Changeset 1347 for trunk/gsdl/src/recpt/z3950proto.cpp

Timestamp:

2000-08-03T17:21:26+12:00 (24 years ago)

Author:

jrm21

Message:

merged z39.50 receptionist stuff into main trunk (along with the mgpp stuff)

File:

• trunk/gsdl/src/recpt/z3950proto.cpp (modified) (10 diffs)

trunk/gsdl/src/recpt/z3950proto.cpp

@@ -26 +26 @@
 #include "z3950proto.h"
 #include "comtypes.h"
-#include <stdio.h>
-
-/*** 
-    We will use the "collection" argument as the name of the database
-    to query on the z3950 server.
-***/
-
-
-z3950_server::z3950_server() {
-  count=0;
-  info=NULL;
-  meta["iconcollection"]="/~johnmcp/gsdl/images/zserver.png";
-}
-
-
-
-
-// z3950 Protocol functions
+#include "browsefilter.h"
+#include "queryfilter.h"
+#include "filter.h"
+// z39.50 yaz stuff
+#include <stdio.h> // for (FILE *) type for yyin and fopen.
+// config file parsing stuff
+#include "z3950cfg.h" // for reading in config files - 
+// defines "struct z3950cfg *zserver_list" as the head of the list.
+extern FILE *yyin;
+extern "C" {
+  extern int zconfigparse();
+}
+
+
+
+
 z3950proto::z3950proto() {
   zserver_count=0;
 }
 
-void z3950proto::add_server (const z3950_server& zserver) {
+z3950proto::~z3950proto() {
+}
+
+void z3950proto::add_server (z3950_server& zserver) {
 
   // append the new server
   zserver_count++;
-  zservers.push_back(zserver);
-}
-
-void z3950proto::configure (const text_t &/*key*/, 
-                const text_tarray &/*cfgline*/) {
+  zservers.push_back(&zserver);
+}
+
+void z3950proto::read_config_file(const text_t &filename) {
+  struct z3950cfg *here;
+  struct z3950cfg *oldhere;
+  z3950_server *zserver;
+  ShortColInfo_t *tempinfo;
+
+  // zconfigparse() is defined in zparse.tab.c,
+  // which is the bison output of zparse.y
+
+  yyin=fopen(filename.getcstr(),"r");
+  if (yyin==NULL) {
+    cerr << "Could not open "<<filename.getcstr()<<" for reading.\n";
+  }
+  zconfigparse();
+  // we now have the config files in the ptr zserver_list
+  if (zserver_list==NULL)
+    return; // no valid servers found in the config file - note that
+  // the parser will have already spat out any errors.
+
+  // now create z3950servers for each structure in server_list
+  here=zserver_list;
+  while (here!=NULL) {
+    zserver=new z3950_server;
+    tempinfo=new ShortColInfo_t;
+
+    tempinfo->host.setcstr(here->hostname);
+    tempinfo->port=here->port;
+    tempinfo->name.setcstr(here->dbname);
+    zserver->setInfo(tempinfo);
+    zserver->setName(here->shortname);
+    // now collection metadata.
+    zserver->setMeta("collectionname",here->longname);
+    if (here->icon!=NULL)
+      zserver->setMeta("iconcollection",here->icon);
+    if (here->smallicon!=NULL)
+      zserver->setMeta("iconcollectionsmall",here->smallicon);
+    
+    /*    filterclass *filter = new filterclass ();
+      zserver->add_filter (filter);
+      browsefilterclass *browsefilter = new browsefilterclass();
+      zserver->add_filter (browsefilter);
+      queryfilterclass *queryfilter = new queryfilterclass();
+      zserver->add_filter (queryfilter);
+    */
+
+    // About list
+    if (here->about!=NULL) {
+      struct z3950aboutlist *about_here=here->about;
+      struct z3950aboutlist *oldabout;
+
+      while (about_here!=NULL) {
+    // problem with default lang (null): can't add ("",..)
+    if (about_here->lang==NULL)
+      zserver->addcfgAbout("en",about_here->text);
+    else
+      zserver->addcfgAbout(about_here->lang, about_here->text);
+    oldabout=about_here;
+    about_here=about_here->next;
+    free(oldabout->lang);
+    free(oldabout->text);
+    free(oldabout);
+      }
+    }
+
+    oldhere=here;
+    here=here->next;
+    free(oldhere->shortname);
+    free(oldhere->hostname);
+    free(oldhere->dbname);
+    free(oldhere->longname);
+    free(oldhere->icon);
+    free(oldhere->smallicon);
+    free(oldhere);
+
+    add_server(*zserver);
+  } // end of while loop.
+
+}
+
+void z3950proto::configure (const text_t &key,
+                const text_tarray &cfgline) {
 
   // this is called for each line in the gsdlsite.cfg file
-  /*  cerr << "z3950proto::configure called:"
-      << "key is " << key.getcstr()
-      << "\n1st line is " << cfgline[0].getcstr() << endl;
-  */
+
+  if (0)
+    cerr << "z3950proto::configure called:"
+     << "key is " << key.getcstr()
+     << "\n1st line is " << cfgline[0].getcstr() << endl;
+  
 }
 
@@ -96 +178 @@
   z3950_server_array::iterator end = zservers.end();
   while (here != end) {
-    collist.push_back(here->getName());
+    collist.push_back((*here)->getName());
     //const ShortColInfo_t *info=here->getInfo();
     //collist.push_back(info->name);
@@ -108 +190 @@
   z3950_server_array::iterator end = zservers.end();
   while (here != end) {
-    if(here->getName()==collection) {
+    if((*here)->getName()==collection) {
       hascollection=true;
       return;
@@ -133 +215 @@
   z3950_server_array::iterator end = zservers.end();
   while (here != end) {
-    if(here->getName()==collection) {
+    if((*here)->getName()==collection) {
       break;
     }
@@ -139 +221 @@
   }
 
-  // is this right? ie does end refer to the last element, or AFTER the
-  // last element?
   if (here==end) {
     err=protocolError;
@@ -149 +229 @@
   }
 
-  /*  collectinfo.shortInfo.name="sdfg";
-      collectinfo.shortInfo.host="localhost";
-      collectinfo.shortInfo.port=0; */
-  const ShortColInfo_t *colinfo=here->getInfo();
+  const ShortColInfo_t *colinfo=(*here)->getInfo();
   collectinfo.shortInfo.name=colinfo->name;
   collectinfo.shortInfo.host=colinfo->host;
@@ -163 +240 @@
   // leave ccsCols empty (no cross-coll. searching - for now)
   /*collectinfo.ccsCols=(text_tarray);*/ //not like this!!!
-  collectinfo.languages.push_back("en");
-  collectinfo.languages.push_back("fr");
+  // This info is available from the config file -- johnmcp
+  /*******collectinfo.languages.push_back("en");
+      collectinfo.languages.push_back("fr");********/
   collectinfo.numDocs=0;
   collectinfo.numWords=0;
@@ -170 +248 @@
   // copy the text maps over.
   //  collectinfo.collectionmeta; // text_tmap
-  //  delete collectinfo.collectionmeta;
-  collectinfo.collectionmeta=*(here->getMeta());
-  /* collectinfo.format; //text_tmap
-    collectinfo.building; //text_tmap
+  collectinfo.collectionmeta=*((*here)->getMeta());
+  collectinfo.format=*((*here)->getFormat()); //text_tmap
+  /*    collectinfo.building; //text_tmap  */
+
+  ////collectinfo.receptionist="z3950"; 
+  /* for now... this is a url,  relative to .../cgi-bin.
+     NOTE: if this is empty, it defaults to _gwcgi_?a=p&p=about&c=<colname>
   */
-
-collectinfo.receptionist="z3950"; /* for now... this is a url,
-                    relative to .../cgi-bin   */
 }
 
 void z3950proto::get_filterinfo (const text_t &/*collection*/,
-                 InfoFiltersResponse_t &/*response*/,
+                 InfoFiltersResponse_t &response,
                  comerror_t &/*err*/, ostream &/*logout*/) {
-
+  // we'll fake it here, and say we have set up some filters
+  response.filterNames.insert("BrowseFilter");
+  response.filterNames.insert("QueryFilter");
+  response.filterNames.insert("NullFilter");
+
+}
+
+void z3950proto::get_filteroptions (const text_t &/*collection*/,
+                    const InfoFilterOptionsRequest_t &/*req*/,
+                    InfoFilterOptionsResponse_t &response, 
+                    comerror_t &err, ostream &/*logout*/) {
+  // for now, assume all servers have the same characteristics
+  /*  if (request.filterName=="QueryFilter") { } 
+      else if (request.filterName=="BrowseFilter") { } 
+      else if (request.filterName=="NullFilter") {  } */
+  response.filterOptions["Index"].type=FilterOption_t::stringt;
+  response.filterOptions["Index"].repeatable=FilterOption_t::onePerQuery;
+  response.filterOptions["Index"].defaultValue="any";
+  response.filterOptions["Index"].validValues.push_back(".any");
+  response.filterOptions["Index"].validValues.push_back(".title");
+  response.filterOptions["Index"].validValues.push_back(".author");
+  // and maybe ["Language"] option as well?
+  err=noError;
 }
 
@@ -190 +290 @@
              FilterResponse_t &response,
              comerror_t &err, ostream &logout) {
-  // assume this function is only called when creating the title page,
-  // and looking for the metadata regarding icons.
-  response.numDocs=0;
-  response.isApprox=Approximate; // Exact | Approximate | MoreThan
+  // this function is called when:
+  //  *  creating the title page,(looking for iconcoll* & collectname metadata)
+  //  *  creating the about page (looking for "Title" metadata)
+  //  * doing the query - (note that a request for metadata comes first, then
+  // filterOptions = FRmetadata | FROID | FRtermFreq (64+4+1)
+
+  // metadata-only requests have filterName="NullFilter", else "QueryFilter".
+  // For the title page, we should not create a connection to the target
+  // (target means the actual z39.50 server, origin means us), but
+  // for the about page and query pages, we need to get information from the
+  // origin. (eg for the about page, we will print out some info such as ID,
+  // name and version.
+
+  // cerr now goes to errout.txt in etc directory
+  err=noError;
+
+  // get relevant "collection"
+  z3950_server_array::iterator zserver = zservers.begin();
+  z3950_server_array::iterator zend = zservers.end();
+  while (zserver != zend) {
+    if((*zserver)->getName()==collection) {
+      break;
+    }
+    zserver++;
+  }
+  // now have collection in zserver.
+  
+  ColInfoResponse_t info;
+  ResultDocInfo_t *docInfo;
+  
   // leave response.termInfo empty
-  // leave response.docInfo empty
-  /* got the following for first query:
-     request.filterResultOptions=FRmetadata;
-     request.getParents=getParents
-     request.fields=metadata
-     request.docSet (includes OID);
+  // response.termInfo.push_back(""); ??????? (should be empty if not req.)
+
+  // See if this is for a query action
+  if (request.filterName=="QueryFilter") {
+    /* Sample OptionValue pairs
+      `StartResults'=`1'
+      `EndResults'=`20'
+      `Term'=`firstword secondword' (term is just whatever the user typed in)
+      `QueryType'=`ranked' => 'OR' (cgiarg t=1)
+      `QueryType' = `boolean' => 'AND' (cgiarg t=0)
+      `Casefold'=`true'
+      `Stem'=`false'
+      `Maxdocs'=`50'
+     */
+    // go through options
+    text_t opt_term;            // the term(s) that the user entered 
+    text_t opt_fields;          // which fields to search on
+    int opt_start=1, opt_end=20; // default values
+    int nummatches=0, maxdocs=50; // default values
+    OptionValue_tarray::iterator ov_here=request.filterOptions.begin();
+    OptionValue_tarray::iterator ov_end=request.filterOptions.end();
+    while (ov_here != ov_end) {
+      //      cout << "OV pair: `" << ov_here->name.getcstr() << "'=`" 
+      //   << ov_here->value.getcstr() << "'\n";
+      if (ov_here->name=="Term") 
+    {
+      opt_term=ov_here->value;
+    } else if (ov_here->name=="Index") 
+      {
+        opt_fields=ov_here->value;
+      } else if (ov_here->name=="StartResults") 
+        {
+          opt_start=ov_here->value.getint();
+        } else if (ov_here->name=="EndResults") 
+          {
+        opt_end=ov_here->value.getint();
+          } else if (ov_here->name=="Maxdocs") 
+        {
+          maxdocs=ov_here->value.getint();
+        } 
+      ov_here++;
+    }
+    err=noError;
+    text_tarray *titles=(*zserver)->getrecordTitles(opt_term,
+                            opt_fields,
+                            opt_start, // first to get
+                            opt_end-opt_start, //count
+                            &nummatches,err);
+    if (err!=noError) {
+      // can we return an err msg in a response, or just use
+      // the more drastic Greenstone error mechanism?
+      docInfo=new ResultDocInfo_t;
+      response.docInfo.push_back(*docInfo);
+      docInfo->metadata["Title"].values.push_back("Error - query err?");
+      logout << "\nz3950 filter query: error connecting to server\n";
+      // for now, DON'T use GSDL protocol err.
+      err=noError;
+      return;
+    }
+    // check if (titles==NULL) - only happens on error?
+    if (nummatches>0) {
+      text_tarray::iterator titles_here=titles->begin();
+      text_tarray::iterator titles_end=titles->end();
+      int counter=1;
+      while (titles_here!=titles_end) {
+    docInfo=new ResultDocInfo_t; 
+    docInfo->metadata["Title"].values.push_back(*titles_here);
+    docInfo->result_num=counter;
+    // we need to give some OID, so we'll just use counter for now...
+    // make it the number into the whole possible retrieved set.
+    docInfo->OID=counter+opt_start-1;
+    response.docInfo.push_back(*docInfo);
+    counter++;
+    titles_here++;
+      }
+    }
+
+    if (request.filterResultOptions & FRtermFreq) {
+      if (nummatches>maxdocs) {
+        response.numDocs=maxdocs; // eg "more than 50" (if Maxdocs==50)
+    response.isApprox=MoreThan;
+      } else {
+    response.numDocs=nummatches; // eg "36 documents"
+    response.isApprox=Exact; // Exact | Approximate | MoreThan
+      }
+    } // end of if (... & FRtermFreq)
+    
+  } // end of if (... & FROID)
+  else {
+    // this wasn't a query action
+
+    if (request.filterOptions.size()>0 &&
+    request.filterOptions[0].name=="ParentNode") {
+      // don't want to return anything
+      return;
+      /*    } else if (request.docSet.size() &&
+        request.docSet[0]!="collection") {
+        // documentaction
+        // if docSet is not empty, it is either "collection", or an array
+        // of OIDs 
+        docInfo=new ResultDocInfo_t;
+        response.docInfo.push_back(*docInfo);
+      */
+    } else {
+      // in case we need to return only metadata
+      docInfo=new ResultDocInfo_t;
+      response.docInfo.push_back(*docInfo); 
+    }
+  }
+
+  // Fill in metadata for each response.docInfo (if wanted)
+  if (request.filterResultOptions & FRmetadata) {
+    get_collectinfo (collection, info, err, logout);
+    // should check err returned here....
+
+    // get the Query out of the filterOptions.
+    text_t query="";
+    text_t field="";
+    OptionValue_tarray::iterator opt_here=request.filterOptions.begin();
+    OptionValue_tarray::iterator opt_end=request.filterOptions.end();
+    while (opt_here!=opt_end) {
+      if (opt_here->name=="Query") {
+    query=opt_here->value;
+    if (field!="") break; // break from loop if we've got both
+      } else if (opt_here->name=="Index") {
+    field=opt_here->value;
+    if (query!="") break; // break from loop if we've got both
+      }
+      opt_here++;
+    }
+
+    if (!request.fields.empty()) {
+      // loop on each document being returned
+      ResultDocInfo_tarray::iterator docs_here=response.docInfo.begin();
+      ResultDocInfo_tarray::iterator docs_end=response.docInfo.end();
+      while (docs_here!=docs_end) {
+    
+    // loop on all the metadata fields in request.fields (type text_tset)
+    text_tset::iterator fields_here=request.fields.begin();
+    text_tset::iterator fields_end=request.fields.end();
+    text_tmap::iterator it;
+    while (fields_here!=fields_end) {
+      it=info.collectionmeta.find(*fields_here);
+      ////////// cerr << "filter: getting " << (*fields_here).getcstr();
+      if (it!=info.collectionmeta.end())
+        docs_here->metadata[*fields_here].values.push_back((*it).second);
+      else if (*fields_here=="Title" && !request.docSet.empty()) { 
+        // We only do this for a document action.
+        // (This comes through as a NullQuery).
+        // hopefully docSet is only not empty for documentaction...
+        text_t doctitle;
+        int i;
+        // check that docSet isn't empty first!!!!!!
+        i=request.docSet[0].getint();
+        text_t doctext="unneeded";
+        (*zserver)->getfullrecord(query, field, i, doctitle, doctext, err);
+        // check err value!
+        docs_here->metadata["Title"].values.push_back(doctitle);        
+      } else {
+        docs_here->metadata[*fields_here].values.push_back("");
+        /////// cerr << " (not found)";
+      }
+      ////////cerr << "\n";
+      fields_here++;
+    } // end of inner while loop
+    docs_here++;
+      } // end of outer while loop
+    } // end of if (!request.fields.empty())
+
+    else { // request.fields empty: return all metadata for about page or query
+      // we'll only put it in the first docInfo.
+      text_tmap::iterator colmeta_here=info.collectionmeta.begin();
+      text_tmap::iterator colmeta_end=info.collectionmeta.end();
+      while (colmeta_here!=colmeta_end) {
+    response.docInfo[0].metadata[(*colmeta_here).first].
+      values.push_back((*colmeta_here).second);
+    /////cerr << "\t" << (*colmeta_here).first.getcstr() << "\n";
+    colmeta_here++;
+      }
+
+      // check if "collectionextra" metadata is set. If it isn't, we should
+      // create connection to target to get it.
+      if (info.collectionmeta.find("collectionextra")==colmeta_end) {
+    // it hasn't been set yet...
+    text_t abouttext="<B>Server Online</B><br>\n";
+    abouttext+=(*zserver)->getzAbout();
+    // add in the "About" text we read in from config file.
+    // how do we incorporate multi-lingual metadata?
+    abouttext+="<P>\n";
+    text_t tmpabout;
+
+    if ((*zserver)->getcfgAbout("en", tmpabout)==true)
+      abouttext+=tmpabout;
+
+    (*zserver)->setMeta("collectionextra",abouttext);
+    response.docInfo[0].metadata["collectionextra"].values.push_back(abouttext);
+      }
+    } // end of else
+
+    // do indices' names, regardless of whether asked for or not...
+    if (!response.docInfo.empty()) {
+     response.docInfo[0].metadata[".author"].values.push_back("author fields");
+      response.docInfo[0].metadata[".title"].values.push_back("title fields");
+      response.docInfo[0].metadata[".any"].values.push_back("any fields");
+    }
+  } //end of  if (... & FRmetadata) ...
+}
+
+
+
+void z3950proto::get_document (const text_t &collection,
+                   const DocumentRequest_t &request,
+                   DocumentResponse_t &response,
+                   comerror_t &err, ostream &logout) {
+
+  err=noError;
+
+  // get relevant "collection"
+  z3950_server_array::iterator zserver = zservers.begin();
+  z3950_server_array::iterator zend = zservers.end();
+  while (zserver != zend) {
+    if((*zserver)->getName()==collection) {
+      break;
+    }
+    zserver++;
+  }
+  // now have collection in zserver.
+
+  /*  cout << "get document:\n\tOID: " << request.OID.getcstr()
+      << "\n\tdocType: " << request.docType.getcstr()
+      << "\n\tdocFormat: " << request.docFormat.getcstr() <<"\n";
   */
-  // fill in the metadata for each of the OIDs (if it is requested)
-  if (request.filterResultOptions & FRmetadata) {
-    // request.fields // type text_tset
-    // docInfo is type ResultDocInfo_tarray
-    ColInfoResponse_t *info = new ColInfoResponse_t;
-    get_collectinfo (collection, *info, err, logout);
-    // should check err returned here....
-
-    ResultDocInfo_t *docInfo=new ResultDocInfo_t;
-    
-    text_tmap::iterator it=info->collectionmeta.find("collectionname");
-    // cerr now goes to errout.txt in etc directory
-    // "(*it).first" is key, "(*it).second" is value(s).
-    if (it!=info->collectionmeta.end()) 
-      docInfo->metadata["collectionname"].values.push_back((*it).second);
-    else // needs to exist, but be empty
-      docInfo->metadata["collectionname"].values.push_back("");
-
-    it=info->collectionmeta.find("iconcollection");
-    if (it!=info->collectionmeta.end()) 
-      docInfo->metadata["iconcollection"].values.push_back((*it).second);
-    else // needs to exist, but be empty
-      docInfo->metadata["iconcollection"].values.push_back("");
-
-    it=info->collectionmeta.find("iconcollectionsmall");
-    if (it!=info->collectionmeta.end()) 
-      docInfo->metadata["iconcollectionsmall"].values.push_back((*it).second);
-    else // needs to exist, but be empty
-      docInfo->metadata["iconcollectionsmall"].values.push_back("");
-    
-
-    response.docInfo.push_back(*docInfo);
-  }
-
-  err=noError;
-}
-
-void z3950proto::get_document (const text_t &/*collection*/,
-                   const DocumentRequest_t &/*request*/,
-                   DocumentResponse_t &/*response*/,
-                   comerror_t &err, ostream &logout) {
-
+
+  /* docresponse consists of
+     text_t response.doc */
+  text_t title="unneeded";
+  text_t doctext;
+  text_t query; // this should not be needed, as we have already connected to 
+  // get the  title....
+  text_t field; // ditto...
+  (*zserver)->getfullrecord(query,field,request.OID.getint(),
+                title,doctext,err);
+  // check return value of above? (false=>not connected) 
+  if (err==noError)
+    response.doc=doctext;
+  else {
+    // could print out different messages based on error type....
+    response.doc="<h2>Error</h2>There was an error while connecting to the ";
+    response.doc+="z39.50 server (ie target). Most likely this was a \n";
+    response.doc+="\"Connection Refused\" error.\n";
+
+  }
   if (0) {
     err=protocolError;
@@ -249 +578 @@
   }
 }
-
-
-#endif