realm")) + default_realm = mp::xml::get_text(attr->children); else if (!strcmp((const char *) attr->name, "xsldir")) xsldir = mp::xml::get_text(attr->children); + else if (!strcmp((const char *) attr->name, "element_transform")) + 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 *) @@ -506,6 +609,34 @@ 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 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 @@ -517,37 +648,99 @@ void yf::Zoom::Impl::configure(const xmlNode *ptr, bool test_only) } yf::Zoom::BackendPtr yf::Zoom::Frontend::get_backend_from_databases( - std::string &database, int *error, const char **addinfo) + mp::Package &package, + std::string &database, int *error, char **addinfo, ODR odr) { std::list::const_iterator map_it; if (m_backend && m_backend->m_frontend_database == database) return m_backend; - std::string db_args; - std::string cf_parm; + std::string input_args; std::string torus_db; size_t db_arg_pos = database.find(','); if (db_arg_pos != std::string::npos) { torus_db = database.substr(0, db_arg_pos); - db_args = database.substr(db_arg_pos+1); + input_args = database.substr(db_arg_pos + 1); } else torus_db = database; - + + std::string authentication; + std::string proxy; + std::string realm = m_p->default_realm; + + const char *param_user = 0; + const char *param_password = 0; + int no_parms = 0; + + char **names; + char **values; + int no_out_args = 0; + if (input_args.length()) + no_parms = yaz_uri_to_array(input_args.c_str(), + odr, &names, &values); + // adding 10 because we'll be adding other URL args + const char **out_names = (const char **) + odr_malloc(odr, (10 + no_parms) * sizeof(*out_names)); + const char **out_values = (const char **) + odr_malloc(odr, (10 + no_parms) * sizeof(*out_values)); + + int i; + for (i = 0; i < no_parms; i++) + { + const char *name = names[i]; + const char *value = values[i]; + assert(name); + assert(value); + if (!strcmp(name, "user")) + param_user = value; + else if (!strcmp(name, "password")) + param_password = value; + else if (!strcmp(name, "proxy")) + proxy = value; + else if (!strcmp(name, "cproxysession")) + { + out_names[no_out_args] = name; + out_values[no_out_args++] = value; + } + else if (!strcmp(name, "realm")) + realm = value; + else if (name[0] == 'x' && name[1] == '-') + { + out_names[no_out_args] = name; + out_values[no_out_args++] = 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) + { + authentication = std::string(param_user); + if (param_password) + authentication += "/" + std::string(param_password); + } SearchablePtr sptr; std::map::iterator it; it = m_p->s_map.find(torus_db); if (it != m_p->s_map.end()) sptr = it->second; - else + else if (m_p->torus_url.length() > 0) { - xmlDoc *doc = mp::get_searchable(m_p->torus_url, torus_db); + xmlDoc *doc = mp::get_searchable(m_p->torus_url, torus_db, realm, + m_p->proxy); if (!doc) { *error = YAZ_BIB1_DATABASE_DOES_NOT_EXIST; - *addinfo = database.c_str(); + *addinfo = odr_strdup(odr, database.c_str()); BackendPtr b; return b; } @@ -560,8 +753,17 @@ 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; } } } @@ -571,25 +773,65 @@ yf::Zoom::BackendPtr yf::Zoom::Frontend::get_backend_from_databases( if (!sptr) { *error = YAZ_BIB1_DATABASE_DOES_NOT_EXIST; - *addinfo = database.c_str(); + *addinfo = odr_strdup(odr, database.c_str()); BackendPtr b; return b; } 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; + + 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 = 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 = "xmlParseFile failed"; + *addinfo = (char *) odr_malloc(odr, 40 + fname.length()); + sprintf(*addinfo, "xmlParseFile failed. File: %s", fname.c_str()); BackendPtr b; return b; } @@ -597,7 +839,7 @@ yf::Zoom::BackendPtr yf::Zoom::Frontend::get_backend_from_databases( if (!xsp) { *error = YAZ_BIB1_DATABASE_DOES_NOT_EXIST; - *addinfo = "xsltParseStylesheetDoc failed"; + *addinfo = odr_strdup(odr, "xsltParseStylesheetDoc failed"); BackendPtr b; xmlFreeDoc(xsp_doc); return b; @@ -610,60 +852,142 @@ 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("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"); + + if (authentication.length() == 0) + authentication = sptr->authentication; + + if (proxy.length() == 0) + proxy = sptr->cfProxy; + if (sptr->cfAuth.length()) { - b->set_option("user", sptr->cfAuth.c_str()); + // A CF target + b->set_option("user", sptr->cfAuth); if (authentication.length()) { size_t found = authentication.find('/'); if (found != std::string::npos) { - cf_parm += "user=" + mp::util::uri_encode(authentication.substr(0, found)) - + "&password=" + mp::util::uri_encode(authentication.substr(found+1)); + out_names[no_out_args] = "user"; + out_values[no_out_args++] = + odr_strdup(odr, authentication.substr(0, found).c_str()); + + out_names[no_out_args] = "password"; + out_values[no_out_args++] = + odr_strdup(odr, authentication.substr(found+1).c_str()); } else - cf_parm += "user=" + mp::util::uri_encode(authentication); + { + out_names[no_out_args] = "user"; + out_values[no_out_args++] = + odr_strdup(odr, authentication.c_str()); + } + } + if (proxy.length()) + { + out_names[no_out_args] = "proxy"; + out_values[no_out_args++] = odr_strdup(odr, proxy.c_str()); + } + if (sptr->cfSubDB.length()) + { + out_names[no_out_args] = "subdatabase"; + out_values[no_out_args++] = odr_strdup(odr, sptr->cfSubDB.c_str()); } } - else if (authentication.length()) - b->set_option("user", authentication.c_str()); - - if (sptr->cfProxy.length()) + else { - if (cf_parm.length()) - cf_parm += "&"; - cf_parm += "proxy=" + mp::util::uri_encode(sptr->cfProxy); + 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); } - if (sptr->cfSubDb.length()) + if (b->sptr->contentConnector.length()) { - if (cf_parm.length()) - cf_parm += "&"; - cf_parm += "subdatabase=" + mp::util::uri_encode(sptr->cfSubDb); + char *fname = (char *) xmalloc(m_p->content_tmp_file.length() + 8); + strcpy(fname, m_p->content_tmp_file.c_str()); + 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) + { + package.log("zoom", 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(xx, 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, "auth: %s\n", authentication.c_str()); + if (proxy.length()) + wrbuf_printf(w, "proxy: %s\n", proxy.c_str()); + if (sptr->cfProxy.length()) + wrbuf_printf(w, "cfproxy: %s\n", sptr->cfProxy.c_str()); + + fwrite(wrbuf_buf(w), 1, wrbuf_len(w), file); + fclose(file); + package.log("zoom", YLOG_LOG, "file %s created", 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); + + if (sptr->sru_version.length()) + b->set_option("sru_version", sptr->sru_version); } else { url = sptr->target; } - if (db_args.length()) - url += "," + db_args; - else if (cf_parm.length()) - url += "," + cf_parm; - yaz_log(YLOG_LOG, "url=%s", url.c_str()); - b->connect(url, error, addinfo); + if (no_out_args) + { + char *x_args = 0; + out_names[no_out_args] = 0; // terminate list + + yaz_array_to_uri(&x_args, odr, (char **) out_names, + (char **) out_values); + url += "," + std::string(x_args); + } + package.log("zoom", YLOG_LOG, "url: %s", url.c_str()); + b->connect(url, error, addinfo, odr); if (*error == 0) { m_backend = b; @@ -671,63 +995,100 @@ 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, - const 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_transform = 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; - - if (preferredRecordSyntax) + + if (preferredRecordSyntax && + !oid_oidcmp(preferredRecordSyntax, yaz_oid_recsyn_xml) + && element_set_name) { - if (!oid_oidcmp(preferredRecordSyntax, yaz_oid_recsyn_xml) - && element_set_name && - !strcmp(element_set_name, "pz2")) + if (!strcmp(element_set_name, m_p->element_transform.c_str())) { - if (b->sptr->request_syntax.length()) - syntax_name = b->sptr->request_syntax.c_str(); + enable_pz2_retrieval = true; enable_pz2_transform = true; } - else + else if (!strcmp(element_set_name, m_p->element_raw.c_str())) { - syntax_name = - yaz_oid_to_string_buf(preferredRecordSyntax, 0, oid_name_str); + enable_pz2_retrieval = true; } } + + if (enable_pz2_retrieval) + { + std::string configured_request_syntax = b->sptr->request_syntax; + if (configured_request_syntax.length()) + { + 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) + || !oid_oidcmp(syntax_oid, yaz_oid_recsyn_opac)) + assume_marc8_charset = true; + } + } + else if (preferredRecordSyntax) + 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_transform) + if (enable_pz2_retrieval) { - element_set_name = "F"; + element_set_name = 0; if (b->sptr->element_set.length()) element_set_name = b->sptr->element_set.c_str(); } b->set_option("elementSetName", element_set_name); + if (b->sptr->sru.length() && element_set_name) + b->set_option("schema", element_set_name); +} - b->present(start, number_to_present, recs, error, addinfo); +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; - Odr_int i = 0; + 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); + + int i = 0; if (!*error) { for (i = 0; i < number_to_present; i++) @@ -756,53 +1117,132 @@ Z_Records *yf::Zoom::Frontend::get_records(Odr_int start, npr = zget_surrogateDiagRec(odr, odr_database, sur_error, addinfo); } - else if (enable_pz2_transform) + else if (enable_pz2_retrieval) { char rec_type_str[100]; + const char *record_encoding = 0; + + if (b->sptr->record_encoding.length()) + record_encoding = b->sptr->record_encoding.c_str(); + else if (assume_marc8_charset) + record_encoding = "marc8"; - strcpy(rec_type_str, b->sptr->use_turbomarc ? - "txml" : "xml"); - // prevent buffer overflow ... - if (b->sptr->record_encoding.length() > 0 && - b->sptr->record_encoding.length() < - (sizeof(rec_type_str)-20)) + strcpy(rec_type_str, b->sptr->use_turbomarc ? "txml" : "xml"); + if (record_encoding) { strcat(rec_type_str, "; charset="); - strcat(rec_type_str, b->sptr->record_encoding.c_str()); + strcat(rec_type_str, record_encoding); } int rec_len; + xmlChar *xmlrec_buf = 0; const char *rec_buf = ZOOM_record_get(recs[i], rec_type_str, &rec_len); - if (rec_buf && b->xsp) + 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((xmlChar **) &rec_buf, &rec_len, + { + xsltSaveResultToString(&xmlrec_buf, &rec_len, rec_res, b->xsp); + 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); } } if (rec_buf) { - 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); + 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); + while (ptr && ptr->type != XML_ELEMENT_NODE) + ptr = ptr->next; + xmlNode *c = + xmlNewChild(ptr, 0, BAD_CAST "metadata", 0); + xmlNewProp(c, BAD_CAST "type", BAD_CAST + "generated-url"); + xmlNode * t = xmlNewText(BAD_CAST res.c_str()); + xmlAddChild(c, t); + + if (xmlrec_buf) + xmlFree(xmlrec_buf); + + xmlDocDumpMemory(doc, &xmlrec_buf, &rec_len); + rec_buf = (const char *) xmlrec_buf; + } + xmlFreeDoc(doc); } - else + if (!npr) { - 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); } else { @@ -831,7 +1271,7 @@ 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) { @@ -881,24 +1321,33 @@ void yf::Zoom::Frontend::handle_search(mp::Package &package) } int error = 0; - const char *addinfo = 0; + char *addinfo = 0; std::string db(sr->databaseNames[0]); - BackendPtr b = get_backend_from_databases(db, &error, &addinfo); + BackendPtr b = get_backend_from_databases(package, db, &error, + &addinfo, odr); if (error) { apdu_res = - odr.create_searchResponse( - apdu_req, error, addinfo); + odr.create_searchResponse(apdu_req, error, addinfo); package.response() = apdu_res; return; } 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; WRBUF pqf_wrbuf = 0; + std::string sortkeys; if (query->which == Z_Query_type_1 || query->which == Z_Query_type_101) { @@ -920,9 +1369,11 @@ void yf::Zoom::Frontend::handle_search(mp::Package &package) const char *cql = query->u.type_104->u.cql; CQL_parser cp = cql_parser_create(); int r = cql_parser_string(cp, cql); + package.log("zoom", YLOG_LOG, "CQL: %s", cql); if (r) { cql_parser_destroy(cp); + package.log("zoom", YLOG_WARN, "CQL syntax error"); apdu_res = odr.create_searchResponse(apdu_req, YAZ_BIB1_MALFORMED_QUERY, @@ -953,6 +1404,48 @@ 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); + WRBUF sort_spec_wrbuf = wrbuf_alloc(); + yaz_srw_sortkeys_to_sort_spec(wrbuf_cstr(sru_sortkeys_wrbuf), + sort_spec_wrbuf); + wrbuf_destroy(sru_sortkeys_wrbuf); + + yaz_tok_cfg_t tc = yaz_tok_cfg_create(); + yaz_tok_parse_t tp = + yaz_tok_parse_buf(tc, wrbuf_cstr(sort_spec_wrbuf)); + yaz_tok_cfg_destroy(tc); + + /* go through sortspec and map fields */ + int token = yaz_tok_move(tp); + while (token != YAZ_TOK_EOF) + { + if (token == YAZ_TOK_STRING) + { + const char *field = yaz_tok_parse_string(tp); + std::map::iterator it; + it = b->sptr->sortmap.find(field); + if (it != b->sptr->sortmap.end()) + sortkeys += it->second; + else + sortkeys += field; + } + sortkeys += " "; + token = yaz_tok_move(tp); + if (token == YAZ_TOK_STRING) + { + sortkeys += yaz_tok_parse_string(tp); + } + if (token != YAZ_TOK_EOF) + { + sortkeys += " "; + token = yaz_tok_move(tp); + } + } + yaz_tok_parse_destroy(tp); + wrbuf_destroy(sort_spec_wrbuf); } cql_parser_destroy(cp); if (r) @@ -979,7 +1472,7 @@ void yf::Zoom::Frontend::handle_search(mp::Package &package) assert(pqf_wrbuf == 0); int cerror, cpos; struct ccl_rpn_node *cn; - yaz_log(YLOG_LOG, "CCL: %s", wrbuf_cstr(ccl_wrbuf)); + package.log("zoom", YLOG_LOG, "CCL: %s", wrbuf_cstr(ccl_wrbuf)); cn = ccl_find_str(b->sptr->ccl_bibset, wrbuf_cstr(ccl_wrbuf), &cerror, &cpos); wrbuf_destroy(ccl_wrbuf); @@ -1006,10 +1499,15 @@ void yf::Zoom::Frontend::handle_search(mp::Package &package) } pqf_wrbuf = wrbuf_alloc(); ccl_pquery(pqf_wrbuf, cn); + package.log("zoom", YLOG_LOG, "RPN: %s", wrbuf_cstr(pqf_wrbuf)); ccl_rpn_delete(cn); } assert(pqf_wrbuf); + + ZOOM_query q = ZOOM_query_create(); + ZOOM_query_sortby2(q, b->sptr->sortStrategy.c_str(), sortkeys.c_str()); + if (b->get_option("sru")) { int status = 0; @@ -1035,9 +1533,11 @@ void yf::Zoom::Frontend::handle_search(mp::Package &package) } if (status == 0) { - yaz_log(YLOG_LOG, "search CQL: %s", wrbuf_cstr(wrb)); - b->search_cql(wrbuf_cstr(wrb), &hits, &error, &addinfo); + 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); wrbuf_destroy(wrb); wrbuf_destroy(pqf_wrbuf); @@ -1052,12 +1552,13 @@ void yf::Zoom::Frontend::handle_search(mp::Package &package) } else { - yaz_log(YLOG_LOG, "search PQF: %s", wrbuf_cstr(pqf_wrbuf)); - b->search_pqf(wrbuf_cstr(pqf_wrbuf), &hits, &error, &addinfo); + 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_destroy(q); wrbuf_destroy(pqf_wrbuf); } - - + const char *element_set_name = 0; Odr_int number_to_present = 0; if (!error) @@ -1106,7 +1607,7 @@ void yf::Zoom::Frontend::handle_present(mp::Package &package) element_set_name = comp->u.simple->u.generic; Odr_int number_of_records_returned = 0; int error = 0; - const char *addinfo = 0; + char *addinfo = 0; Z_Records *records = get_records( *pr->resultSetStartPoint - 1, *pr->numberOfRecordsRequested, &error, &addinfo, &number_of_records_returned, odr, m_backend, @@ -1130,6 +1631,9 @@ void yf::Zoom::Frontend::handle_package(mp::Package &package) else if (gdu->which == Z_GDU_Z3950) { Z_APDU *apdu_req = gdu->u.z3950; + + if (m_backend) + wrbuf_rewind(m_backend->m_apdu_wrbuf); if (apdu_req->which == Z_APDU_initRequest) { mp::odr odr; @@ -1155,6 +1659,11 @@ void yf::Zoom::Frontend::handle_package(mp::Package &package) "zoom filter cannot handle this APDU"); package.session().close(); } + if (m_backend) + { + WRBUF w = m_backend->m_apdu_wrbuf; + package.log_write(wrbuf_buf(w), wrbuf_len(w)); + } } else {