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