X-Git-Url: http://git.indexdata.com/?a=blobdiff_plain;f=src%2Ffilter_zoom.cpp;h=bbd8302c05319ced402f616b3a0dd237ae6c1cdd;hb=1601cfdde92b0311701c1208f7b322087d2ac9cf;hp=99ed354fd642f7b47ecf824fd595856773de4f8c;hpb=803ba02705871ad2ad4b29b764c64e7a83d958ea;p=metaproxy-moved-to-github.git diff --git a/src/filter_zoom.cpp b/src/filter_zoom.cpp index 99ed354..bbd8302 100644 --- a/src/filter_zoom.cpp +++ b/src/filter_zoom.cpp @@ -1,5 +1,5 @@ /* This file is part of Metaproxy. - Copyright (C) 2005-2011 Index Data + Copyright (C) 2005-2012 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 @@ -92,8 +92,11 @@ namespace metaproxy_1 { SearchablePtr sptr; xsltStylesheetPtr xsp; std::string content_session_id; + bool enable_cproxy; + bool enable_explain; + xmlDoc *explain_doc; public: - Backend(SearchablePtr sptr); + Backend(); ~Backend(); void connect(std::string zurl, int *error, char **addinfo, ODR odr); @@ -115,12 +118,21 @@ namespace metaproxy_1 { BackendPtr m_backend; void handle_package(mp::Package &package); void handle_search(mp::Package &package); + + BackendPtr explain_search(mp::Package &package, + std::string &database, + int *error, + char **addinfo, + mp::odr &odr, + std::string &torus_db, + std::string &realm); void handle_present(mp::Package &package); BackendPtr get_backend_from_databases(mp::Package &package, std::string &database, int *error, char **addinfo, - ODR odr); + mp::odr &odr, + int *proxy_step); bool create_content_session(mp::Package &package, BackendPtr b, @@ -128,16 +140,19 @@ namespace metaproxy_1 { char **addinfo, ODR odr, std::string authentication, - std::string proxy); + std::string proxy, + std::string realm); void prepare_elements(BackendPtr b, Odr_oid *preferredRecordSyntax, const char *element_set_name, bool &enable_pz2_retrieval, bool &enable_pz2_transform, + bool &enable_record_transform, bool &assume_marc8_charset); - Z_Records *get_records(Odr_int start, + Z_Records *get_records(Package &package, + Odr_int start, Odr_int number_to_present, int *error, char **addinfo, @@ -145,6 +160,18 @@ namespace metaproxy_1 { ODR odr, BackendPtr b, Odr_oid *preferredRecordSyntax, const char *element_set_name); + Z_Records *get_explain_records(Package &package, + 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); + + void log_diagnostic(mp::Package &package, + int error, const char *addinfo); public: Frontend(Impl *impl); ~Frontend(); @@ -166,7 +193,8 @@ namespace metaproxy_1 { std::map m_clients; boost::mutex m_mutex; boost::condition m_cond_session_ready; - std::string torus_url; + std::string torus_searchable_url; + std::string torus_content_url; std::string default_realm; std::map fieldmap; std::string xsldir; @@ -178,11 +206,37 @@ namespace metaproxy_1 { std::string element_transform; std::string element_raw; std::string proxy; + xsltStylesheetPtr explain_xsp; + xsltStylesheetPtr record_xsp; std::map s_map; + std::string zoom_timeout; }; } } + +static xmlNode *xml_node_search(xmlNode *ptr, int *num, int m) +{ + while (ptr) + { + if (ptr->type == XML_ELEMENT_NODE && + !strcmp((const char *) ptr->name, "recordData")) + { + (*num)++; + if (m == *num) + return ptr; + } + else // else: we don't want to find nested nodes + { + xmlNode *ret_node = xml_node_search(ptr->children, num, m); + if (ret_node) + return ret_node; + } + ptr = ptr->next; + } + return 0; +} + // define Pimpl wrapper forwarding to Impl yf::Zoom::Zoom() : m_p(new Impl) @@ -207,21 +261,27 @@ void yf::Zoom::process(mp::Package &package) const // define Implementation stuff -yf::Zoom::Backend::Backend(SearchablePtr ptr) : sptr(ptr) +yf::Zoom::Backend::Backend() { 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; + enable_cproxy = true; + enable_explain = false; + explain_doc = 0; } yf::Zoom::Backend::~Backend() { if (xsp) xsltFreeStylesheet(xsp); + if (explain_doc) + xmlFreeDoc(explain_doc); ZOOM_connection_destroy(m_connection); ZOOM_resultset_destroy(m_resultset); + wrbuf_destroy(m_apdu_wrbuf); } @@ -231,37 +291,30 @@ void yf::Zoom::Backend::get_zoom_error(int *error, char **addinfo, const char *msg = 0; const char *zoom_addinfo = 0; const char *dset = 0; - *error = ZOOM_connection_error_x(m_connection, &msg, &zoom_addinfo, &dset); - if (*error) + int error0 = ZOOM_connection_error_x(m_connection, &msg, + &zoom_addinfo, &dset); + if (error0) { - if (*error >= ZOOM_ERROR_CONNECT) - { - // turn ZOOM diagnostic into a Bib-1 2: with addinfo=zoom err msg - *error = YAZ_BIB1_TEMPORARY_SYSTEM_ERROR; - *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); - } - } + if (!dset) + dset = "Unknown"; + + if (!strcmp(dset, "info:srw/diagnostic/1")) + *error = yaz_diag_srw_to_bib1(error0); + else if (!strcmp(dset, "Bib-1")) + *error = error0; else + *error = YAZ_BIB1_TEMPORARY_SYSTEM_ERROR; + + *addinfo = (char *) odr_malloc( + odr, 30 + strlen(dset) + strlen(msg) + + (zoom_addinfo ? strlen(zoom_addinfo) : 0)); + **addinfo = '\0'; + if (zoom_addinfo && *zoom_addinfo) { - 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)"); + strcpy(*addinfo, zoom_addinfo); + strcat(*addinfo, " "); } + sprintf(*addinfo + strlen(*addinfo), "(%s %d %s)", dset, error0, msg); } } @@ -276,6 +329,7 @@ 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) { + ZOOM_resultset_destroy(m_resultset); m_resultset = ZOOM_connection_search(m_connection, q); get_zoom_error(error, addinfo, odr); if (*error == 0) @@ -378,15 +432,20 @@ void yf::Zoom::Impl::release_frontend(mp::Package &package) } yf::Zoom::Impl::Impl() : - apdu_log(false), element_transform("pz2") , element_raw("raw") + apdu_log(false), element_transform("pz2") , element_raw("raw"), + zoom_timeout("40") { bibset = ccl_qual_mk(); + explain_xsp = 0; + record_xsp = 0; srand((unsigned int) time(0)); } yf::Zoom::Impl::~Impl() -{ +{ + if (explain_xsp) + xsltFreeStylesheet(explain_xsp); ccl_qual_rm(&bibset); } @@ -491,8 +550,11 @@ yf::Zoom::SearchablePtr yf::Zoom::Impl::parse_torus_record(const xmlNode *ptr) "cclmap_", 7)) { std::string value = mp::xml::get_text(ptr); - ccl_qual_fitem(s->ccl_bibset, value.c_str(), - (const char *) ptr->name + 7); + if (value.length() > 0) + { + ccl_qual_fitem(s->ccl_bibset, value.c_str(), + (const char *) ptr->name + 7); + } } else if (!strncmp((const char *) ptr->name, "sortmap_", 8)) @@ -560,6 +622,9 @@ 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) { + std::string explain_xslt_fname; + std::string record_xslt_fname; + content_tmp_file = "/tmp/cf.XXXXXX.p"; if (path && *path) { @@ -575,7 +640,9 @@ void yf::Zoom::Impl::configure(const xmlNode *ptr, bool test_only, for (attr = ptr->properties; attr; attr = attr->next) { if (!strcmp((const char *) attr->name, "url")) - torus_url = mp::xml::get_text(attr->children); + torus_searchable_url = mp::xml::get_text(attr->children); + else if (!strcmp((const char *) attr->name, "content_url")) + torus_content_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")) @@ -586,11 +653,19 @@ void yf::Zoom::Impl::configure(const xmlNode *ptr, bool test_only, element_raw = mp::xml::get_text(attr->children); else if (!strcmp((const char *) attr->name, "proxy")) proxy = mp::xml::get_text(attr->children); + else if (!strcmp((const char *) attr->name, "explain_xsl")) + explain_xslt_fname = mp::xml::get_text(attr->children); + else if (!strcmp((const char *) attr->name, "record_xsl")) + record_xslt_fname = mp::xml::get_text(attr->children); else throw mp::filter::FilterException( "Bad attribute " + std::string((const char *) attr->name)); } + // If content_url is not given, use value of searchable, to + // ensure backwards compatibility + if (!torus_content_url.length()) + torus_content_url = torus_searchable_url; configure_local_records(ptr->children, test_only); } else if (!strcmp((const char *) ptr->name, "cclmap")) @@ -645,6 +720,19 @@ void yf::Zoom::Impl::configure(const xmlNode *ptr, bool test_only, attr->name)); } } + else if (!strcmp((const char *) ptr->name, "zoom")) + { + const struct _xmlAttr *attr; + for (attr = ptr->properties; attr; attr = attr->next) + { + if (!strcmp((const char *) attr->name, "timeout")) + zoom_timeout = mp::xml::get_text(attr->children); + else + throw mp::filter::FilterException( + "Bad attribute " + std::string((const char *) + attr->name)); + } + } else { throw mp::filter::FilterException @@ -653,6 +741,76 @@ void yf::Zoom::Impl::configure(const xmlNode *ptr, bool test_only, + " in zoom filter"); } } + + if (explain_xslt_fname.length()) + { + const char *path = 0; + + if (xsldir.length()) + path = xsldir.c_str(); + else + path = file_path.c_str(); + + char fullpath[1024]; + char *cp = yaz_filepath_resolve(explain_xslt_fname.c_str(), + path, 0, fullpath); + if (!cp) + { + throw mp::filter::FilterException + ("Cannot read XSLT " + explain_xslt_fname); + } + + xmlDoc *xsp_doc = xmlParseFile(cp); + if (!xsp_doc) + { + throw mp::filter::FilterException + ("Cannot parse XSLT " + explain_xslt_fname); + } + + explain_xsp = xsltParseStylesheetDoc(xsp_doc); + if (!explain_xsp) + { + xmlFreeDoc(xsp_doc); + throw mp::filter::FilterException + ("Cannot parse XSLT " + explain_xslt_fname); + + } + } + + if (record_xslt_fname.length()) + { + const char *path = 0; + + if (xsldir.length()) + path = xsldir.c_str(); + else + path = file_path.c_str(); + + char fullpath[1024]; + char *cp = yaz_filepath_resolve(record_xslt_fname.c_str(), + path, 0, fullpath); + if (!cp) + { + throw mp::filter::FilterException + ("Cannot read XSLT " + record_xslt_fname); + } + + xmlDoc *xsp_doc = xmlParseFile(cp); + if (!xsp_doc) + { + throw mp::filter::FilterException + ("Cannot parse XSLT " + record_xslt_fname); + } + + record_xsp = xsltParseStylesheetDoc(xsp_doc); + if (!record_xsp) + { + xmlFreeDoc(xsp_doc); + throw mp::filter::FilterException + ("Cannot parse XSLT " + record_xslt_fname); + + } + } } bool yf::Zoom::Frontend::create_content_session(mp::Package &package, @@ -660,7 +818,8 @@ bool yf::Zoom::Frontend::create_content_session(mp::Package &package, int *error, char **addinfo, ODR odr, std::string authentication, - std::string proxy) + std::string proxy, + std::string realm) { if (b->sptr->contentConnector.length()) { @@ -694,21 +853,26 @@ bool yf::Zoom::Frontend::create_content_session(mp::Package &package, wrbuf_printf(w, "auth: %s\n", authentication.c_str()); if (proxy.length()) wrbuf_printf(w, "proxy: %s\n", proxy.c_str()); + if (realm.length()) + wrbuf_printf(w, "realm: %s\n", realm.c_str()); fwrite(wrbuf_buf(w), 1, wrbuf_len(w), file); fclose(file); package.log("zoom", YLOG_LOG, "content file: %s", fname); xfree(fname); + wrbuf_destroy(w); } return true; } yf::Zoom::BackendPtr yf::Zoom::Frontend::get_backend_from_databases( mp::Package &package, - std::string &database, int *error, char **addinfo, ODR odr) + std::string &database, int *error, char **addinfo, mp::odr &odr, + int *proxy_step) { std::list::const_iterator map_it; - if (m_backend && m_backend->m_frontend_database == database) + if (m_backend && !m_backend->enable_explain && + m_backend->m_frontend_database == database) return m_backend; std::string input_args; @@ -723,11 +887,16 @@ yf::Zoom::BackendPtr yf::Zoom::Frontend::get_backend_from_databases( torus_db = database; std::string authentication; + std::string content_authentication; std::string proxy; + std::string content_proxy; std::string realm = m_p->default_realm; const char *param_user = 0; const char *param_password = 0; + const char *param_content_user = 0; + const char *param_content_password = 0; + const char *param_nocproxy = 0; int no_parms = 0; char **names; @@ -742,6 +911,8 @@ yf::Zoom::BackendPtr yf::Zoom::Frontend::get_backend_from_databases( const char **out_values = (const char **) odr_malloc(odr, (10 + no_parms) * sizeof(*out_values)); + // may be changed if it's a content connection + std::string torus_url = m_p->torus_searchable_url; int i; for (i = 0; i < no_parms; i++) { @@ -753,12 +924,35 @@ yf::Zoom::BackendPtr yf::Zoom::Frontend::get_backend_from_databases( param_user = value; else if (!strcmp(name, "password")) param_password = value; + else if (!strcmp(name, "content-user")) + param_content_user = value; + else if (!strcmp(name, "content-password")) + param_content_password = value; + else if (!strcmp(name, "content-proxy")) + content_proxy = value; + else if (!strcmp(name, "nocproxy")) + param_nocproxy = value; else if (!strcmp(name, "proxy")) - proxy = value; + { + char **dstr; + int dnum = 0; + nmem_strsplit(((ODR) odr)->mem, ",", value, &dstr, &dnum); + if (*proxy_step >= dnum) + *proxy_step = 0; + else + { + proxy = dstr[*proxy_step]; + + (*proxy_step)++; + if (*proxy_step == dnum) + *proxy_step = 0; + } + } else if (!strcmp(name, "cproxysession")) { out_names[no_out_args] = name; out_values[no_out_args++] = value; + torus_url = m_p->torus_content_url; } else if (!strcmp(name, "realm")) realm = value; @@ -783,16 +977,29 @@ yf::Zoom::BackendPtr yf::Zoom::Frontend::get_backend_from_databases( if (param_password) authentication += "/" + std::string(param_password); } + if (param_content_user) + { + content_authentication = std::string(param_content_user); + if (param_content_password) + content_authentication += "/" + std::string(param_content_password); + } + + if (torus_db.compare("IR-Explain---1") == 0) + return explain_search(package, database, error, addinfo, odr, torus_db, + realm); + 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 if (m_p->torus_url.length() > 0) + else if (torus_url.length() > 0) { - xmlDoc *doc = mp::get_searchable(m_p->torus_url, torus_db, realm, - m_p->proxy); + std::string torus_query = "udb==" + torus_db; + xmlDoc *doc = mp::get_searchable(package,torus_url, torus_db, + torus_query, + realm, m_p->proxy); if (!doc) { *error = YAZ_BIB1_DATABASE_DOES_NOT_EXIST; @@ -904,20 +1111,22 @@ yf::Zoom::BackendPtr yf::Zoom::Frontend::get_backend_from_databases( m_backend.reset(); - BackendPtr b(new Backend(sptr)); + BackendPtr b(new Backend); + b->sptr = sptr; b->xsp = xsp; b->m_frontend_database = database; + b->enable_cproxy = param_nocproxy ? false : true; if (sptr->query_encoding.length()) b->set_option("rpnCharset", sptr->query_encoding); - b->set_option("timeout", "40"); + b->set_option("timeout", m_p->zoom_timeout.c_str()); if (m_p->apdu_log) b->set_option("apdulog", "1"); - if (sptr->piggyback) + if (sptr->piggyback && sptr->sru.length()) b->set_option("count", "1"); /* some SRU servers INSIST on getting maximumRecords > 0 */ b->set_option("piggyback", sptr->piggyback ? "1" : "0"); @@ -962,6 +1171,11 @@ 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) + { + out_names[no_out_args] = "nocproxy"; + out_values[no_out_args++] = odr_strdup(odr, param_nocproxy); + } } else { @@ -978,6 +1192,9 @@ yf::Zoom::BackendPtr yf::Zoom::Frontend::get_backend_from_databases( if (proxy.length()) b->set_option("proxy", proxy); } + if (proxy.length()) + package.log("zoom", YLOG_LOG, "proxy: %s", proxy.c_str()); + std::string url; if (sptr->sru.length()) { @@ -1002,9 +1219,12 @@ yf::Zoom::BackendPtr yf::Zoom::Frontend::get_backend_from_databases( } package.log("zoom", YLOG_LOG, "url: %s", url.c_str()); b->connect(url, error, addinfo, odr); - if (*error == 0) + if (*error == 0 && b->enable_cproxy) create_content_session(package, b, error, addinfo, odr, - authentication, proxy); + content_authentication.length() ? + content_authentication : authentication, + content_proxy.length() ? content_proxy : proxy, + realm); if (*error == 0) m_backend = b; return b; @@ -1015,25 +1235,32 @@ void yf::Zoom::Frontend::prepare_elements(BackendPtr b, const char *element_set_name, bool &enable_pz2_retrieval, bool &enable_pz2_transform, + bool &enable_record_transform, bool &assume_marc8_charset) - { char oid_name_str[OID_STR_MAX]; const char *syntax_name = 0; if (preferredRecordSyntax && - !oid_oidcmp(preferredRecordSyntax, yaz_oid_recsyn_xml) - && element_set_name) + !oid_oidcmp(preferredRecordSyntax, yaz_oid_recsyn_xml)) { - if (!strcmp(element_set_name, m_p->element_transform.c_str())) + if (element_set_name && + !strcmp(element_set_name, m_p->element_transform.c_str())) { enable_pz2_retrieval = true; enable_pz2_transform = true; } - else if (!strcmp(element_set_name, m_p->element_raw.c_str())) + else if (element_set_name && + !strcmp(element_set_name, m_p->element_raw.c_str())) { enable_pz2_retrieval = true; } + else if (m_p->record_xsp) + { + enable_pz2_retrieval = true; + enable_pz2_transform = true; + enable_record_transform = true; + } } if (enable_pz2_retrieval) @@ -1070,7 +1297,68 @@ void yf::Zoom::Frontend::prepare_elements(BackendPtr b, b->set_option("schema", element_set_name); } -Z_Records *yf::Zoom::Frontend::get_records(Odr_int start, +Z_Records *yf::Zoom::Frontend::get_explain_records( + Package &package, + 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) +{ + Odr_int i; + Z_Records *records = 0; + + if (!b->explain_doc) + { + return records; + } + if (number_to_present > 10000) + number_to_present = 10000; + + xmlNode *ptr = xmlDocGetRootElement(b->explain_doc); + + Z_NamePlusRecordList *npl = (Z_NamePlusRecordList *) + odr_malloc(odr, sizeof(*npl)); + npl->records = (Z_NamePlusRecord **) + odr_malloc(odr, number_to_present * sizeof(*npl->records)); + + for (i = 0; i < number_to_present; i++) + { + int num = 0; + xmlNode *res = xml_node_search(ptr, &num, start + i + 1); + if (!res) + break; + xmlBufferPtr xml_buf = xmlBufferCreate(); + xmlNode *tmp_node = xmlCopyNode(res->children, 1); + xmlNodeDump(xml_buf, tmp_node->doc, tmp_node, 0, 0); + + Z_NamePlusRecord *npr = + (Z_NamePlusRecord *) odr_malloc(odr, sizeof(*npr)); + npr->databaseName = odr_strdup(odr, b->m_frontend_database.c_str()); + npr->which = Z_NamePlusRecord_databaseRecord; + npr->u.databaseRecord = + z_ext_record_xml(odr, + (const char *) xml_buf->content, xml_buf->use); + npl->records[i] = npr; + xmlFreeNode(tmp_node); + xmlBufferFree(xml_buf); + } + records = (Z_Records*) odr_malloc(odr, sizeof(*records)); + records->which = Z_Records_DBOSD; + records->u.databaseOrSurDiagnostics = npl; + + npl->num_records = i; + *number_of_records_returned = i; + return records; +} + + +Z_Records *yf::Zoom::Frontend::get_records(Package &package, + Odr_int start, Odr_int number_to_present, int *error, char **addinfo, @@ -1085,13 +1373,19 @@ Z_Records *yf::Zoom::Frontend::get_records(Odr_int start, 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; + bool enable_record_transform = false; prepare_elements(b, preferredRecordSyntax, element_set_name, enable_pz2_retrieval, enable_pz2_transform, + enable_record_transform, assume_marc8_charset); + package.log("zoom", YLOG_LOG, "pz2_retrieval: %s . pz2_transform: %s", + enable_pz2_retrieval ? "yes" : "no", + enable_pz2_transform ? "yes" : "no"); + if (start < 0 || number_to_present <=0) return records; @@ -1112,6 +1406,32 @@ Z_Records *yf::Zoom::Frontend::get_records(Odr_int start, } if (i > 0) { // only return records if no error and at least one record + + const char *xsl_parms[3]; + char cproxy_host[1024]; + + if (b->enable_cproxy && b->content_session_id.length()) + { + sprintf(cproxy_host, "%s.%s/", + b->content_session_id.c_str(), + m_p->content_proxy_server.c_str()); + + char *q_cproxy_host = (char *) + odr_malloc(odr, strlen(cproxy_host) + 3); + strcpy(q_cproxy_host, "\""); + strcat(q_cproxy_host, cproxy_host); + strcat(q_cproxy_host, "\""); + + xsl_parms[0] = "cproxyhost"; + xsl_parms[1] = q_cproxy_host; + xsl_parms[2] = 0; + } + else + { + xsl_parms[0] = 0; + *cproxy_host = '\0'; + } + char *odr_database = odr_strdup(odr, b->m_frontend_database.c_str()); Z_NamePlusRecordList *npl = (Z_NamePlusRecordList *) @@ -1124,11 +1444,13 @@ Z_Records *yf::Zoom::Frontend::get_records(Odr_int start, { Z_NamePlusRecord *npr = 0; const char *addinfo; + int sur_error = ZOOM_record_error(recs[i], 0 /* msg */, &addinfo, 0 /* diagset */); if (sur_error) { + log_diagnostic(package, sur_error, addinfo); npr = zget_surrogateDiagRec(odr, odr_database, sur_error, addinfo); } @@ -1148,7 +1470,9 @@ Z_Records *yf::Zoom::Frontend::get_records(Odr_int start, strcat(rec_type_str, "; charset="); strcat(rec_type_str, record_encoding); } - + + package.log("zoom", YLOG_LOG, "Getting record of type %s", + rec_type_str); int rec_len; xmlChar *xmlrec_buf = 0; const char *rec_buf = ZOOM_record_get(recs[i], rec_type_str, @@ -1157,11 +1481,16 @@ Z_Records *yf::Zoom::Frontend::get_records(Odr_int start, { std::string addinfo("ZOOM_record_get failed for type "); + int error = YAZ_BIB1_SYSTEM_ERROR_IN_PRESENTING_RECORDS; addinfo += rec_type_str; - npr = zget_surrogateDiagRec( - odr, odr_database, - YAZ_BIB1_SYSTEM_ERROR_IN_PRESENTING_RECORDS, - addinfo.c_str()); + log_diagnostic(package, error, addinfo.c_str()); + npr = zget_surrogateDiagRec(odr, odr_database, + error, addinfo.c_str()); + } + else + { + package.log_write(rec_buf, rec_len); + package.log_write("\r\n", 2); } if (rec_buf && b->xsp && enable_pz2_transform) @@ -1169,76 +1498,75 @@ Z_Records *yf::Zoom::Frontend::get_records(Odr_int start, xmlDoc *rec_doc = xmlParseMemory(rec_buf, rec_len); if (!rec_doc) { + const char *addinfo = "xml parse failed for record"; + int error = YAZ_BIB1_SYSTEM_ERROR_IN_PRESENTING_RECORDS; + log_diagnostic(package, error, addinfo); npr = zget_surrogateDiagRec( - odr, odr_database, - YAZ_BIB1_SYSTEM_ERROR_IN_PRESENTING_RECORDS, - "xml parse failed for record"); + odr, odr_database, error, addinfo); } else { - xmlDoc *rec_res = - xsltApplyStylesheet(b->xsp, rec_doc, 0); + // first stage XSLT - per target + xsltStylesheetPtr xsp = b->xsp; + xmlDoc *rec_res = xsltApplyStylesheet(xsp, rec_doc, + xsl_parms); + // insert generated-url + if (rec_res) + { + std::string res = + mp::xml::url_recipe_handle(rec_res, + b->sptr->urlRecipe); + if (res.length()) + { + xmlNode *ptr = xmlDocGetRootElement(rec_res); + 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); + } + } + // second stage XSLT - common + if (rec_res && m_p->record_xsp && + enable_record_transform) + { + xmlDoc *tmp_doc = rec_res; + xsp = m_p->record_xsp; + rec_res = xsltApplyStylesheet(xsp, tmp_doc, + xsl_parms); + xmlFreeDoc(tmp_doc); + } + // get result out of it if (rec_res) { xsltSaveResultToString(&xmlrec_buf, &rec_len, - rec_res, b->xsp); + rec_res, xsp); rec_buf = (const char *) xmlrec_buf; + package.log("zoom", YLOG_LOG, "xslt successful"); + package.log_write(rec_buf, rec_len); xmlFreeDoc(rec_res); } if (!rec_buf) { std::string addinfo; + int error = + YAZ_BIB1_SYSTEM_ERROR_IN_PRESENTING_RECORDS; addinfo = "xslt apply failed for " + b->sptr->transform_xsl_fname; + log_diagnostic(package, error, addinfo.c_str()); npr = zget_surrogateDiagRec( - odr, odr_database, - YAZ_BIB1_SYSTEM_ERROR_IN_PRESENTING_RECORDS, - addinfo.c_str()); + odr, odr_database, error, addinfo.c_str()); } xmlFreeDoc(rec_doc); } } - if (rec_buf) - { - 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); - } if (!npr) { if (!rec_buf) @@ -1320,6 +1648,92 @@ struct cql_node *yf::Zoom::Impl::convert_cql_fields(struct cql_node *cn, return r; } +void yf::Zoom::Frontend::log_diagnostic(mp::Package &package, + int error, const char *addinfo) +{ + const char *err_msg = yaz_diag_bib1_str(error); + if (addinfo) + package.log("zoom", YLOG_WARN, "Diagnostic %d %s: %s", + error, err_msg, addinfo); + else + package.log("zoom", YLOG_WARN, "Diagnostic %d %s:", + error, err_msg); +} + +yf::Zoom::BackendPtr yf::Zoom::Frontend::explain_search(mp::Package &package, + std::string &database, + int *error, + char **addinfo, + mp::odr &odr, + std::string &torus_db, + std::string &realm) +{ + m_backend.reset(); + + BackendPtr b(new Backend); + + b->m_frontend_database = database; + b->enable_explain = true; + + Z_GDU *gdu = package.request().get(); + Z_APDU *apdu_req = gdu->u.z3950; + Z_SearchRequest *sr = apdu_req->u.searchRequest; + Z_Query *query = sr->query; + + if (!m_p->explain_xsp) + { + *error = YAZ_BIB1_UNSPECIFIED_ERROR; + *addinfo = + odr_strdup(odr, "IR-Explain---1 unsupported. " + "Torus explain_xsl not defined"); + return m_backend; + } + else if (query->which == Z_Query_type_104 && + query->u.type_104->which == Z_External_CQL) + { + std::string torus_url = m_p->torus_searchable_url; + std::string torus_query(query->u.type_104->u.cql); + xmlDoc *doc = mp::get_searchable(package, torus_url, "", + torus_query, + realm, m_p->proxy); + if (m_p->explain_xsp) + { + xmlDoc *rec_res = xsltApplyStylesheet(m_p->explain_xsp, doc, 0); + + xmlFreeDoc(doc); + doc = rec_res; + } + if (!doc) + { + *error = YAZ_BIB1_UNSPECIFIED_ERROR; + *addinfo = odr_strdup(odr, "IR-Explain---1 problem. " + "Could not obtain Torus records for Explain"); + } + else + { + xmlNode *ptr = xmlDocGetRootElement(doc); + int hits = 0; + + xml_node_search(ptr, &hits, 0); + + Z_APDU *apdu_res = odr.create_searchResponse(apdu_req, 0, 0); + apdu_res->u.searchResponse->resultCount = odr_intdup(odr, hits); + package.response() = apdu_res; + m_backend = b; + } + if (b->explain_doc) + xmlFreeDoc(b->explain_doc); + b->explain_doc = doc; + return m_backend; + } + else + { + *error = YAZ_BIB1_QUERY_TYPE_UNSUPP; + *addinfo = odr_strdup(odr, "IR-Explain---1 only supports CQL"); + return m_backend; + } +} + void yf::Zoom::Frontend::handle_search(mp::Package &package) { Z_GDU *gdu = package.request().get(); @@ -1329,33 +1743,48 @@ void yf::Zoom::Frontend::handle_search(mp::Package &package) Z_SearchRequest *sr = apdu_req->u.searchRequest; if (sr->num_databaseNames != 1) { - apdu_res = odr.create_searchResponse( - apdu_req, YAZ_BIB1_TOO_MANY_DATABASES_SPECIFIED, 0); + int error = YAZ_BIB1_TOO_MANY_DATABASES_SPECIFIED; + log_diagnostic(package, error, 0); + apdu_res = odr.create_searchResponse(apdu_req, error, 0); package.response() = apdu_res; return; } + int proxy_step = 0; + +next_proxy: int error = 0; char *addinfo = 0; std::string db(sr->databaseNames[0]); + BackendPtr b = get_backend_from_databases(package, db, &error, - &addinfo, odr); + &addinfo, odr, &proxy_step); + if (error && proxy_step) + { + package.log("zoom", YLOG_WARN, + "create backend failed: trying next proxy"); + goto next_proxy; + } if (error) { - apdu_res = - odr.create_searchResponse(apdu_req, error, addinfo); + log_diagnostic(package, error, addinfo); + apdu_res = odr.create_searchResponse(apdu_req, error, addinfo); package.response() = apdu_res; return; } + if (!b || b->enable_explain) + return; b->set_option("setname", "default"); bool enable_pz2_retrieval = false; bool enable_pz2_transform = false; + bool enable_record_transform = false; bool assume_marc8_charset = false; prepare_elements(b, sr->preferredRecordSyntax, 0 /*element_set_name */, enable_pz2_retrieval, enable_pz2_transform, + enable_record_transform, assume_marc8_charset); Odr_int hits = 0; @@ -1388,11 +1817,11 @@ void yf::Zoom::Frontend::handle_search(mp::Package &package) if (r) { cql_parser_destroy(cp); - package.log("zoom", YLOG_WARN, "CQL syntax error"); + error = YAZ_BIB1_MALFORMED_QUERY; + const char *addinfo = "CQL syntax error"; + log_diagnostic(package, error, addinfo); apdu_res = - odr.create_searchResponse(apdu_req, - YAZ_BIB1_MALFORMED_QUERY, - "CQL syntax error"); + odr.create_searchResponse(apdu_req, error, addinfo); package.response() = apdu_res; return; } @@ -1401,82 +1830,94 @@ void yf::Zoom::Frontend::handle_search(mp::Package &package) if (cn_error) { // hopefully we are getting a ptr to a index+relation+term node + error = YAZ_BIB1_UNSUPP_USE_ATTRIBUTE; addinfo = 0; if (cn_error->which == CQL_NODE_ST) addinfo = cn_error->u.st.index; - - apdu_res = - odr.create_searchResponse(apdu_req, - YAZ_BIB1_UNSUPP_USE_ATTRIBUTE, - addinfo); + + log_diagnostic(package, error, addinfo); + apdu_res = odr.create_searchResponse(apdu_req, error, addinfo); package.response() = apdu_res; + cql_parser_destroy(cp); return; } char ccl_buf[1024]; - r = cql_to_ccl_buf(cn, ccl_buf, sizeof(ccl_buf)); - if (r == 0) + if (r) { - ccl_wrbuf = wrbuf_alloc(); - wrbuf_puts(ccl_wrbuf, ccl_buf); - - WRBUF sru_sortkeys_wrbuf = wrbuf_alloc(); + error = YAZ_BIB1_MALFORMED_QUERY; + const char *addinfo = "CQL to CCL conversion error"; - 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); + log_diagnostic(package, error, addinfo); + apdu_res = odr.create_searchResponse(apdu_req, error, addinfo); + package.response() = apdu_res; + cql_parser_destroy(cp); + return; + } + + WRBUF sru_sortkeys_wrbuf = wrbuf_alloc(); + if (cql_sortby_to_sortkeys(cn, wrbuf_vp_puts, sru_sortkeys_wrbuf)) + { + error = YAZ_BIB1_ILLEGAL_SORT_RELATION; + const char *addinfo = "CQL to CCL sortby conversion"; - 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); + log_diagnostic(package, error, addinfo); + apdu_res = odr.create_searchResponse(apdu_req, error, addinfo); + package.response() = apdu_res; + wrbuf_destroy(sru_sortkeys_wrbuf); + cql_parser_destroy(cp); + return; + } + 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); - /* go through sortspec and map fields */ - int token = yaz_tok_move(tp); - while (token != YAZ_TOK_EOF) + ccl_wrbuf = wrbuf_alloc(); + wrbuf_puts(ccl_wrbuf, ccl_buf); + + 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) { - 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); } + yaz_tok_parse_destroy(tp); + wrbuf_destroy(sort_spec_wrbuf); + cql_parser_destroy(cp); - if (r) - { - apdu_res = - odr.create_searchResponse(apdu_req, - YAZ_BIB1_MALFORMED_QUERY, - "CQL to CCL conversion error"); - package.response() = apdu_res; - return; - } } else { - apdu_res = - odr.create_searchResponse(apdu_req, YAZ_BIB1_QUERY_TYPE_UNSUPP, 0); + error = YAZ_BIB1_QUERY_TYPE_UNSUPP; + const char *addinfo = 0; + log_diagnostic(package, error, addinfo); + apdu_res = odr.create_searchResponse(apdu_req, error, addinfo); package.response() = apdu_res; return; } @@ -1494,21 +1935,25 @@ void yf::Zoom::Frontend::handle_search(mp::Package &package) if (!cn) { char *addinfo = odr_strdup(odr, ccl_err_msg(cerror)); - int z3950_diag = YAZ_BIB1_MALFORMED_QUERY; + error = YAZ_BIB1_MALFORMED_QUERY; switch (cerror) { case CCL_ERR_UNKNOWN_QUAL: - z3950_diag = YAZ_BIB1_UNSUPP_USE_ATTRIBUTE; - break; case CCL_ERR_TRUNC_NOT_LEFT: case CCL_ERR_TRUNC_NOT_RIGHT: case CCL_ERR_TRUNC_NOT_BOTH: - z3950_diag = YAZ_BIB1_UNSUPP_TRUNCATION_ATTRIBUTE; +#ifdef CCL_ERR_TRUNC_NOT_EMBED + case CCL_ERR_TRUNC_NOT_EMBED: +#endif +#ifdef CCL_ERR_TRUNC_NOT_SINGLE + case CCL_ERR_TRUNC_NOT_SINGLE: +#endif + error = YAZ_BIB1_UNSUPP_SEARCH; break; } - apdu_res = - odr.create_searchResponse(apdu_req, z3950_diag, addinfo); + log_diagnostic(package, error, addinfo); + apdu_res = odr.create_searchResponse(apdu_req, error, addinfo); package.response() = apdu_res; return; } @@ -1558,9 +2003,10 @@ void yf::Zoom::Frontend::handle_search(mp::Package &package) wrbuf_destroy(pqf_wrbuf); if (status) { - apdu_res = - odr.create_searchResponse(apdu_req, YAZ_BIB1_MALFORMED_QUERY, - "can not convert from RPN to CQL/SOLR"); + 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); package.response() = apdu_res; return; } @@ -1574,6 +2020,14 @@ void yf::Zoom::Frontend::handle_search(mp::Package &package) wrbuf_destroy(pqf_wrbuf); } + if (error && proxy_step) + { + // reset below prevent reuse in get_backend_from_databases + m_backend.reset(); + package.log("zoom", YLOG_WARN, "search failed: trying next proxy"); + goto next_proxy; + } + const char *element_set_name = 0; Odr_int number_to_present = 0; if (!error) @@ -1581,9 +2035,12 @@ void yf::Zoom::Frontend::handle_search(mp::Package &package) Odr_int number_of_records_returned = 0; Z_Records *records = get_records( + package, 0, number_to_present, &error, &addinfo, &number_of_records_returned, odr, b, sr->preferredRecordSyntax, element_set_name); + if (error) + log_diagnostic(package, error, addinfo); apdu_res = odr.create_searchResponse(apdu_req, error, addinfo); if (records) { @@ -1623,19 +2080,42 @@ void yf::Zoom::Frontend::handle_present(mp::Package &package) Odr_int number_of_records_returned = 0; int error = 0; char *addinfo = 0; - Z_Records *records = get_records( - *pr->resultSetStartPoint - 1, *pr->numberOfRecordsRequested, - &error, &addinfo, &number_of_records_returned, odr, m_backend, - pr->preferredRecordSyntax, element_set_name); - apdu_res = odr.create_presentResponse(apdu_req, error, addinfo); - if (records) + if (m_backend->enable_explain) { - apdu_res->u.presentResponse->records = records; - apdu_res->u.presentResponse->numberOfRecordsReturned = - odr_intdup(odr, number_of_records_returned); + Z_Records *records = + get_explain_records( + package, + *pr->resultSetStartPoint - 1, *pr->numberOfRecordsRequested, + &error, &addinfo, &number_of_records_returned, odr, m_backend, + pr->preferredRecordSyntax, element_set_name); + + apdu_res = odr.create_presentResponse(apdu_req, error, addinfo); + if (records) + { + apdu_res->u.presentResponse->records = records; + apdu_res->u.presentResponse->numberOfRecordsReturned = + odr_intdup(odr, number_of_records_returned); + } + package.response() = apdu_res; + } + else + { + Z_Records *records = + get_records(package, + *pr->resultSetStartPoint - 1, *pr->numberOfRecordsRequested, + &error, &addinfo, &number_of_records_returned, odr, m_backend, + pr->preferredRecordSyntax, element_set_name); + + apdu_res = odr.create_presentResponse(apdu_req, error, addinfo); + if (records) + { + apdu_res->u.presentResponse->records = records; + apdu_res->u.presentResponse->numberOfRecordsReturned = + odr_intdup(odr, number_of_records_returned); + } + package.response() = apdu_res; } - package.response() = apdu_res; } void yf::Zoom::Frontend::handle_package(mp::Package &package)