src/metaproxy_filter_xquery.cpp

   1 /* This file is part of mp-xquery
   2    Copyright (C) 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 <metaproxy/package.hpp>
  20 #include <metaproxy/util.hpp>
  21 #include <yaz/log.h>
  22 #include <yaz/oid_db.h>
  23 #include <map>
  24
  25 #include <iostream>
  26 #include <fstream>
  27 #include <sstream>
  28
  29 #include <zorba/zorba.h>
  30 #include <zorba/store_manager.h>
  31 #include <zorba/serializer.h>
  32 #include <zorba/singleton_item_sequence.h>
  33 #include <zorba/zorba_exception.h>
  34
  35
  36 namespace mp = metaproxy_1;
  37 namespace yf = mp::filter;
  38 namespace mp_util = metaproxy_1::util;
  39 using namespace mp;
  40 using namespace zorba;
  41
  42 namespace metaproxy_1 {
  43     namespace filter {
  44         class XQuery : public Base {
  45         public:
  46             ~XQuery();
  47             XQuery();
  48             void process(metaproxy_1::Package & package) const;
  49             void configure(const xmlNode * ptr, bool test_only,
  50                            const char *path);
  51             void start() const;
  52             void stop(int signo) const;
  53         private:
  54             bool convert_one_record(const char *input_buf,
  55                                     size_t input_len,
  56                                     std::string &result) const;
  57             std::map<std::string, std::string> zorba_variables;
  58             std::string zorba_filename;
  59             std::string zorba_script;
  60             std::string zorba_record_variable;
  61             Zorba *lZorba;
  62             XQuery_t lQuery;
  63         };
  64     }
  65 }
  66
  67 yf::XQuery::XQuery()
  68 {
  69     lZorba = 0;
  70 }
  71
  72 yf::XQuery::~XQuery()
  73 {
  74     if (lZorba)
  75         lZorba->shutdown();
  76 }
  77
  78 void yf::XQuery::start() const
  79 {
  80 }
  81
  82 void yf::XQuery::stop(int signo) const
  83 {
  84 }
  85
  86 bool yf::XQuery::convert_one_record(const char *input_buf,
  87                                     size_t input_len,
  88                                     std::string &result) const
  89 {
  90     XQuery_t tQuery = lQuery->clone();
  91
  92     zorba::DynamicContext* lDynamicContext = tQuery->getDynamicContext();
  93
  94     zorba::Item lItem;
  95     std::map<std::string, std::string>::const_iterator it;
  96     for (it = zorba_variables.begin(); it != zorba_variables.end(); it++)
  97     {
  98         lItem = lZorba->getItemFactory()->createString(it->second);
  99         lDynamicContext->setVariable(it->first, lItem);
 100     }
 101     std::string rec_content = "raw:" + std::string(input_buf, input_len);
 102     lItem = lZorba->getItemFactory()->createString(rec_content);
 103     lDynamicContext->setVariable(zorba_record_variable, lItem);
 104
 105     std::stringstream ss;
 106     tQuery->execute(ss);
 107     result = ss.str();
 108     return true;
 109 }
 110
 111 void yf::XQuery::process(Package &package) const
 112 {
 113     Z_GDU *gdu_req = package.request().get();
 114     Z_PresentRequest *pr_req = 0;
 115     Z_SearchRequest *sr_req = 0;
 116
 117     const char *input_schema = 0;
 118     Odr_oid *input_syntax = 0;
 119
 120     if (gdu_req && gdu_req->which == Z_GDU_Z3950 &&
 121         gdu_req->u.z3950->which == Z_APDU_presentRequest)
 122     {
 123         pr_req = gdu_req->u.z3950->u.presentRequest;
 124
 125         input_schema =
 126             mp_util::record_composition_to_esn(pr_req->recordComposition);
 127         input_syntax = pr_req->preferredRecordSyntax;
 128     }
 129     else if (gdu_req && gdu_req->which == Z_GDU_Z3950 &&
 130              gdu_req->u.z3950->which == Z_APDU_searchRequest)
 131     {
 132         sr_req = gdu_req->u.z3950->u.searchRequest;
 133
 134         input_syntax = sr_req->preferredRecordSyntax;
 135
 136         // we don't know how many hits we're going to get and therefore
 137         // the effective element set name.. Therefore we can only allow
 138         // two cases.. Both equal or absent.. If not, we'll just have to
 139         // disable the piggyback!
 140         if (sr_req->smallSetElementSetNames
 141             &&
 142             sr_req->mediumSetElementSetNames
 143             &&
 144             sr_req->smallSetElementSetNames->which == Z_ElementSetNames_generic
 145             &&
 146             sr_req->mediumSetElementSetNames->which == Z_ElementSetNames_generic
 147             &&
 148             !strcmp(sr_req->smallSetElementSetNames->u.generic,
 149                     sr_req->mediumSetElementSetNames->u.generic))
 150         {
 151             input_schema = sr_req->smallSetElementSetNames->u.generic;
 152         }
 153         else if (!sr_req->smallSetElementSetNames &&
 154                  !sr_req->mediumSetElementSetNames)
 155             ; // input_schema is 0 already
 156         else
 157         {
 158             // disable piggyback (perhaps it was disabled already)
 159             *sr_req->smallSetUpperBound = 0;
 160             *sr_req->largeSetLowerBound = 0;
 161             *sr_req->mediumSetPresentNumber = 0;
 162             package.move();
 163             return;
 164         }
 165         // we can handle it in record_transform.
 166     }
 167     else
 168     {
 169         package.move();
 170         return;
 171     }
 172
 173     mp::odr odr_en(ODR_ENCODE);
 174
 175     const char *backend_schema = 0;
 176     const Odr_oid *backend_syntax = 0;
 177
 178     if (input_schema && !strcmp(input_schema, "bibframe") &&
 179         (!input_syntax || !oid_oidcmp(input_syntax, yaz_oid_recsyn_xml)))
 180     {
 181         backend_schema = "marcxml";
 182         backend_syntax = yaz_oid_recsyn_xml;
 183     }
 184     else
 185     {
 186         package.move();
 187         return;
 188     }
 189
 190     if (sr_req)
 191     {
 192         if (backend_syntax)
 193             sr_req->preferredRecordSyntax = odr_oiddup(odr_en, backend_syntax);
 194         else
 195             sr_req->preferredRecordSyntax = 0;
 196         if (backend_schema)
 197         {
 198             sr_req->smallSetElementSetNames
 199                 = (Z_ElementSetNames *)
 200                 odr_malloc(odr_en, sizeof(Z_ElementSetNames));
 201             sr_req->smallSetElementSetNames->which = Z_ElementSetNames_generic;
 202             sr_req->smallSetElementSetNames->u.generic
 203                 = odr_strdup(odr_en, backend_schema);
 204             sr_req->mediumSetElementSetNames = sr_req->smallSetElementSetNames;
 205         }
 206         else
 207         {
 208             sr_req->smallSetElementSetNames = 0;
 209             sr_req->mediumSetElementSetNames = 0;
 210         }
 211     }
 212     else if (pr_req)
 213     {
 214         if (backend_syntax)
 215             pr_req->preferredRecordSyntax = odr_oiddup(odr_en, backend_syntax);
 216         else
 217             pr_req->preferredRecordSyntax = 0;
 218
 219         if (backend_schema)
 220         {
 221             pr_req->recordComposition
 222                 = (Z_RecordComposition *)
 223                 odr_malloc(odr_en, sizeof(Z_RecordComposition));
 224             pr_req->recordComposition->which
 225                 = Z_RecordComp_simple;
 226             pr_req->recordComposition->u.simple
 227                 = (Z_ElementSetNames *)
 228                 odr_malloc(odr_en, sizeof(Z_ElementSetNames));
 229             pr_req->recordComposition->u.simple->which = Z_ElementSetNames_generic;
 230             pr_req->recordComposition->u.simple->u.generic
 231                 = odr_strdup(odr_en, backend_schema);
 232         }
 233         else
 234             pr_req->recordComposition = 0;
 235     }
 236     package.move();
 237
 238     Z_GDU *gdu_res = package.response().get();
 239
 240     // see if we have a records list to patch!
 241     Z_NamePlusRecordList *records = 0;
 242     if (gdu_res && gdu_res->which == Z_GDU_Z3950 &&
 243         gdu_res->u.z3950->which == Z_APDU_presentResponse)
 244     {
 245         Z_PresentResponse * pr_res = gdu_res->u.z3950->u.presentResponse;
 246
 247         if (pr_res
 248             && pr_res->numberOfRecordsReturned
 249             && *(pr_res->numberOfRecordsReturned) > 0
 250             && pr_res->records
 251             && pr_res->records->which == Z_Records_DBOSD)
 252         {
 253             records = pr_res->records->u.databaseOrSurDiagnostics;
 254         }
 255     }
 256     if (gdu_res && gdu_res->which == Z_GDU_Z3950 &&
 257         gdu_res->u.z3950->which == Z_APDU_searchResponse)
 258     {
 259         Z_SearchResponse *sr_res = gdu_res->u.z3950->u.searchResponse;
 260
 261         if (sr_res
 262             && sr_res->numberOfRecordsReturned
 263             && *(sr_res->numberOfRecordsReturned) > 0
 264             && sr_res->records
 265             && sr_res->records->which == Z_Records_DBOSD)
 266         {
 267             records = sr_res->records->u.databaseOrSurDiagnostics;
 268         }
 269     }
 270     if (records)
 271     {
 272         int i;
 273         for (i = 0; i < records->num_records; i++)
 274         {
 275             Z_NamePlusRecord *npr = records->records[i];
 276             if (npr->which == Z_NamePlusRecord_databaseRecord)
 277             {
 278                 const char *details = 0;
 279                 Z_External *r = npr->u.databaseRecord;
 280                 int ret_trans = -1;
 281                 if (r->which == Z_External_octet &&
 282                     !oid_oidcmp(r->direct_reference, yaz_oid_recsyn_xml))
 283                 {
 284                     std::string result;
 285                     if (convert_one_record(
 286                         r->u.octet_aligned->buf, r->u.octet_aligned->len,
 287                         result))
 288                     {
 289                         npr->u.databaseRecord =
 290                             z_ext_record_oid(odr_en, yaz_oid_recsyn_xml,
 291                                              result.c_str(),
 292                                              result.length());
 293                     }
 294                 }
 295             }
 296         }
 297         package.response() = gdu_res;
 298     }
 299 }
 300
 301 void yf::XQuery::configure(const xmlNode * ptr, bool test_only,
 302                            const char *path)
 303 {
 304     for (ptr = ptr->children; ptr; ptr = ptr->next)
 305     {
 306         if (ptr->type != XML_ELEMENT_NODE)
 307             continue;
 308         if (!strcmp((const char *) ptr->name, "setVariable"))
 309         {
 310             std::string name;
 311             std::string value;
 312             struct _xmlAttr *attr;
 313             for (attr = ptr->properties; attr; attr = attr->next)
 314                 if (!strcmp((const char *) attr->name, "name"))
 315                     name = mp::xml::get_text(attr->children);
 316                 else if (!strcmp((const char *) attr->name, "value"))
 317                     value = mp::xml::get_text(attr->children);
 318                 else
 319                     throw mp::filter::FilterException(
 320                         "Bad attribute " + std::string((const char *)
 321                                                        attr->name));
 322             if (name.length() > 0)
 323                 zorba_variables[name] = value;
 324         }
 325         else if (!strcmp((const char *) ptr->name, "filename"))
 326         {
 327             std::string value;
 328             struct _xmlAttr *attr;
 329             for (attr = ptr->properties; attr; attr = attr->next)
 330                 if (!strcmp((const char *) attr->name, "value"))
 331                     value = mp::xml::get_text(attr->children);
 332                 else
 333                     throw mp::filter::FilterException(
 334                         "Bad attribute " + std::string((const char *)
 335                                                        attr->name));
 336             zorba_filename = value;
 337         }
 338         else if (!strcmp((const char *) ptr->name, "script"))
 339         {
 340             std::string value;
 341             struct _xmlAttr *attr;
 342             for (attr = ptr->properties; attr; attr = attr->next)
 343                 if (!strcmp((const char *) attr->name, "value"))
 344                     value = mp::xml::get_text(attr->children);
 345                 else
 346                     throw mp::filter::FilterException(
 347                         "Bad attribute " + std::string((const char *)
 348                                                        attr->name));
 349             zorba_script = value;
 350         }
 351         else if (!strcmp((const char *) ptr->name, "record"))
 352         {
 353             std::string value;
 354             struct _xmlAttr *attr;
 355             for (attr = ptr->properties; attr; attr = attr->next)
 356                 if (!strcmp((const char *) attr->name, "value"))
 357                     value = mp::xml::get_text(attr->children);
 358                 else
 359                     throw mp::filter::FilterException(
 360                         "Bad attribute " + std::string((const char *)
 361                                                        attr->name));
 362             zorba_record_variable = value;
 363         }
 364         else
 365         {
 366             throw mp::filter::FilterException("Bad element "
 367                                                + std::string((const char *)
 368                                                              ptr->name));
 369         }
 370     }
 371     if (zorba_script.length() == 0)
 372         throw mp::filter::FilterException("Missing element script");
 373     if (zorba_record_variable.length() == 0)
 374         throw mp::filter::FilterException("Missing element record");
 375     if (zorba_filename.length() == 0)
 376         throw mp::filter::FilterException("Missing element filename");
 377     if (!test_only)
 378     {
 379         void* lStore = StoreManager::getStore();
 380         lZorba = Zorba::getInstance(lStore);
 381
 382         lQuery = lZorba->createQuery();
 383
 384         lQuery->setFileName(zorba_filename);
 385
 386         std::unique_ptr<std::istream> qfile;
 387         qfile.reset(new std::ifstream(zorba_script.c_str()));
 388
 389         Zorba_CompilerHints lHints;
 390         lQuery->compile(*qfile, lHints);
 391     }
 392 }
 393
 394 static yf::Base* filter_creator()
 395 {
 396     return new mp::filter::XQuery;
 397 }
 398
 399 extern "C" {
 400     struct metaproxy_1_filter_struct metaproxy_1_filter_xquery = {
 401         0,
 402         "xquery",
 403         filter_creator
 404     };
 405 }
 406
 407
 408 /*
 409  * Local variables:
 410  * c-basic-offset: 4
 411  * c-file-style: "Stroustrup"
 412  * indent-tabs-mode: nil
 413  * End:
 414  * vim: shiftwidth=4 tabstop=8 expandtab
 415  */
 416