src/filter_zoom.cpp

   1 /* This file is part of Metaproxy.
   2    Copyright (C) 2005-2011 Index Data
   3
   4 Metaproxy is free software; you can redistribute it and/or modify it under
   5 the terms of the GNU General Public License as published by the Free
   6 Software Foundation; either version 2, or (at your option) any later
   7 version.
   8
   9 Metaproxy is distributed in the hope that it will be useful, but WITHOUT ANY
  10 WARRANTY; without even the implied warranty of MERCHANTABILITY or
  11 FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License
  12 for more details.
  13
  14 You should have received a copy of the GNU General Public License
  15 along with this program; if not, write to the Free Software
  16 Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
  17 */
  18
  19 #include "config.hpp"
  20 #include "filter_zoom.hpp"
  21 #include <yaz/zoom.h>
  22 #include <yaz/yaz-version.h>
  23 #include <yaz/srw.h>
  24 #include <metaproxy/package.hpp>
  25 #include <metaproxy/util.hpp>
  26 #include <metaproxy/xmlutil.hpp>
  27 #include "torus.hpp"
  28
  29 #include <libxslt/xsltutils.h>
  30 #include <libxslt/transform.h>
  31
  32 #include <boost/thread/mutex.hpp>
  33 #include <boost/thread/condition.hpp>
  34 #include <yaz/ccl_xml.h>
  35 #include <yaz/ccl.h>
  36 #include <yaz/rpn2cql.h>
  37 #include <yaz/rpn2solr.h>
  38 #include <yaz/pquery.h>
  39 #include <yaz/cql.h>
  40 #include <yaz/oid_db.h>
  41 #include <yaz/diagbib1.h>
  42 #include <yaz/log.h>
  43 #include <yaz/zgdu.h>
  44 #include <yaz/querytowrbuf.h>
  45
  46 namespace mp = metaproxy_1;
  47 namespace yf = mp::filter;
  48
  49 namespace metaproxy_1 {
  50     namespace filter {
  51         struct Zoom::Searchable : boost::noncopyable {
  52             std::string authentication;
  53             std::string cfAuth;
  54             std::string cfProxy;
  55             std::string cfSubDb;
  56             std::string udb;
  57             std::string target;
  58             std::string query_encoding;
  59             std::string sru;
  60             std::string request_syntax;
  61             std::string element_set;
  62             std::string record_encoding;
  63             std::string transform_xsl_fname;
  64             std::string urlRecipe;
  65             bool use_turbomarc;
  66             bool piggyback;
  67             CCL_bibset ccl_bibset;
  68             Searchable(CCL_bibset base);
  69             ~Searchable();
  70         };
  71         class Zoom::Backend : boost::noncopyable {
  72             friend class Impl;
  73             friend class Frontend;
  74             std::string zurl;
  75             ZOOM_connection m_connection;
  76             ZOOM_resultset m_resultset;
  77             std::string m_frontend_database;
  78             SearchablePtr sptr;
  79             xsltStylesheetPtr xsp;
  80         public:
  81             Backend(SearchablePtr sptr);
  82             ~Backend();
  83             void connect(std::string zurl, int *error, const char **addinfo);
  84             void search_pqf(const char *pqf, Odr_int *hits,
  85                             int *error, const char **addinfo);
  86             void search_cql(const char *cql, Odr_int *hits,
  87                             int *error, const char **addinfo);
  88             void present(Odr_int start, Odr_int number, ZOOM_record *recs,
  89                          int *error, const char **addinfo);
  90             void set_option(const char *name, const char *value);
  91             const char *get_option(const char *name);
  92             void get_zoom_error(int *error, const char **addinfo);
  93         };
  94         class Zoom::Frontend : boost::noncopyable {
  95             friend class Impl;
  96             Impl *m_p;
  97             bool m_is_virtual;
  98             bool m_in_use;
  99             yazpp_1::GDU m_init_gdu;
 100             BackendPtr m_backend;
 101             void handle_package(mp::Package &package);
 102             void handle_search(mp::Package &package);
 103             void handle_present(mp::Package &package);
 104             BackendPtr get_backend_from_databases(std::string &database,
 105                                                   int *error,
 106                                                   char **addinfo,
 107                                                   ODR odr);
 108             Z_Records *get_records(Odr_int start,
 109                                    Odr_int number_to_present,
 110                                    int *error,
 111                                    const char **addinfo,
 112                                    Odr_int *number_of_records_returned,
 113                                    ODR odr, BackendPtr b,
 114                                    Odr_oid *preferredRecordSyntax,
 115                                    const char *element_set_name);
 116         public:
 117             Frontend(Impl *impl);
 118             ~Frontend();
 119         };
 120         class Zoom::Impl {
 121             friend class Frontend;
 122         public:
 123             Impl();
 124             ~Impl();
 125             void process(metaproxy_1::Package & package);
 126             void configure(const xmlNode * ptr, bool test_only);
 127         private:
 128             void configure_local_records(const xmlNode * ptr, bool test_only);
 129             FrontendPtr get_frontend(mp::Package &package);
 130             void release_frontend(mp::Package &package);
 131             SearchablePtr parse_torus_record(const xmlNode *ptr);
 132             struct cql_node *convert_cql_fields(struct cql_node *cn, ODR odr);
 133             std::map<mp::Session, FrontendPtr> m_clients;
 134             boost::mutex m_mutex;
 135             boost::condition m_cond_session_ready;
 136             std::string torus_url;
 137             std::map<std::string,std::string> fieldmap;
 138             std::string xsldir;
 139             CCL_bibset bibset;
 140             std::string element_transform;
 141             std::string element_raw;
 142             std::map<std::string,SearchablePtr> s_map;
 143         };
 144     }
 145 }
 146
 147 // define Pimpl wrapper forwarding to Impl
 148
 149 yf::Zoom::Zoom() : m_p(new Impl)
 150 {
 151 }
 152
 153 yf::Zoom::~Zoom()
 154 {  // must have a destructor because of boost::scoped_ptr
 155 }
 156
 157 void yf::Zoom::configure(const xmlNode *xmlnode, bool test_only)
 158 {
 159     m_p->configure(xmlnode, test_only);
 160 }
 161
 162 void yf::Zoom::process(mp::Package &package) const
 163 {
 164     m_p->process(package);
 165 }
 166
 167
 168 // define Implementation stuff
 169
 170 yf::Zoom::Backend::Backend(SearchablePtr ptr) : sptr(ptr)
 171 {
 172     m_connection = ZOOM_connection_create(0);
 173     m_resultset = 0;
 174     xsp = 0;
 175 }
 176
 177 yf::Zoom::Backend::~Backend()
 178 {
 179     if (xsp)
 180         xsltFreeStylesheet(xsp);
 181     ZOOM_connection_destroy(m_connection);
 182     ZOOM_resultset_destroy(m_resultset);
 183 }
 184
 185
 186 void yf::Zoom::Backend::get_zoom_error(int *error, const char **addinfo)
 187 {
 188     const char *msg = 0;
 189     *error = ZOOM_connection_error(m_connection, &msg, addinfo);
 190     if (*error)
 191     {
 192         if (*error >= ZOOM_ERROR_CONNECT)
 193         {
 194             // turn ZOOM diagnostic into a Bib-1 2: with addinfo=zoom err msg
 195             *error = YAZ_BIB1_TEMPORARY_SYSTEM_ERROR;
 196             if (addinfo)
 197                 *addinfo = msg;
 198         }
 199     }
 200 }
 201
 202 void yf::Zoom::Backend::connect(std::string zurl,
 203                                 int *error, const char **addinfo)
 204 {
 205     ZOOM_connection_connect(m_connection, zurl.c_str(), 0);
 206     get_zoom_error(error, addinfo);
 207 }
 208
 209 void yf::Zoom::Backend::search_pqf(const char *pqf, Odr_int *hits,
 210                                    int *error, const char **addinfo)
 211 {
 212     m_resultset = ZOOM_connection_search_pqf(m_connection, pqf);
 213     get_zoom_error(error, addinfo);
 214     if (*error == 0)
 215         *hits = ZOOM_resultset_size(m_resultset);
 216     else
 217         *hits = 0;
 218 }
 219
 220 void yf::Zoom::Backend::search_cql(const char *cql, Odr_int *hits,
 221                                    int *error, const char **addinfo)
 222 {
 223     ZOOM_query q = ZOOM_query_create();
 224
 225     ZOOM_query_cql(q, cql);
 226
 227     m_resultset = ZOOM_connection_search(m_connection, q);
 228     ZOOM_query_destroy(q);
 229     get_zoom_error(error, addinfo);
 230     if (*error == 0)
 231         *hits = ZOOM_resultset_size(m_resultset);
 232     else
 233         *hits = 0;
 234 }
 235
 236 void yf::Zoom::Backend::present(Odr_int start, Odr_int number,
 237                                 ZOOM_record *recs,
 238                                 int *error, const char **addinfo)
 239 {
 240     ZOOM_resultset_records(m_resultset, recs, start, number);
 241     get_zoom_error(error, addinfo);
 242 }
 243
 244 void yf::Zoom::Backend::set_option(const char *name, const char *value)
 245 {
 246     ZOOM_connection_option_set(m_connection, name, value);
 247     if (m_resultset)
 248         ZOOM_resultset_option_set(m_resultset, name, value);
 249 }
 250
 251 const char *yf::Zoom::Backend::get_option(const char *name)
 252 {
 253     return ZOOM_connection_option_get(m_connection, name);
 254 }
 255
 256 yf::Zoom::Searchable::Searchable(CCL_bibset base)
 257 {
 258     piggyback = true;
 259     use_turbomarc = true;
 260     ccl_bibset = ccl_qual_dup(base);
 261 }
 262
 263 yf::Zoom::Searchable::~Searchable()
 264 {
 265     ccl_qual_rm(&ccl_bibset);
 266 }
 267
 268 yf::Zoom::Frontend::Frontend(Impl *impl) :
 269     m_p(impl), m_is_virtual(false), m_in_use(true)
 270 {
 271 }
 272
 273 yf::Zoom::Frontend::~Frontend()
 274 {
 275 }
 276
 277 yf::Zoom::FrontendPtr yf::Zoom::Impl::get_frontend(mp::Package &package)
 278 {
 279     boost::mutex::scoped_lock lock(m_mutex);
 280
 281     std::map<mp::Session,yf::Zoom::FrontendPtr>::iterator it;
 282
 283     while(true)
 284     {
 285         it = m_clients.find(package.session());
 286         if (it == m_clients.end())
 287             break;
 288
 289         if (!it->second->m_in_use)
 290         {
 291             it->second->m_in_use = true;
 292             return it->second;
 293         }
 294         m_cond_session_ready.wait(lock);
 295     }
 296     FrontendPtr f(new Frontend(this));
 297     m_clients[package.session()] = f;
 298     f->m_in_use = true;
 299     return f;
 300 }
 301
 302 void yf::Zoom::Impl::release_frontend(mp::Package &package)
 303 {
 304     boost::mutex::scoped_lock lock(m_mutex);
 305     std::map<mp::Session,yf::Zoom::FrontendPtr>::iterator it;
 306
 307     it = m_clients.find(package.session());
 308     if (it != m_clients.end())
 309     {
 310         if (package.session().is_closed())
 311         {
 312             m_clients.erase(it);
 313         }
 314         else
 315         {
 316             it->second->m_in_use = false;
 317         }
 318         m_cond_session_ready.notify_all();
 319     }
 320 }
 321
 322 yf::Zoom::Impl::Impl() : element_transform("pz2") , element_raw("raw")
 323 {
 324     bibset = ccl_qual_mk();
 325 }
 326
 327 yf::Zoom::Impl::~Impl()
 328 {
 329     ccl_qual_rm(&bibset);
 330 }
 331
 332 yf::Zoom::SearchablePtr yf::Zoom::Impl::parse_torus_record(const xmlNode *ptr)
 333 {
 334     Zoom::SearchablePtr s(new Searchable(bibset));
 335
 336     for (ptr = ptr->children; ptr; ptr = ptr->next)
 337     {
 338         if (ptr->type != XML_ELEMENT_NODE)
 339             continue;
 340         if (!strcmp((const char *) ptr->name, "layer"))
 341             ptr = ptr->children;
 342         else if (!strcmp((const char *) ptr->name,
 343                          "authentication"))
 344         {
 345             s->authentication = mp::xml::get_text(ptr);
 346         }
 347         else if (!strcmp((const char *) ptr->name,
 348                          "cfAuth"))
 349         {
 350             s->cfAuth = mp::xml::get_text(ptr);
 351         }
 352         else if (!strcmp((const char *) ptr->name,
 353                          "cfProxy"))
 354         {
 355             s->cfProxy = mp::xml::get_text(ptr);
 356         }
 357         else if (!strcmp((const char *) ptr->name,
 358                          "cfSubDb"))
 359         {
 360             s->cfSubDb = mp::xml::get_text(ptr);
 361         }
 362         else if (!strcmp((const char *) ptr->name, "udb"))
 363         {
 364             s->udb = mp::xml::get_text(ptr);
 365         }
 366         else if (!strcmp((const char *) ptr->name, "zurl"))
 367         {
 368             s->target = mp::xml::get_text(ptr);
 369         }
 370         else if (!strcmp((const char *) ptr->name, "sru"))
 371         {
 372             s->sru = mp::xml::get_text(ptr);
 373         }
 374         else if (!strcmp((const char *) ptr->name,
 375                          "queryEncoding"))
 376         {
 377             s->query_encoding = mp::xml::get_text(ptr);
 378         }
 379         else if (!strcmp((const char *) ptr->name,
 380                          "piggyback"))
 381         {
 382             s->piggyback = mp::xml::get_bool(ptr, true);
 383         }
 384         else if (!strcmp((const char *) ptr->name,
 385                          "requestSyntax"))
 386         {
 387             s->request_syntax = mp::xml::get_text(ptr);
 388         }
 389         else if (!strcmp((const char *) ptr->name,
 390                          "elementSet"))
 391         {
 392             s->element_set = mp::xml::get_text(ptr);
 393         }
 394         else if (!strcmp((const char *) ptr->name,
 395                          "recordEncoding"))
 396         {
 397             s->record_encoding = mp::xml::get_text(ptr);
 398         }
 399         else if (!strcmp((const char *) ptr->name,
 400                          "transform"))
 401         {
 402             s->transform_xsl_fname = mp::xml::get_text(ptr);
 403         }
 404         else if (!strcmp((const char *) ptr->name,
 405                          "urlRecipe"))
 406         {
 407             s->urlRecipe = mp::xml::get_text(ptr);
 408         }
 409         else if (!strcmp((const char *) ptr->name,
 410                          "useTurboMarc"))
 411         {
 412             ; // useTurboMarc is ignored
 413         }
 414         else if (!strncmp((const char *) ptr->name,
 415                           "cclmap_", 7))
 416         {
 417             std::string value = mp::xml::get_text(ptr);
 418             ccl_qual_fitem(s->ccl_bibset, value.c_str(),
 419                            (const char *) ptr->name + 7);
 420         }
 421     }
 422     return s;
 423 }
 424
 425 void yf::Zoom::Impl::configure_local_records(const xmlNode *ptr, bool test_only)
 426 {
 427     while (ptr && ptr->type != XML_ELEMENT_NODE)
 428         ptr = ptr->next;
 429
 430     if (ptr)
 431     {
 432         if (!strcmp((const char *) ptr->name, "records"))
 433         {
 434             for (ptr = ptr->children; ptr; ptr = ptr->next)
 435             {
 436                 if (ptr->type != XML_ELEMENT_NODE)
 437                     continue;
 438                 if (!strcmp((const char *) ptr->name, "record"))
 439                 {
 440                     SearchablePtr s = parse_torus_record(ptr);
 441                     if (s)
 442                     {
 443                         std::string udb = s->udb;
 444                         if (udb.length())
 445                             s_map[s->udb] = s;
 446                         else
 447                         {
 448                             throw mp::filter::FilterException
 449                                 ("No udb for local torus record");
 450                         }
 451                     }
 452                 }
 453                 else
 454                 {
 455                     throw mp::filter::FilterException
 456                         ("Bad element "
 457                          + std::string((const char *) ptr->name)
 458                          + " in zoom filter inside element "
 459                          "<torus><records>");
 460                 }
 461             }
 462         }
 463         else
 464         {
 465             throw mp::filter::FilterException
 466                 ("Bad element "
 467                  + std::string((const char *) ptr->name)
 468                  + " in zoom filter inside element <torus>");
 469         }
 470     }
 471 }
 472
 473 void yf::Zoom::Impl::configure(const xmlNode *ptr, bool test_only)
 474 {
 475     for (ptr = ptr->children; ptr; ptr = ptr->next)
 476     {
 477         if (ptr->type != XML_ELEMENT_NODE)
 478             continue;
 479         else if (!strcmp((const char *) ptr->name, "torus"))
 480         {
 481             const struct _xmlAttr *attr;
 482             for (attr = ptr->properties; attr; attr = attr->next)
 483             {
 484                 if (!strcmp((const char *) attr->name, "url"))
 485                     torus_url = mp::xml::get_text(attr->children);
 486                 else if (!strcmp((const char *) attr->name, "xsldir"))
 487                     xsldir = mp::xml::get_text(attr->children);
 488                 else if (!strcmp((const char *) attr->name, "element_transform"))
 489                     element_transform = mp::xml::get_text(attr->children);
 490                 else if (!strcmp((const char *) attr->name, "element_raw"))
 491                     element_raw = mp::xml::get_text(attr->children);
 492                 else
 493                     throw mp::filter::FilterException(
 494                         "Bad attribute " + std::string((const char *)
 495                                                        attr->name));
 496             }
 497             configure_local_records(ptr->children, test_only);
 498         }
 499         else if (!strcmp((const char *) ptr->name, "cclmap"))
 500         {
 501             const char *addinfo = 0;
 502             ccl_xml_config(bibset, ptr, &addinfo);
 503         }
 504         else if (!strcmp((const char *) ptr->name, "fieldmap"))
 505         {
 506             const struct _xmlAttr *attr;
 507             std::string ccl_field;
 508             std::string cql_field;
 509             for (attr = ptr->properties; attr; attr = attr->next)
 510             {
 511                 if (!strcmp((const char *) attr->name, "ccl"))
 512                     ccl_field = mp::xml::get_text(attr->children);
 513                 else if (!strcmp((const char *) attr->name, "cql"))
 514                     cql_field = mp::xml::get_text(attr->children);
 515                 else
 516                     throw mp::filter::FilterException(
 517                         "Bad attribute " + std::string((const char *)
 518                                                        attr->name));
 519             }
 520             if (cql_field.length())
 521                 fieldmap[cql_field] = ccl_field;
 522         }
 523         else
 524         {
 525             throw mp::filter::FilterException
 526                 ("Bad element "
 527                  + std::string((const char *) ptr->name)
 528                  + " in zoom filter");
 529         }
 530     }
 531 }
 532
 533 yf::Zoom::BackendPtr yf::Zoom::Frontend::get_backend_from_databases(
 534     std::string &database, int *error, char **addinfo, ODR odr)
 535 {
 536     std::list<BackendPtr>::const_iterator map_it;
 537     if (m_backend && m_backend->m_frontend_database == database)
 538         return m_backend;
 539
 540     std::string db_args;
 541     std::string torus_db;
 542     size_t db_arg_pos = database.find(',');
 543     if (db_arg_pos != std::string::npos)
 544     {
 545         torus_db = database.substr(0, db_arg_pos);
 546         db_args = database.substr(db_arg_pos + 1);
 547     }
 548     else
 549         torus_db = database;
 550
 551     SearchablePtr sptr;
 552
 553     std::map<std::string,SearchablePtr>::iterator it;
 554     it = m_p->s_map.find(torus_db);
 555     if (it != m_p->s_map.end())
 556         sptr = it->second;
 557     else
 558     {
 559         xmlDoc *doc = mp::get_searchable(m_p->torus_url, torus_db);
 560         if (!doc)
 561         {
 562             *error = YAZ_BIB1_DATABASE_DOES_NOT_EXIST;
 563             *addinfo = odr_strdup(odr, database.c_str());
 564             BackendPtr b;
 565             return b;
 566         }
 567         const xmlNode *ptr = xmlDocGetRootElement(doc);
 568         if (ptr)
 569         {   // presumably ptr is a records element node
 570             // parse first record in document
 571             for (ptr = ptr->children; ptr; ptr = ptr->next)
 572             {
 573                 if (ptr->type == XML_ELEMENT_NODE
 574                     && !strcmp((const char *) ptr->name, "record"))
 575                 {
 576                     sptr = m_p->parse_torus_record(ptr);
 577                     break;
 578                 }
 579             }
 580         }
 581         xmlFreeDoc(doc);
 582     }
 583
 584     if (!sptr)
 585     {
 586         *error = YAZ_BIB1_DATABASE_DOES_NOT_EXIST;
 587         *addinfo = odr_strdup(odr, database.c_str());
 588         BackendPtr b;
 589         return b;
 590     }
 591
 592     xsltStylesheetPtr xsp = 0;
 593     if (sptr->transform_xsl_fname.length())
 594     {
 595         std::string fname;
 596
 597         if (m_p->xsldir.length())
 598             fname = m_p->xsldir + "/" + sptr->transform_xsl_fname;
 599         else
 600             fname = sptr->transform_xsl_fname;
 601         xmlDoc *xsp_doc = xmlParseFile(fname.c_str());
 602         if (!xsp_doc)
 603         {
 604             *error = YAZ_BIB1_TEMPORARY_SYSTEM_ERROR;
 605             *addinfo = (char *) odr_malloc(odr, 40 + strlen(fname.c_str()));
 606             sprintf(*addinfo, "xmlParseFile failed. File %s", fname.c_str());
 607             BackendPtr b;
 608             return b;
 609         }
 610         xsp = xsltParseStylesheetDoc(xsp_doc);
 611         if (!xsp)
 612         {
 613             *error = YAZ_BIB1_DATABASE_DOES_NOT_EXIST;
 614             *addinfo = odr_strdup(odr, "xsltParseStylesheetDoc failed");
 615             BackendPtr b;
 616             xmlFreeDoc(xsp_doc);
 617             return b;
 618         }
 619     }
 620
 621     m_backend.reset();
 622
 623     BackendPtr b(new Backend(sptr));
 624
 625     b->xsp = xsp;
 626     b->m_frontend_database = database;
 627     std::string authentication = sptr->authentication;
 628
 629     b->set_option("timeout", "40");
 630
 631     if (sptr->query_encoding.length())
 632         b->set_option("rpnCharset", sptr->query_encoding.c_str());
 633
 634     if (sptr->cfAuth.length())
 635     {
 636         // A CF target
 637         b->set_option("user", sptr->cfAuth.c_str());
 638         if (authentication.length() && db_args.length() == 0)
 639         {
 640             // no database (auth) args specified already.. and the
 641             // Torus authentication has it.. Generate the args that CF
 642             // understands..
 643             size_t found = authentication.find('/');
 644             if (found != std::string::npos)
 645             {
 646                 db_args += "user=" + mp::util::uri_encode(authentication.substr(0, found))
 647                     + "&password=" + mp::util::uri_encode(authentication.substr(found+1));
 648             }
 649             else
 650                 db_args += "user=" + mp::util::uri_encode(authentication);
 651         }
 652     }
 653     else
 654     {
 655         // A non-CF target
 656         if (db_args.length())
 657         {
 658             // user has specified backend authentication
 659             const char *param_user = 0;
 660             const char *param_password = 0;
 661             char **names;
 662             char **values;
 663             int i;
 664             int no_parms = yaz_uri_to_array(db_args.c_str(),
 665                                             odr, &names, &values);
 666             for (i = 0; i < no_parms; i++)
 667             {
 668                 const char *name = names[i];
 669                 const char *value = values[i];
 670                 if (!strcmp(name, "user"))
 671                     param_user = value;
 672                 else if (!strcmp(name, "password"))
 673                     param_password = value;
 674                 else
 675                 {
 676                     BackendPtr notfound;
 677                     char *msg = (char*) odr_malloc(odr, strlen(name) + 30);
 678                     *error = YAZ_BIB1_TEMPORARY_SYSTEM_ERROR;
 679                     sprintf(msg, "Bad database argument: %s", name);
 680                     *addinfo = msg;
 681                     return notfound;
 682                 }
 683             }
 684             if (param_user && param_password)
 685             {
 686                 char *auth = (char*) odr_malloc(
 687                     odr, strlen(param_user) + strlen(param_password) + 2);
 688                 strcpy(auth, param_user);
 689                 strcat(auth, "/");
 690                 strcat(auth, param_password);
 691                 b->set_option("user", auth);
 692             }
 693             db_args.clear(); // no arguments to be passed (non-CF)
 694         }
 695         else
 696         {
 697             // use authentication from Torus, if given
 698             if (authentication.length())
 699                 b->set_option("user", authentication.c_str());
 700         }
 701     }
 702     if (sptr->cfProxy.length())
 703     {
 704         if (db_args.length())
 705             db_args += "&";
 706         db_args += "proxy=" + mp::util::uri_encode(sptr->cfProxy);
 707     }
 708     if (sptr->cfSubDb.length())
 709     {
 710         if (db_args.length())
 711             db_args += "&";
 712         db_args += "subdatabase=" + mp::util::uri_encode(sptr->cfSubDb);
 713     }
 714
 715     std::string url;
 716     if (sptr->sru.length())
 717     {
 718         url = "http://" + sptr->target;
 719         b->set_option("sru", sptr->sru.c_str());
 720     }
 721     else
 722     {
 723         url = sptr->target;
 724     }
 725     if (db_args.length())
 726         url += "," + db_args;
 727     yaz_log(YLOG_LOG, "url=%s", url.c_str());
 728     const char *addinfo_c = 0;
 729     b->connect(url, error, &addinfo_c);
 730     if (addinfo_c)
 731         *addinfo = odr_strdup(odr, addinfo_c);
 732     if (*error == 0)
 733     {
 734         m_backend = b;
 735     }
 736     return b;
 737 }
 738
 739 Z_Records *yf::Zoom::Frontend::get_records(Odr_int start,
 740                                            Odr_int number_to_present,
 741                                            int *error,
 742                                            const char **addinfo,
 743                                            Odr_int *number_of_records_returned,
 744                                            ODR odr,
 745                                            BackendPtr b,
 746                                            Odr_oid *preferredRecordSyntax,
 747                                            const char *element_set_name)
 748 {
 749     *number_of_records_returned = 0;
 750     Z_Records *records = 0;
 751     bool enable_pz2_retrieval = false; // whether target profile is used
 752     bool enable_pz2_transform = false; // whether XSLT is used as well
 753     bool assume_marc8_charset = false;
 754
 755     if (start < 0 || number_to_present <= 0)
 756         return records;
 757
 758     if (number_to_present > 10000)
 759         number_to_present = 10000;
 760
 761     ZOOM_record *recs = (ZOOM_record *)
 762         odr_malloc(odr, number_to_present * sizeof(*recs));
 763
 764     char oid_name_str[OID_STR_MAX];
 765     const char *syntax_name = 0;
 766
 767     if (preferredRecordSyntax &&
 768         !oid_oidcmp(preferredRecordSyntax, yaz_oid_recsyn_xml)
 769         && element_set_name)
 770     {
 771         if (!strcmp(element_set_name, m_p->element_transform.c_str()))
 772         {
 773             enable_pz2_retrieval = true;
 774             enable_pz2_transform = true;
 775         }
 776         else if (!strcmp(element_set_name, m_p->element_raw.c_str()))
 777         {
 778             enable_pz2_retrieval = true;
 779         }
 780     }
 781
 782     if (enable_pz2_retrieval)
 783     {
 784         if (b->sptr->request_syntax.length())
 785         {
 786             syntax_name = b->sptr->request_syntax.c_str();
 787             const Odr_oid *syntax_oid =
 788                 yaz_string_to_oid(yaz_oid_std(), CLASS_RECSYN, syntax_name);
 789             if (!oid_oidcmp(syntax_oid, yaz_oid_recsyn_usmarc)
 790                 || !oid_oidcmp(syntax_oid, yaz_oid_recsyn_opac))
 791                 assume_marc8_charset = true;
 792         }
 793     }
 794     else if (preferredRecordSyntax)
 795         syntax_name =
 796             yaz_oid_to_string_buf(preferredRecordSyntax, 0, oid_name_str);
 797
 798     b->set_option("preferredRecordSyntax", syntax_name);
 799
 800     if (enable_pz2_retrieval)
 801     {
 802         element_set_name = 0;
 803         if (b->sptr->element_set.length())
 804             element_set_name = b->sptr->element_set.c_str();
 805     }
 806
 807     b->set_option("elementSetName", element_set_name);
 808
 809     b->present(start, number_to_present, recs, error, addinfo);
 810
 811     Odr_int i = 0;
 812     if (!*error)
 813     {
 814         for (i = 0; i < number_to_present; i++)
 815             if (!recs[i])
 816                 break;
 817     }
 818     if (i > 0)
 819     {  // only return records if no error and at least one record
 820         char *odr_database = odr_strdup(odr,
 821                                         b->m_frontend_database.c_str());
 822         Z_NamePlusRecordList *npl = (Z_NamePlusRecordList *)
 823             odr_malloc(odr, sizeof(*npl));
 824         *number_of_records_returned = i;
 825         npl->num_records = i;
 826         npl->records = (Z_NamePlusRecord **)
 827             odr_malloc(odr, i * sizeof(*npl->records));
 828         for (i = 0; i < number_to_present; i++)
 829         {
 830             Z_NamePlusRecord *npr = 0;
 831             const char *addinfo;
 832             int sur_error = ZOOM_record_error(recs[i], 0 /* msg */,
 833                                               &addinfo, 0 /* diagset */);
 834
 835             if (sur_error)
 836             {
 837                 npr = zget_surrogateDiagRec(odr, odr_database, sur_error,
 838                                             addinfo);
 839             }
 840             else if (enable_pz2_retrieval)
 841             {
 842                 char rec_type_str[100];
 843                 const char *record_encoding = 0;
 844
 845                 if (b->sptr->record_encoding.length())
 846                     record_encoding = b->sptr->record_encoding.c_str();
 847                 else if (assume_marc8_charset)
 848                     record_encoding = "marc8";
 849
 850                 strcpy(rec_type_str, b->sptr->use_turbomarc ? "txml" : "xml");
 851                 if (record_encoding)
 852                 {
 853                     strcat(rec_type_str, "; charset=");
 854                     strcat(rec_type_str, record_encoding);
 855                 }
 856
 857                 int rec_len;
 858                 const char *rec_buf = ZOOM_record_get(recs[i], rec_type_str,
 859                                                       &rec_len);
 860                 if (rec_buf && b->xsp && enable_pz2_transform)
 861                 {
 862                     xmlDoc *rec_doc = xmlParseMemory(rec_buf, rec_len);
 863                     if (rec_doc)
 864                     {
 865                         xmlDoc *rec_res;
 866                         rec_res = xsltApplyStylesheet(b->xsp, rec_doc, 0);
 867
 868                         if (rec_res)
 869                             xsltSaveResultToString((xmlChar **) &rec_buf, &rec_len,
 870                                                    rec_res, b->xsp);
 871                     }
 872                 }
 873
 874                 if (rec_buf)
 875                 {
 876                     xmlDoc *doc = xmlParseMemory(rec_buf, rec_len);
 877                     mp::xml::url_recipe_handle(doc, b->sptr->urlRecipe);
 878                     xmlFreeDoc(doc);
 879                 }
 880                 if (rec_buf)
 881                 {
 882                     npr = (Z_NamePlusRecord *) odr_malloc(odr, sizeof(*npr));
 883                     npr->databaseName = odr_database;
 884                     npr->which = Z_NamePlusRecord_databaseRecord;
 885                     npr->u.databaseRecord =
 886                         z_ext_record_xml(odr, rec_buf, rec_len);
 887                 }
 888                 else
 889                 {
 890                     npr = zget_surrogateDiagRec(
 891                         odr, odr_database,
 892                         YAZ_BIB1_SYSTEM_ERROR_IN_PRESENTING_RECORDS,
 893                         rec_type_str);
 894                 }
 895             }
 896             else
 897             {
 898                 Z_External *ext =
 899                     (Z_External *) ZOOM_record_get(recs[i], "ext", 0);
 900                 if (ext)
 901                 {
 902                     npr = (Z_NamePlusRecord *) odr_malloc(odr, sizeof(*npr));
 903                     npr->databaseName = odr_database;
 904                     npr->which = Z_NamePlusRecord_databaseRecord;
 905                     npr->u.databaseRecord = ext;
 906                 }
 907                 else
 908                 {
 909                     npr = zget_surrogateDiagRec(
 910                         odr, odr_database,
 911                         YAZ_BIB1_SYSTEM_ERROR_IN_PRESENTING_RECORDS,
 912                         "ZOOM_record, type ext");
 913                 }
 914             }
 915             npl->records[i] = npr;
 916         }
 917         records = (Z_Records*) odr_malloc(odr, sizeof(*records));
 918         records->which = Z_Records_DBOSD;
 919         records->u.databaseOrSurDiagnostics = npl;
 920     }
 921     return records;
 922 }
 923
 924 struct cql_node *yf::Zoom::Impl::convert_cql_fields(struct cql_node *cn,
 925                                                     ODR odr)
 926 {
 927     struct cql_node *r = 0;
 928     if (!cn)
 929         return 0;
 930     switch (cn->which)
 931     {
 932     case CQL_NODE_ST:
 933         if (cn->u.st.index)
 934         {
 935             std::map<std::string,std::string>::const_iterator it;
 936             it = fieldmap.find(cn->u.st.index);
 937             if (it == fieldmap.end())
 938                 return cn;
 939             if (it->second.length())
 940                 cn->u.st.index = odr_strdup(odr, it->second.c_str());
 941             else
 942                 cn->u.st.index = 0;
 943         }
 944         break;
 945     case CQL_NODE_BOOL:
 946         r = convert_cql_fields(cn->u.boolean.left, odr);
 947         if (!r)
 948             r = convert_cql_fields(cn->u.boolean.right, odr);
 949         break;
 950     case CQL_NODE_SORT:
 951         r = convert_cql_fields(cn->u.sort.search, odr);
 952         break;
 953     }
 954     return r;
 955 }
 956
 957 void yf::Zoom::Frontend::handle_search(mp::Package &package)
 958 {
 959     Z_GDU *gdu = package.request().get();
 960     Z_APDU *apdu_req = gdu->u.z3950;
 961     Z_APDU *apdu_res = 0;
 962     mp::odr odr;
 963     Z_SearchRequest *sr = apdu_req->u.searchRequest;
 964     if (sr->num_databaseNames != 1)
 965     {
 966         apdu_res = odr.create_searchResponse(
 967             apdu_req, YAZ_BIB1_TOO_MANY_DATABASES_SPECIFIED, 0);
 968         package.response() = apdu_res;
 969         return;
 970     }
 971
 972     int error = 0;
 973     char *addinfo_s = 0;
 974     std::string db(sr->databaseNames[0]);
 975     BackendPtr b = get_backend_from_databases(db, &error, &addinfo_s, odr);
 976     if (error)
 977     {
 978         apdu_res =
 979             odr.create_searchResponse(
 980                 apdu_req, error, addinfo_s);
 981         package.response() = apdu_res;
 982         return;
 983     }
 984
 985     const char *addinfo_c = 0;
 986     b->set_option("setname", "default");
 987
 988     Odr_int hits = 0;
 989     Z_Query *query = sr->query;
 990     WRBUF ccl_wrbuf = 0;
 991     WRBUF pqf_wrbuf = 0;
 992
 993     if (query->which == Z_Query_type_1 || query->which == Z_Query_type_101)
 994     {
 995         // RPN
 996         pqf_wrbuf = wrbuf_alloc();
 997         yaz_rpnquery_to_wrbuf(pqf_wrbuf, query->u.type_1);
 998     }
 999     else if (query->which == Z_Query_type_2)
1000     {
1001         // CCL
1002         ccl_wrbuf = wrbuf_alloc();
1003         wrbuf_write(ccl_wrbuf, (const char *) query->u.type_2->buf,
1004                     query->u.type_2->len);
1005     }
1006     else if (query->which == Z_Query_type_104 &&
1007              query->u.type_104->which == Z_External_CQL)
1008     {
1009         // CQL
1010         const char *cql = query->u.type_104->u.cql;
1011         CQL_parser cp = cql_parser_create();
1012         int r = cql_parser_string(cp, cql);
1013         if (r)
1014         {
1015             cql_parser_destroy(cp);
1016             apdu_res =
1017                 odr.create_searchResponse(apdu_req,
1018                                           YAZ_BIB1_MALFORMED_QUERY,
1019                                           "CQL syntax error");
1020             package.response() = apdu_res;
1021             return;
1022         }
1023         struct cql_node *cn = cql_parser_result(cp);
1024         struct cql_node *cn_error = m_p->convert_cql_fields(cn, odr);
1025         if (cn_error)
1026         {
1027             // hopefully we are getting a ptr to a index+relation+term node
1028             addinfo_c = 0;
1029             if (cn_error->which == CQL_NODE_ST)
1030                 addinfo_c = cn_error->u.st.index;
1031
1032             apdu_res =
1033                 odr.create_searchResponse(apdu_req,
1034                                           YAZ_BIB1_UNSUPP_USE_ATTRIBUTE,
1035                                           addinfo_c);
1036             package.response() = apdu_res;
1037             return;
1038         }
1039         char ccl_buf[1024];
1040
1041         r = cql_to_ccl_buf(cn, ccl_buf, sizeof(ccl_buf));
1042         if (r == 0)
1043         {
1044             ccl_wrbuf = wrbuf_alloc();
1045             wrbuf_puts(ccl_wrbuf, ccl_buf);
1046         }
1047         cql_parser_destroy(cp);
1048         if (r)
1049         {
1050             apdu_res =
1051                 odr.create_searchResponse(apdu_req,
1052                                           YAZ_BIB1_MALFORMED_QUERY,
1053                                           "CQL to CCL conversion error");
1054             package.response() = apdu_res;
1055             return;
1056         }
1057     }
1058     else
1059     {
1060         apdu_res =
1061             odr.create_searchResponse(apdu_req, YAZ_BIB1_QUERY_TYPE_UNSUPP, 0);
1062         package.response() = apdu_res;
1063         return;
1064     }
1065
1066     if (ccl_wrbuf)
1067     {
1068         // CCL to PQF
1069         assert(pqf_wrbuf == 0);
1070         int cerror, cpos;
1071         struct ccl_rpn_node *cn;
1072         yaz_log(YLOG_LOG, "CCL: %s", wrbuf_cstr(ccl_wrbuf));
1073         cn = ccl_find_str(b->sptr->ccl_bibset, wrbuf_cstr(ccl_wrbuf),
1074                           &cerror, &cpos);
1075         wrbuf_destroy(ccl_wrbuf);
1076         if (!cn)
1077         {
1078             char *addinfo = odr_strdup(odr, ccl_err_msg(cerror));
1079             int z3950_diag = YAZ_BIB1_MALFORMED_QUERY;
1080
1081             switch (cerror)
1082             {
1083             case CCL_ERR_UNKNOWN_QUAL:
1084                 z3950_diag = YAZ_BIB1_UNSUPP_USE_ATTRIBUTE;
1085                 break;
1086             case CCL_ERR_TRUNC_NOT_LEFT:
1087             case CCL_ERR_TRUNC_NOT_RIGHT:
1088             case CCL_ERR_TRUNC_NOT_BOTH:
1089                 z3950_diag = YAZ_BIB1_UNSUPP_TRUNCATION_ATTRIBUTE;
1090                 break;
1091             }
1092             apdu_res =
1093                 odr.create_searchResponse(apdu_req, z3950_diag, addinfo);
1094             package.response() = apdu_res;
1095             return;
1096         }
1097         pqf_wrbuf = wrbuf_alloc();
1098         ccl_pquery(pqf_wrbuf, cn);
1099         ccl_rpn_delete(cn);
1100     }
1101
1102     assert(pqf_wrbuf);
1103     if (b->get_option("sru"))
1104     {
1105         int status = 0;
1106         Z_RPNQuery *zquery;
1107         zquery = p_query_rpn(odr, wrbuf_cstr(pqf_wrbuf));
1108         WRBUF wrb = wrbuf_alloc();
1109
1110         if (!strcmp(b->get_option("sru"), "solr"))
1111         {
1112             solr_transform_t cqlt = solr_transform_create();
1113
1114             status = solr_transform_rpn2solr_wrbuf(cqlt, wrb, zquery);
1115
1116             solr_transform_close(cqlt);
1117         }
1118         else
1119         {
1120             cql_transform_t cqlt = cql_transform_create();
1121
1122             status = cql_transform_rpn2cql_wrbuf(cqlt, wrb, zquery);
1123
1124             cql_transform_close(cqlt);
1125         }
1126         if (status == 0)
1127         {
1128             yaz_log(YLOG_LOG, "search CQL: %s", wrbuf_cstr(wrb));
1129             b->search_cql(wrbuf_cstr(wrb), &hits, &error, &addinfo_c);
1130         }
1131
1132         wrbuf_destroy(wrb);
1133         wrbuf_destroy(pqf_wrbuf);
1134         if (status)
1135         {
1136             apdu_res =
1137                 odr.create_searchResponse(apdu_req, YAZ_BIB1_MALFORMED_QUERY,
1138                                           "can not convert from RPN to CQL/SOLR");
1139             package.response() = apdu_res;
1140             return;
1141         }
1142     }
1143     else
1144     {
1145         yaz_log(YLOG_LOG, "search PQF: %s", wrbuf_cstr(pqf_wrbuf));
1146         b->search_pqf(wrbuf_cstr(pqf_wrbuf), &hits, &error, &addinfo_c);
1147         wrbuf_destroy(pqf_wrbuf);
1148     }
1149
1150
1151     const char *element_set_name = 0;
1152     Odr_int number_to_present = 0;
1153     if (!error)
1154         mp::util::piggyback_sr(sr, hits, number_to_present, &element_set_name);
1155
1156     Odr_int number_of_records_returned = 0;
1157     Z_Records *records = get_records(
1158         0, number_to_present, &error, &addinfo_c,
1159         &number_of_records_returned, odr, b, sr->preferredRecordSyntax,
1160         element_set_name);
1161     apdu_res = odr.create_searchResponse(apdu_req, error, addinfo_c);
1162     if (records)
1163     {
1164         apdu_res->u.searchResponse->records = records;
1165         apdu_res->u.searchResponse->numberOfRecordsReturned =
1166             odr_intdup(odr, number_of_records_returned);
1167     }
1168     apdu_res->u.searchResponse->resultCount = odr_intdup(odr, hits);
1169     package.response() = apdu_res;
1170 }
1171
1172 void yf::Zoom::Frontend::handle_present(mp::Package &package)
1173 {
1174     Z_GDU *gdu = package.request().get();
1175     Z_APDU *apdu_req = gdu->u.z3950;
1176     Z_APDU *apdu_res = 0;
1177     Z_PresentRequest *pr = apdu_req->u.presentRequest;
1178
1179     mp::odr odr;
1180     if (!m_backend)
1181     {
1182         package.response() = odr.create_presentResponse(
1183             apdu_req, YAZ_BIB1_SPECIFIED_RESULT_SET_DOES_NOT_EXIST, 0);
1184         return;
1185     }
1186     const char *element_set_name = 0;
1187     Z_RecordComposition *comp = pr->recordComposition;
1188     if (comp && comp->which != Z_RecordComp_simple)
1189     {
1190         package.response() = odr.create_presentResponse(
1191             apdu_req,
1192             YAZ_BIB1_PRESENT_COMP_SPEC_PARAMETER_UNSUPP, 0);
1193         return;
1194     }
1195     if (comp && comp->u.simple->which == Z_ElementSetNames_generic)
1196         element_set_name = comp->u.simple->u.generic;
1197     Odr_int number_of_records_returned = 0;
1198     int error = 0;
1199     const char *addinfo = 0;
1200     Z_Records *records = get_records(
1201         *pr->resultSetStartPoint - 1, *pr->numberOfRecordsRequested,
1202         &error, &addinfo, &number_of_records_returned, odr, m_backend,
1203         pr->preferredRecordSyntax, element_set_name);
1204
1205     apdu_res = odr.create_presentResponse(apdu_req, error, addinfo);
1206     if (records)
1207     {
1208         apdu_res->u.presentResponse->records = records;
1209         apdu_res->u.presentResponse->numberOfRecordsReturned =
1210             odr_intdup(odr, number_of_records_returned);
1211     }
1212     package.response() = apdu_res;
1213 }
1214
1215 void yf::Zoom::Frontend::handle_package(mp::Package &package)
1216 {
1217     Z_GDU *gdu = package.request().get();
1218     if (!gdu)
1219         ;
1220     else if (gdu->which == Z_GDU_Z3950)
1221     {
1222         Z_APDU *apdu_req = gdu->u.z3950;
1223         if (apdu_req->which == Z_APDU_initRequest)
1224         {
1225             mp::odr odr;
1226             package.response() = odr.create_close(
1227                 apdu_req,
1228                 Z_Close_protocolError,
1229                 "double init");
1230         }
1231         else if (apdu_req->which == Z_APDU_searchRequest)
1232         {
1233             handle_search(package);
1234         }
1235         else if (apdu_req->which == Z_APDU_presentRequest)
1236         {
1237             handle_present(package);
1238         }
1239         else
1240         {
1241             mp::odr odr;
1242             package.response() = odr.create_close(
1243                 apdu_req,
1244                 Z_Close_protocolError,
1245                 "zoom filter cannot handle this APDU");
1246             package.session().close();
1247         }
1248     }
1249     else
1250     {
1251         package.session().close();
1252     }
1253 }
1254
1255 void yf::Zoom::Impl::process(mp::Package &package)
1256 {
1257     FrontendPtr f = get_frontend(package);
1258     Z_GDU *gdu = package.request().get();
1259
1260     if (f->m_is_virtual)
1261     {
1262         f->handle_package(package);
1263     }
1264     else if (gdu && gdu->which == Z_GDU_Z3950 && gdu->u.z3950->which ==
1265              Z_APDU_initRequest)
1266     {
1267         Z_InitRequest *req = gdu->u.z3950->u.initRequest;
1268         f->m_init_gdu = gdu;
1269
1270         mp::odr odr;
1271         Z_APDU *apdu = odr.create_initResponse(gdu->u.z3950, 0, 0);
1272         Z_InitResponse *resp = apdu->u.initResponse;
1273
1274         int i;
1275         static const int masks[] = {
1276             Z_Options_search,
1277             Z_Options_present,
1278             -1
1279         };
1280         for (i = 0; masks[i] != -1; i++)
1281             if (ODR_MASK_GET(req->options, masks[i]))
1282                 ODR_MASK_SET(resp->options, masks[i]);
1283
1284         static const int versions[] = {
1285             Z_ProtocolVersion_1,
1286             Z_ProtocolVersion_2,
1287             Z_ProtocolVersion_3,
1288             -1
1289         };
1290         for (i = 0; versions[i] != -1; i++)
1291             if (ODR_MASK_GET(req->protocolVersion, versions[i]))
1292                 ODR_MASK_SET(resp->protocolVersion, versions[i]);
1293             else
1294                 break;
1295
1296         *resp->preferredMessageSize = *req->preferredMessageSize;
1297         *resp->maximumRecordSize = *req->maximumRecordSize;
1298
1299         package.response() = apdu;
1300         f->m_is_virtual = true;
1301     }
1302     else
1303         package.move();
1304
1305     release_frontend(package);
1306 }
1307
1308
1309 static mp::filter::Base* filter_creator()
1310 {
1311     return new mp::filter::Zoom;
1312 }
1313
1314 extern "C" {
1315     struct metaproxy_1_filter_struct metaproxy_1_filter_zoom = {
1316         0,
1317         "zoom",
1318         filter_creator
1319     };
1320 }
1321
1322
1323 /*
1324  * Local variables:
1325  * c-basic-offset: 4
1326  * c-file-style: "Stroustrup"
1327  * indent-tabs-mode: nil
1328  * End:
1329  * vim: shiftwidth=4 tabstop=8 expandtab
1330  */
1331