Changeset 8460

Timestamp:

2004-11-05T10:08:31+13:00 (19 years ago)

Author:

kjdon

Message:

added in Chi's changes for METS doc handling

File:

• trunk/gsdl3/src/java/org/greenstone/gsdl3/gs3build/doctypes/METSDocument.java (modified) (5 diffs)

trunk/gsdl3/src/java/org/greenstone/gsdl3/gs3build/doctypes/METSDocument.java

@@ -18 +18 @@
 
 import org.greenstone.gsdl3.gs3build.metadata.*;
+import org.greenstone.gsdl3.gs3build.util.HTMLDoc;
+import org.greenstone.gsdl3.gs3build.doctypes.DocumentInterface;
+import org.greenstone.gsdl3.gs3build.xpointer.XPointer;
 
 public class METSDocument extends AbstractDocument {
 
     public static final String METS_DOCUMENT_TYPE = "METS";
+    Document domDocument;
+    public METSDocument(DocumentID id) {
+        super(id);
+    //domDocument = null;
+    }
 
     public METSDocument(URL url) {
@@ -28 +36 @@
     if (url.getProtocol().equals("file")) {
         this._parseFile(new File(url.getPath()));
-    } 
+    } else {
+        System.err.println("Error: METSDocument must currently be a file");
+        System.err.println("       URL = " + url.getPath());
+    }
     }
     
     private void _parseFile(File file) {
     
+    String parseFilePath;
     try {
         DocumentBuilderFactory factory = DocumentBuilderFactory.newInstance();
         DocumentBuilder builder = factory.newDocumentBuilder();
-        Document document = builder.parse(file);
-
+        domDocument = builder.parse(file);
+
+        int filePosition = file.getPath().indexOf("import/")+7;
+        parseFilePath = file.getPath().substring(0, filePosition);
         // TODO: get all the types in the tree
       
@@ -46 +60 @@
         System.out.println("Read");
       
-        NodeList fileSecs = document.getElementsByTagName("mets:fileSec");
-      
-        for (int g = 0; g < fileSecs.getLength(); g ++) {
+        // Get METSDmdSec metadata sections
+        NodeList dmdSecs = domDocument.getElementsByTagName("mets:dmdSec");
+        
+        this.metadata = METSDescriptiveSet.parseXML(dmdSecs);
+
+        // Get METSFileSec sections
+        NodeList fileSecs = domDocument.getElementsByTagName("mets:fileSec");
+        this.fileSet = METSFileSet.parseXML(fileSecs, fileSet, parseFilePath);
+        // Get METSStructureMap Sections
+        NodeList structMapSecs = domDocument.getElementsByTagName("mets:structMap");
+        this.structureSet = METSStructureSet.parseXML(structMapSecs);
+
+        //for (int g = 0; g < fileSecs.getLength(); g ++) {
         //Schema schema = new Schema(schemas.item(s));
-        this._parseFileSec(fileSecs.item(g));
-        }
+        //this._parseFileSec(fileSecs.item(g));
+        //}
       
         // Get document metadata sections
-        NodeList dmdSecs = document.getElementsByTagName("mets:dmdSec");
+        //NodeList dmdSecs = document.getElementsByTagName("mets:dmdSec");
       
         //Schema schema = new Schema(schemas.item(s));
-        this.metadata = METSDescriptiveSet.parseXML(fileSecs);
-    }
-    catch (FactoryConfigurationError e) {
+        //this.metadata = METSDescriptiveSet.parseXML(fileSecs);
+    } catch (FactoryConfigurationError e) {
         System.out.println(e);
-    }
-    catch (ParserConfigurationException ex) {
+    } catch (ParserConfigurationException ex) {
         System.out.println(ex);
-    }
-    catch (SAXException ex) {
+    } catch (SAXException ex) {
         System.out.println(ex);
-    }
-    catch (IOException ex) {
+    } catch (IOException ex) {
         System.out.println(ex);
     }
     }
-
+    /*
     private void _parseFileSec(Node fileSec) {
     // this is in effect a group without a sense of 'self'...
@@ -108 +128 @@
         }
     }
-    }
+    }*/
     
     public String getDocumentType() {
@@ -116 +136 @@
     public String getDocumentText() {
     // TODO: make this more than a dummy function!
-    return null;
-    }
-
-    public String getSectionText(String sectionId) {
-    return null;
-    }
-    
+    HTMLDoc metsDoc;
+    URL url = (URL) this.fileSet.getFile(0).getLocation();
+
+    this.getSectionText("1");
+    
+    if (url.getProtocol().equals("file")) {
+        metsDoc = new HTMLDoc(url,url.getPath());
+    } else {
+        metsDoc = new HTMLDoc(url);
+    }
+    return metsDoc.getContent();
+    }
+    public Document getDOMDocument() 
+    {
+    if (domDocument == null) {
+        URL url = (URL) this.fileSet.getFile(0).getLocation();
+        try {
+        DocumentBuilderFactory factory = DocumentBuilderFactory.newInstance();
+        DocumentBuilder builder = factory.newDocumentBuilder();
+        domDocument = builder.parse(new File(url.getPath()));
+        } catch (FactoryConfigurationError e) {
+        System.out.println(e);
+        } catch (ParserConfigurationException ex) {
+        System.out.println(ex);
+        } catch (SAXException ex) {
+        System.out.println(ex);
+        } catch (IOException ex) {
+        System.out.println(ex);
+        }
+    }
+    return domDocument;
+    }
+
+    private XPointer getXPointer(METSDivision division)
+    {
+    String fileId = division.getDefaultFileReference();
+
+    if (fileId == null) {
+        System.err.println("Unable to obtain file reference for section " + division.getLabel());
+        return null;
+    }
+    
+    METSFileGroup fileGroup = this.getDocumentFiles().getGroup(fileId);
+    if (fileGroup == null) {
+        System.err.println("Unable to obtain file reference for filegroup " + fileId);
+        return null;
+    }
+    METSFile file = fileGroup.getFile(0);
+    if (file == null) {
+        System.err.println("Unable to obtain any files within filegroup " + fileId);
+        return null;
+    }
+    URL url = file.getLocation();
+    XPointer xpointer = XPointer.processXPointer(domDocument, url);
+    
+    return xpointer;
+    }
+    
+    private XPointer getXPointer(String sectionId)
+    {
+    if (domDocument == null) {
+        URL url = (URL) this.fileSet.getFile(0).getLocation();
+        try {
+        DocumentBuilderFactory factory = DocumentBuilderFactory.newInstance();
+        DocumentBuilder builder = factory.newDocumentBuilder();
+        domDocument = builder.parse(new File(url.getPath()));
+        } catch (FactoryConfigurationError e) {
+        System.out.println(e);
+        } catch (ParserConfigurationException ex) {
+        System.out.println(ex);
+        } catch (SAXException ex) {
+        System.out.println(ex);
+        } catch (IOException ex) {
+        System.out.println(ex);       
+        }
+    }
+    
+    METSDivision division = 
+        this.getDocumentStructure().getDivision(METSStructureSet.GSDL3_SECTION_STRUCTURE, sectionId);
+    
+    if(division == null) {
+        return null;
+    }
+        return this.getXPointer(division);
+    }
+
+    public Node getSectionStartNode(METSDivision division)
+    {
+    XPointer xpointer = this.getXPointer(division);
+    return xpointer.getStartNode();
+    }
+
+    public Node getSectionStartNode(String sectionId)
+    {
+    XPointer xpointer = this.getXPointer(sectionId);
+    return xpointer.getStartNode();
+    }
+    
+    public String getSectionText(String sectionId) 
+    {
+    XPointer xpointer = this.getXPointer(sectionId);
+    
+    if (xpointer == null) {
+        return "";
+    }
+
+    //get the XML content of the xpointer
+    return xpointer.toString();
+    }
 }
+
+