add additional options and features to HWPFLister, enabling PAPX and paragraphs (including SPRMs) information dump

git-svn-id: https://svn.apache.org/repos/asf/poi/trunk@1142861 13f79535-47bb-0310-9956-ffa450edef68

src/scratchpad/src/org/apache/poi/hwpf/dev/HWPFLister.java

@@ -18,36 +18,125 @@
 package org.apache.poi.hwpf.dev;
 
 import java.io.FileInputStream;
+import java.util.Arrays;
 
 import org.apache.poi.hwpf.HWPFDocument;
+import org.apache.poi.hwpf.HWPFDocumentCore;
 import org.apache.poi.hwpf.model.FileInformationBlock;
+import org.apache.poi.hwpf.model.PAPX;
+import org.apache.poi.hwpf.sprm.SprmIterator;
+import org.apache.poi.hwpf.sprm.SprmOperation;
+import org.apache.poi.hwpf.usermodel.Paragraph;
+import org.apache.poi.hwpf.usermodel.Range;
 
 /**
- * Used by developers to list out key information on a
+ * Used by developers to list out key information on a HWPF file. End users will
- *  HWPF file. End users will probably never need to
+ * probably never need to use this program.
- *  use this program.
+ * 
+ * @author Nick Burch (nick at torchbox dot com)
+ * @author Sergey Vladimirov (vlsergey at gmail dot com)
  */
-public final class HWPFLister {
+public final class HWPFLister
-	private final HWPFDocument _doc;
+{
-	public HWPFLister(HWPFDocument doc) {
+    public static void main( String[] args ) throws Exception
-		_doc = doc;
+    {
-	}
+        if ( args.length == 0 )
+        {
+            System.err.println( "Use:" );
+            System.err.println( "   HWPFLister <filename> "
+                    + "[--papx] [--papxProperties] "
+                    + "[--paragraphs] [--paragraphsSprms] [--paragraphsText]" );
+            System.exit( 1 );
+        }
 
-	public static void main(String[] args) throws Exception {
+        boolean outputParagraphs = false;
-		if(args.length == 0) {
+        boolean outputParagraphsSprms = false;
-			System.err.println("Use:");
+        boolean outputParagraphsText = false;
-			System.err.println("   HWPFLister <filename>");
-			System.exit(1);
-		}
 
-		HWPFLister l = new HWPFLister(
+        boolean outputPapx = false;
-				new HWPFDocument(new FileInputStream(args[0]))
+        boolean outputPapxProperties = false;
-		);
-		l.dumpFIB();
-	}
 
-	public void dumpFIB() {
+        for ( String arg : Arrays.asList( args ).subList( 1, args.length ) )
-		FileInformationBlock fib = _doc.getFileInformationBlock();
+        {
-		System.out.println(fib.toString());
+            if ( "--paragraphs".equals( arg ) )
-	}
+                outputParagraphs = true;
+            if ( "--paragraphsSprms".equals( arg ) )
+                outputParagraphsSprms = true;
+            if ( "--paragraphsText".equals( arg ) )
+                outputParagraphsText = true;
+
+            if ( "--papx".equals( arg ) )
+                outputPapx = true;
+            if ( "--papxProperties".equals( arg ) )
+                outputPapxProperties = true;
+        }
+
+        HWPFLister lister = new HWPFLister( new HWPFDocument(
+                new FileInputStream( args[0] ) ) );
+        lister.dumpFIB();
+
+        if ( outputParagraphs )
+        {
+            System.out.println( "== Paragraphs ==" );
+            lister.dumpParagraphs( outputParagraphsSprms, outputPapx,
+                    outputParagraphsText );
+        }
+
+        if ( !outputParagraphs && outputPapx )
+        {
+            System.out.println( "== PAPX ==" );
+            lister.dumpPapx( outputPapxProperties );
+        }
+    }
+
+    private final HWPFDocumentCore _doc;
+
+    public HWPFLister( HWPFDocumentCore doc )
+    {
+        _doc = doc;
+    }
+
+    public void dumpFIB()
+    {
+        FileInformationBlock fib = _doc.getFileInformationBlock();
+        System.out.println( fib );
+    }
+
+    public void dumpPapx( boolean withProperties )
+    {
+        for ( PAPX papx : _doc.getParagraphTable().getParagraphs() )
+        {
+            System.out.println( papx );
+
+            if ( withProperties )
+                System.out.println( papx.getParagraphProperties( _doc
+                        .getStyleSheet() ) );
+        }
+    }
+
+    public void dumpParagraphs( boolean withSprms, boolean withPapx,
+            boolean withText )
+    {
+        Range range = _doc.getRange();
+        for ( int p = 0; p < range.numParagraphs(); p++ )
+        {
+            Paragraph paragraph = range.getParagraph( p );
+            System.out.println( paragraph.toString( withPapx ) );
+
+            if ( withSprms )
+            {
+                PAPX papx = _doc.getParagraphTable().getParagraphs().get( p );
+
+                SprmIterator sprmIt = new SprmIterator( papx.getGrpprl(), 2 );
+                while ( sprmIt.hasNext() )
+                {
+                    SprmOperation sprm = sprmIt.next();
+                    System.out.println( "\t" + sprm.toString() );
+                }
+            }
+
+            if ( withText )
+                System.out.println( paragraph.text() );
+        }
+    }
 }