Changeset 12191 for trunk/greenstone3-extensions/gs3build/src/org/greenstone/gsdl3/gs3build/indexers/MGIndexer.java

Timestamp:

2006-07-13T10:29:55+12:00 (18 years ago)

Author:

kjdon

Message:

committed some changes that I had made ages ago. Not sure if it still compiles - I need to write an ant build file for this, and check compilation. Will do it once I need to - its unclear whether anyone will ever use this again

File:

• trunk/greenstone3-extensions/gs3build/src/org/greenstone/gsdl3/gs3build/indexers/MGIndexer.java (modified) (12 diffs)

trunk/greenstone3-extensions/gs3build/src/org/greenstone/gsdl3/gs3build/indexers/MGIndexer.java

@@ -18 +18 @@
 import org.greenstone.gsdl3.gs3build.doctypes.DocumentID;
 import org.greenstone.gsdl3.gs3build.doctypes.DocumentInterface;
+import org.greenstone.gsdl3.gs3build.doctypes.AbstractDocument;
 import org.greenstone.gsdl3.gs3build.doctypes.HTMLDocument;
 import org.greenstone.gsdl3.gs3build.doctypes.METSDocument;
@@ -24 +25 @@
 import org.greenstone.gsdl3.gs3build.util.DOMUtils;
 import org.greenstone.gsdl3.util.GSXML;
+import org.greenstone.gsdl3.util.GSFile;
 import org.greenstone.gsdl3.util.Misc;
 import org.greenstone.gsdl3.util.Processing;
@@ -293 +295 @@
         //metsDoc.setModified(true);
         //  System.out.println("Assigning " + this.sectionSeqNo + " to " + metsDoc.getID() + " " + division.getLabel());
-        } // section level
+        } // first pass
     
         // append an 'end of section' marker
@@ -408 +410 @@
     public boolean indexDocument(DocumentID docID, DocumentInterface document)
     {
-    
-    if (!this.firstDocument) {
-        this.indexBuffer.append(END_OF_DOCUMENT);
-        mgPasses.processDocument(indexBuffer.toString());
-        this.indexBuffer.delete(0, this.indexBuffer.length());
-        
-    }
-
+    int count = ((AbstractDocument)document).getNumSections();
     String docText = null;
     // set the mgseqno if first pass
@@ -426 +421 @@
     
     //long start = System.currentTimeMillis();
-    Document domDocument = document.getDOMDocument();
-    if (domDocument != null) {
-        System.err.println("dom doc is not null");
-        METSStructure sections = document.getDocumentStructure().getStructure("Section");
-        if (sections != null) {
-        System.err.println("sections are not null");
-        docText = this.prepareDOM(document, domDocument, sections, "gsdl3"); //this.name, "gsdl3", this.field);
-        //  System.out.println(docText);
-        }
-    }
+    if (this.current_index.getLevel().equals("section")) {
+
+        Document domDocument = document.getDOMDocument();
+        if (domDocument != null) {
+        System.err.println("dom doc is not null");
+        METSStructure sections = document.getDocumentStructure().getStructure("Section");
+        if (sections != null) {
+            System.err.println("sections are not null");
+            docText = this.prepareDOM(document, domDocument, sections, "gsdl3"); //this.name, "gsdl3", this.field);
+            //  System.out.println(docText);
+        }
+        }
+    } 
+        
+    
     //long finish = System.currentTimeMillis();
     //System.err.println("dom doc = "+ Long.toString(finish-start));
@@ -447 +447 @@
         if (field.equals("text")) {
             doc_text_buffer.append(document.getDocumentText());
+            doc_text_buffer.append(" ");
         }  else {
             // its a metadata - do namespace properly!!
@@ -455 +456 @@
                 String value = valueIter.next().toString();
                 doc_text_buffer.append(value);
+                doc_text_buffer.append(" ");
             }
             }
@@ -461 +463 @@
         docText = doc_text_buffer.toString();
         sectionSeqNo ++;
+        int num_secs = 0;
     }
     //finish = System.currentTimeMillis();
@@ -467 +470 @@
     this.indexBuffer.append(docText);
     // remember that we're not on the first document, 
-    this.firstDocument = false;
+    //this.firstDocument = false;
     this.documentSeqNo ++;
+    //if (!this.firstDocument) {
+    this.indexBuffer.append(END_OF_DOCUMENT);
+    mgPasses.processDocument(indexBuffer.toString());
+    String filename="";
+    try {
+        filename = "pass"+this.pass+"doc"+this.documentSeqNo+".txt";
+        System.err.println("trying to write to "+filename);
+        GSFile.writeFile(indexBuffer.toString().getBytes(), filename);
+    } catch (Exception e) {
+        System.err.println("COUldn't write to file, "+filename);
+    }
+    this.indexBuffer.delete(0, this.indexBuffer.length());
+        
+    
 
     return true;
@@ -487 +504 @@
     this.indexBuffer = new StringBuffer();
     int indexNo = this.pass/2;
-    this.current_index = null;
     
     this.current_index = (MGIndex) this.indexes.get(indexNo);
@@ -512 +528 @@
         this.textStem = this.indexStem;
     }
+    
     mgPasses.setFileName(this.indexStem);
     if (!Misc.isWindows()) {
@@ -573 +590 @@
     } catch (Exception e) {}
     
-    int exit_value = 0;
+    int exit_value = mgPasses.exitValue();
     System.out.println("Pass " + this.pass + " completed with " + exit_value);
     if (exit_value !=0) {