X-Git-Url: http://jsfdemo.indexdata.com/?a=blobdiff_plain;f=src%2Ffilter_zoom.cpp;h=5860e15f6806067fdc1eb9adef9a0e2ccf3ddd88;hb=0a7f68cbd61a5cd143834ee3c62da6256c8e9648;hp=bce817c74e07f5d608950e452efc4d6bedef05c6;hpb=cd54d9cec3b4db6f3b39e7d506c7d88877d54fc7;p=metaproxy-moved-to-github.git diff --git a/src/filter_zoom.cpp b/src/filter_zoom.cpp index bce817c..5860e15 100644 --- a/src/filter_zoom.cpp +++ b/src/filter_zoom.cpp @@ -19,6 +19,7 @@ Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA #include "config.hpp" #include "filter_zoom.hpp" #include +#include #include #include #include "torus.hpp" @@ -29,6 +30,7 @@ Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA #include #include #include +#include #include #include #include @@ -41,6 +43,10 @@ namespace yf = mp::filter; namespace metaproxy_1 { namespace filter { struct Zoom::Searchable : boost::noncopyable { + std::string authentication; + std::string cfAuth; + std::string cfProxy; + std::string cfSubDb; std::string database; std::string target; std::string query_encoding; @@ -110,14 +116,14 @@ namespace metaproxy_1 { private: FrontendPtr get_frontend(mp::Package &package); void release_frontend(mp::Package &package); - void parse_torus(const xmlNode *ptr); - - std::listm_searchables; - + SearchablePtr parse_torus(const xmlNode *ptr); + struct cql_node *convert_cql_fields(struct cql_node *cn, ODR odr); std::map m_clients; boost::mutex m_mutex; boost::condition m_cond_session_ready; - mp::Torus torus; + std::string torus_url; + std::map fieldmap; + std::string xsldir; }; } } @@ -201,7 +207,7 @@ int yf::Zoom::Backend::get_error(const char **addinfo) yf::Zoom::Searchable::Searchable() { piggyback = true; - use_turbomarc = false; + use_turbomarc = true; ccl_bibset = ccl_qual_mk(); } @@ -272,10 +278,11 @@ yf::Zoom::Impl::~Impl() { } -void yf::Zoom::Impl::parse_torus(const xmlNode *ptr1) +yf::Zoom::SearchablePtr yf::Zoom::Impl::parse_torus(const xmlNode *ptr1) { + SearchablePtr notfound; if (!ptr1) - return ; + return notfound; for (ptr1 = ptr1->children; ptr1; ptr1 = ptr1->next) { if (ptr1->type != XML_ELEMENT_NODE) @@ -296,7 +303,27 @@ void yf::Zoom::Impl::parse_torus(const xmlNode *ptr1) { if (ptr3->type != XML_ELEMENT_NODE) continue; - if (!strcmp((const char *) ptr3->name, "id")) + if (!strcmp((const char *) ptr3->name, + "authentication")) + { + s->authentication = mp::xml::get_text(ptr3); + } + else if (!strcmp((const char *) ptr3->name, + "cfAuth")) + { + s->cfAuth = mp::xml::get_text(ptr3); + } + else if (!strcmp((const char *) ptr3->name, + "cfProxy")) + { + s->cfProxy = mp::xml::get_text(ptr3); + } + else if (!strcmp((const char *) ptr3->name, + "cfSubDb")) + { + s->cfSubDb = mp::xml::get_text(ptr3); + } + else if (!strcmp((const char *) ptr3->name, "id")) { s->database = mp::xml::get_text(ptr3); } @@ -341,7 +368,7 @@ void yf::Zoom::Impl::parse_torus(const xmlNode *ptr1) else if (!strcmp((const char *) ptr3->name, "useTurboMarc")) { - s->use_turbomarc = mp::xml::get_bool(ptr3, false); + ; // useTurboMarc is ignored } else if (!strncmp((const char *) ptr3->name, "cclmap_", 7)) @@ -351,17 +378,12 @@ void yf::Zoom::Impl::parse_torus(const xmlNode *ptr1) (const char *) ptr3->name + 7); } } - if (s->database.length() && s->target.length()) - { - yaz_log(YLOG_LOG, "add db=%s target=%s turbomarc=%s", - s->database.c_str(), s->target.c_str(), - s->use_turbomarc ? "1" : "0"); - m_searchables.push_back(s); - } + return s; } } } } + return notfound; } void yf::Zoom::Impl::configure(const xmlNode *ptr, bool test_only) @@ -370,30 +392,43 @@ void yf::Zoom::Impl::configure(const xmlNode *ptr, bool test_only) { if (ptr->type != XML_ELEMENT_NODE) continue; - if (!strcmp((const char *) ptr->name, "records")) - { - parse_torus(ptr); - } else if (!strcmp((const char *) ptr->name, "torus")) { - std::string url; const struct _xmlAttr *attr; for (attr = ptr->properties; attr; attr = attr->next) { if (!strcmp((const char *) attr->name, "url")) - url = mp::xml::get_text(attr->children); + torus_url = mp::xml::get_text(attr->children); + else if (!strcmp((const char *) attr->name, "xsldir")) + xsldir = mp::xml::get_text(attr->children); else throw mp::filter::FilterException( "Bad attribute " + std::string((const char *) attr->name)); } - torus.read_searchables(url); - xmlDoc *doc = torus.get_doc(); - if (doc) + } + else if (!strcmp((const char *) ptr->name, "fieldmap")) + { + const struct _xmlAttr *attr; + std::string ccl_field; + std::string cql_field; + for (attr = ptr->properties; attr; attr = attr->next) { - xmlNode *ptr = xmlDocGetRootElement(doc); - parse_torus(ptr); + if (!strcmp((const char *) attr->name, "ccl")) + ccl_field = mp::xml::get_text(attr->children); + else if (!strcmp((const char *) attr->name, "cql")) + cql_field = mp::xml::get_text(attr->children); + else + throw mp::filter::FilterException( + "Bad attribute " + std::string((const char *) + attr->name)); } + if (cql_field.length()) + fieldmap[cql_field] = ccl_field; + } + else if (!strcmp((const char *) ptr->name, "records")) + { + yaz_log(YLOG_WARN, "records ignored!"); } else { @@ -412,29 +447,34 @@ yf::Zoom::BackendPtr yf::Zoom::Frontend::get_backend_from_databases( if (m_backend && m_backend->m_frontend_database == database) return m_backend; - std::list::iterator map_s = - m_p->m_searchables.begin(); - - std::string c_db = mp::util::database_name_normalize(database); - - while (map_s != m_p->m_searchables.end()) + xmlDoc *doc = mp::get_searchable(m_p->torus_url, database); + if (!doc) { - if (c_db.compare((*map_s)->database) == 0) - break; - map_s++; + *error = YAZ_BIB1_DATABASE_DOES_NOT_EXIST; + *addinfo = database.c_str(); + BackendPtr b; + return b; } - if (map_s == m_p->m_searchables.end()) + SearchablePtr sptr = m_p->parse_torus(xmlDocGetRootElement(doc)); + xmlFreeDoc(doc); + if (!sptr) { *error = YAZ_BIB1_DATABASE_DOES_NOT_EXIST; *addinfo = database.c_str(); BackendPtr b; return b; } - + xsltStylesheetPtr xsp = 0; - if ((*map_s)->transform_xsl_fname.length()) + if (sptr->transform_xsl_fname.length()) { - xmlDoc *xsp_doc = xmlParseFile((*map_s)->transform_xsl_fname.c_str()); + std::string fname; + + if (m_p->xsldir.length()) + fname = m_p->xsldir + "/" + sptr->transform_xsl_fname; + else + fname = sptr->transform_xsl_fname; + xmlDoc *xsp_doc = xmlParseFile(fname.c_str()); if (!xsp_doc) { *error = YAZ_BIB1_TEMPORARY_SYSTEM_ERROR; @@ -453,18 +493,49 @@ yf::Zoom::BackendPtr yf::Zoom::Frontend::get_backend_from_databases( } } - SearchablePtr sptr = *map_s; - m_backend.reset(); BackendPtr b(new Backend(sptr)); + std::string cf_parm; b->xsp = xsp; b->m_frontend_database = database; + std::string authentication = sptr->authentication; if (sptr->query_encoding.length()) b->set_option("rpnCharset", sptr->query_encoding.c_str()); + if (sptr->cfAuth.length()) + { + b->set_option("user", sptr->cfAuth.c_str()); + 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)); + } + else + cf_parm += "user=" + mp::util::uri_encode(authentication); + } + } + else if (authentication.length()) + b->set_option("user", authentication.c_str()); + + if (sptr->cfProxy.length()) + { + if (cf_parm.length()) + cf_parm += "&"; + cf_parm += "proxy=" + mp::util::uri_encode(sptr->cfProxy); + } + if (sptr->cfSubDb.length()) + { + if (cf_parm.length()) + cf_parm += "&"; + cf_parm += "subdatabase=" + mp::util::uri_encode(sptr->cfSubDb); + } + std::string url; if (sptr->sru.length()) { @@ -472,8 +543,13 @@ yf::Zoom::BackendPtr yf::Zoom::Frontend::get_backend_from_databases( b->set_option("sru", sptr->sru.c_str()); } else + { url = sptr->target; - + } + if (cf_parm.length()) + { + url += "," + cf_parm; + } b->connect(url, error, addinfo); if (*error == 0) { @@ -511,7 +587,8 @@ Z_Records *yf::Zoom::Frontend::get_records(Odr_int start, if (preferredRecordSyntax) { if (!oid_oidcmp(preferredRecordSyntax, yaz_oid_recsyn_xml) - && !strcmp(element_set_name, "pz2")) + && element_set_name && + !strcmp(element_set_name, "pz2")) { if (b->sptr->request_syntax.length()) { @@ -645,6 +722,38 @@ 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) +{ + struct cql_node *r = 0; + if (!cn) + return 0; + switch (cn->which) + { + case CQL_NODE_ST: + if (cn->u.st.index) + { + std::map::const_iterator it; + it = fieldmap.find(cn->u.st.index); + if (it == fieldmap.end()) + return cn; + if (it->second.length()) + cn->u.st.index = odr_strdup(odr, it->second.c_str()); + else + cn->u.st.index = 0; + } + break; + case CQL_NODE_BOOL: + r = convert_cql_fields(cn->u.boolean.left, odr); + if (!r) + r = convert_cql_fields(cn->u.boolean.right, odr); + break; + case CQL_NODE_SORT: + r = convert_cql_fields(cn->u.sort.search, odr); + break; + } + return r; +} void yf::Zoom::Frontend::handle_search(mp::Package &package) { @@ -664,7 +773,14 @@ void yf::Zoom::Frontend::handle_search(mp::Package &package) int error = 0; const char *addinfo = 0; std::string db(sr->databaseNames[0]); - BackendPtr b = get_backend_from_databases(db, &error, &addinfo); + std::string torus_db; + size_t db_arg_pos = db.find(','); + if (db_arg_pos != std::string::npos) + torus_db = db.substr(0, db_arg_pos); + else + torus_db = db; + + BackendPtr b = get_backend_from_databases(torus_db, &error, &addinfo); if (error) { apdu_res = @@ -678,45 +794,73 @@ void yf::Zoom::Frontend::handle_search(mp::Package &package) Odr_int hits = 0; Z_Query *query = sr->query; + WRBUF ccl_wrbuf = 0; + WRBUF pqf_wrbuf = 0; + if (query->which == Z_Query_type_1 || query->which == Z_Query_type_101) { // RPN - WRBUF w = wrbuf_alloc(); - yaz_rpnquery_to_wrbuf(w, query->u.type_1); - - b->search_pqf(wrbuf_cstr(w), &hits, &error, &addinfo); - - wrbuf_destroy(w); + pqf_wrbuf = wrbuf_alloc(); + yaz_rpnquery_to_wrbuf(pqf_wrbuf, query->u.type_1); } else if (query->which == Z_Query_type_2) { // CCL - WRBUF w = wrbuf_alloc(); - wrbuf_write(w, (const char *) query->u.type_2->buf, + ccl_wrbuf = wrbuf_alloc(); + wrbuf_write(ccl_wrbuf, (const char *) query->u.type_2->buf, query->u.type_2->len); - int cerror, cpos; - struct ccl_rpn_node *cn; - cn = ccl_find_str(b->sptr->ccl_bibset, wrbuf_cstr(w), &cerror, &cpos); - wrbuf_destroy(w); - - if (!cn) + } + else if (query->which == Z_Query_type_104 && + query->u.type_104->which == Z_External_CQL) + { + // CQL + const char *cql = query->u.type_104->u.cql; + CQL_parser cp = cql_parser_create(); + int r = cql_parser_string(cp, cql); + if (r) { - char *addinfo = odr_strdup(odr, ccl_err_msg(cerror)); - + cql_parser_destroy(cp); apdu_res = odr.create_searchResponse(apdu_req, YAZ_BIB1_MALFORMED_QUERY, + "CQL syntax error"); + package.response() = apdu_res; + return; + } + struct cql_node *cn = cql_parser_result(cp); + struct cql_node *cn_error = m_p->convert_cql_fields(cn, odr); + if (cn_error) + { + // hopefully we are getting a ptr to a index+relation+term node + 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); package.response() = apdu_res; return; } - w = wrbuf_alloc(); - ccl_pquery(w, cn); - - b->search_pqf(wrbuf_cstr(w), &hits, &error, &addinfo); - - ccl_rpn_delete(cn); - wrbuf_destroy(w); + char ccl_buf[1024]; + + r = cql_to_ccl_buf(cn, ccl_buf, sizeof(ccl_buf)); + if (r == 0) + { + ccl_wrbuf = wrbuf_alloc(); + wrbuf_puts(ccl_wrbuf, ccl_buf); + } + 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 { @@ -725,6 +869,47 @@ void yf::Zoom::Frontend::handle_search(mp::Package &package) package.response() = apdu_res; return; } + + if (ccl_wrbuf) + { + // CCL to PQF + assert(pqf_wrbuf == 0); + int cerror, cpos; + struct ccl_rpn_node *cn; + yaz_log(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); + if (!cn) + { + char *addinfo = odr_strdup(odr, ccl_err_msg(cerror)); + int z3950_diag = 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; + break; + } + apdu_res = + odr.create_searchResponse(apdu_req, z3950_diag, addinfo); + package.response() = apdu_res; + return; + } + pqf_wrbuf = wrbuf_alloc(); + ccl_pquery(pqf_wrbuf, cn); + ccl_rpn_delete(cn); + } + + assert(pqf_wrbuf); + b->search_pqf(wrbuf_cstr(pqf_wrbuf), &hits, &error, &addinfo); + + wrbuf_destroy(pqf_wrbuf); const char *element_set_name = 0; Odr_int number_to_present = 0;