Changeset 18895

Timestamp:

2009-04-06T13:18:45+12:00 (15 years ago)

Author:

kjdon

Message:

added is_valid_element to metaformat, which checks elementSet for the element name. metaformat subclass constructors must set up elementSet. Currently they do this in the code. would be good if it could read in from a file eventually, maybe?? When metaformat is outputting the m,etadata, it checks whether the element is valid before outputting. otherwise it will invalidate the response as it won't conform to the schema. also, changed where we lowercased the entire name to lowercasing only the first letter - some qdc fields have an internal upper case letter.

Location:

gsdl/trunk/runtime-src/src/oaiservr

Files:

• dublincore.cpp (modified) (1 diff)
• dublincore.h (modified) (1 diff)
• metaformat.cpp (modified) (3 diffs)
• metaformat.h (modified) (2 diffs)
• qualified_dublincore.cpp (modified) (1 diff)
• qualified_dublincore.h (modified) (1 diff)
• rfc1807.cpp (modified) (2 diffs)
• rfc1807.h (modified) (1 diff)

gsdl/trunk/runtime-src/src/oaiservr/dublincore.cpp

@@ -1 +1 @@
 
 #include "dublincore.h"
+
+dublin_core::dublin_core() : metaformat() {
+  // These element names taken from the schema
+  // http://www.openarchives.org/OAI/2.0/oai_dc.xsd
+  elementSet.insert("contributor");
+  elementSet.insert("coverage");
+  elementSet.insert("creator");
+  elementSet.insert("date");
+  elementSet.insert("description");
+  elementSet.insert("format");
+  elementSet.insert("identifier");
+  elementSet.insert("language");
+  elementSet.insert("publisher");
+  elementSet.insert("relation");
+  elementSet.insert("rights");
+  elementSet.insert("source");
+  elementSet.insert("subject");
+  elementSet.insert("title");
+  elementSet.insert("type");
+ 
+
+}
 
 const text_t dublin_core::formatName() {

gsdl/trunk/runtime-src/src/oaiservr/dublincore.h

@@ -6 +6 @@
   virtual void output_metadata_header(ostream &output);
   virtual void output_metadata_footer(ostream &output);
+
  public:
-  dublin_core() : metaformat(){};
+  dublin_core(); 
   virtual const text_t formatName();
   virtual const text_t formatPrefix();

gsdl/trunk/runtime-src/src/oaiservr/metaformat.cpp

@@ -69 +69 @@
     if (metaItem == this->formatPrefix()) {
       metaItem = substr(last+1, here->first.end()); // Get the rest of the metadata tag (it's name) but without the '.'
-      lc(metaItem); // Convert it to lowercase for putting in the xml tags
-
+      lc(metaItem.begin(),metaItem.begin()+1); // We want lowercase, but some of the fields in qualified dublin core have internal upper case, eg instructionalMethod. So we assume that lowercasing the first letter is enough
       if (doOutput) {
-    this->output_item(output, utf8convert, headerDone, metaItem, here->second.values);
+    if (this->is_valid_element(metaItem)) {
+      
+      this->output_item(output, utf8convert, headerDone, metaItem, here->second.values);
+    }
       }
       else {
@@ -86 +88 @@
     // particular metadata is stuff we don't want?
     if (doOutput) {
-      this->output_item(output, utf8convert, headerDone, mapTo, here->second.values);
+      if (this->is_valid_element(mapTo)) {
+        this->output_item(output, utf8convert, headerDone, mapTo, here->second.values);
+      }
     }
     else {
@@ -115 +119 @@
   ofstream o("dummy", ios::out);
   return this->scan_metadata(o, collection, docInfo, false);
+}
+
+bool metaformat::is_valid_element(text_t &meta_name)
+{
+  if (elementSet.count(meta_name)==1) return true;
+  return false;
+  
 }
 

gsdl/trunk/runtime-src/src/oaiservr/metaformat.h

@@ -19 +19 @@
  protected:
   virtual void output_metadata_header(ostream &output) = 0;
-  virtual void output_item(ostream &output, outconvertclass &outconvert, bool &headerDone, const text_t &label, 
+  virtual void output_item(ostream &output, outconvertclass &outconvert, 
+               bool &headerDone, const text_t &label, 
                const text_tarray &values);
   virtual void output_metadata_footer(ostream &output) = 0;
-  virtual bool scan_metadata(ostream &output, const text_t &collection,  ResultDocInfo_t &docInfo, 
-                 bool doOutput);
+  virtual bool is_valid_element(text_t &meta_name);
+  virtual bool scan_metadata(ostream &output, const text_t &collection,  
+                 ResultDocInfo_t &docInfo, bool doOutput);
   oaiconfig *    oaiConfigure;
+
+  // a set containing all the valid element names for the set. Must be 
+  // initialized by each metadata class
+  text_tset elementSet; 
 
  public:
@@ -30 +36 @@
   virtual const text_t formatName() = 0;
   virtual const text_t formatPrefix() = 0;
-  virtual bool output_record(ostream &output, recptproto *protocol, const text_t &collection, 
+  virtual bool output_record(ostream &output, recptproto *protocol, 
+                 const text_t &collection, 
                  const text_t &record_OID);
   virtual bool is_available(const text_t &collection, ResultDocInfo_t &docInfo);
-  virtual bool output_metadata(ostream &output, const text_t &collection,  ResultDocInfo_t &docInfo);
+  virtual bool output_metadata(ostream &output, const text_t &collection,  
+                   ResultDocInfo_t &docInfo);
   virtual bool output_formatdata(ostream &output) = 0;
   void    set_configuration(oaiconfig *config) { this->oaiConfigure = config; }

gsdl/trunk/runtime-src/src/oaiservr/qualified_dublincore.cpp

@@ -2 +2 @@
 #include "qualified_dublincore.h"
 
+qualified_dublin_core::qualified_dublin_core() : metaformat() {
+  // These element names taken from the schema
+  //http://www.greenstone.org/namespace/gsdl_qdc/1.0/gsdl_qdc.xsd
+
+  elementSet.insert("title");
+   elementSet.insert("creator");
+   elementSet.insert("subject");
+   elementSet.insert("description");
+   elementSet.insert("publisher");
+   elementSet.insert("contributor");
+   elementSet.insert("date");
+   elementSet.insert("type");
+   elementSet.insert("format");
+   elementSet.insert("identifier");
+   elementSet.insert("source");
+   elementSet.insert("language");
+   elementSet.insert("relation");
+   elementSet.insert("coverage");
+   elementSet.insert("rights");
+   elementSet.insert("alternative");
+   elementSet.insert("tableOfContents");
+   elementSet.insert("abstract");
+   elementSet.insert("created");
+   elementSet.insert("valid");
+   elementSet.insert("available");
+   elementSet.insert("issued"); 
+   elementSet.insert("modified");
+   elementSet.insert("dateAccepted");
+   elementSet.insert("dateCopyrighted");
+   elementSet.insert("dateSubmitted");
+   elementSet.insert("extent");
+   elementSet.insert("medium");
+   elementSet.insert("isVersionOf");
+   elementSet.insert("hasVersion");
+   elementSet.insert("isReplacedBy");
+   elementSet.insert("replaces"); 
+   elementSet.insert("isRequiredBy");
+   elementSet.insert("requires"); 
+   elementSet.insert("isPartOf");
+   elementSet.insert("hasPart"); 
+   elementSet.insert("isReferencedBy");
+   elementSet.insert("references");
+   elementSet.insert("isFormatOf");
+   elementSet.insert("hasFormat");
+   elementSet.insert("conformsTo");
+   elementSet.insert("spatial");
+   elementSet.insert("temporal");
+   elementSet.insert("audience");
+   elementSet.insert("accrualMethod");
+   elementSet.insert("accrualPeriodicity");
+   elementSet.insert("accrualPolicy");
+   elementSet.insert("instructionalMethod");
+   elementSet.insert("provenance");
+   elementSet.insert("rightsHolder");
+   elementSet.insert("mediator");
+   elementSet.insert("educationLevel");
+   elementSet.insert("accessRights");
+   elementSet.insert("license"); 
+   elementSet.insert("bibliographicCitation");
+}
 const text_t qualified_dublin_core::formatName() {
   return "gsdl_qdc";

gsdl/trunk/runtime-src/src/oaiservr/qualified_dublincore.h

@@ -7 +7 @@
   virtual void output_metadata_footer(ostream &output);
  public:
-  qualified_dublin_core() : metaformat(){}
+  qualified_dublin_core();
   virtual const text_t formatName();
   virtual const text_t formatPrefix();

gsdl/trunk/runtime-src/src/oaiservr/rfc1807.cpp

@@ -1 +1 @@
 #include "rfc1807.h"
+
+rfc1807::rfc1807() : metaformat(){
+  // These element names taken from the schema
+  // http://www.openarchives.org/OAI/1.1/rfc1807.xsd
+  elementSet.insert("bib-version");
+  elementSet.insert("id");
+  elementSet.insert("entry"); 
+  elementSet.insert("organization");
+  elementSet.insert("title");
+  elementSet.insert("type");
+  elementSet.insert("revision");
+  elementSet.insert("withdraw");
+  elementSet.insert("author"); 
+  elementSet.insert("corp-author");
+  elementSet.insert("contact"); 
+  elementSet.insert("date");
+  elementSet.insert("pages");
+  elementSet.insert("copyright");
+  elementSet.insert("handle"); 
+  elementSet.insert("other_access");
+  elementSet.insert("retrieval");
+  elementSet.insert("keyword"); 
+  elementSet.insert("cr-category");
+  elementSet.insert("period"); 
+  elementSet.insert("series"); 
+  elementSet.insert("monitoring");
+  elementSet.insert("funding");
+  elementSet.insert("contract");
+  elementSet.insert("grant");
+  elementSet.insert("language");
+  elementSet.insert("notes"); 
+  elementSet.insert("abstract");
+ 
+}; 
 
 const text_t rfc1807::formatName() 
@@ -56 +90 @@
   return true;
 }
+
+

gsdl/trunk/runtime-src/src/oaiservr/rfc1807.h

@@ -6 +6 @@
   virtual void output_metadata_footer(ostream &output);
  public:
-  rfc1807() : metaformat(){}; // call parent constructor
+  rfc1807(); 
   virtual const text_t formatName();
   virtual const text_t formatPrefix();