rewrite table bounds detection for Word 97, including inner table support

git-svn-id: https://svn.apache.org/repos/asf/poi/trunk@1143070 13f79535-47bb-0310-9956-ffa450edef68

src/scratchpad/src/org/apache/poi/hwpf/converter/AbstractWordConverter.java

@@ -241,6 +241,14 @@ public abstract class AbstractWordConverter
             if ( paragraph.isInTable()
                     && paragraph.getTableLevel() != currentTableLevel )
             {
+                if ( paragraph.getTableLevel() < currentTableLevel )
+                    throw new IllegalStateException(
+                            "Trying to process table cell with higher level ("
+                                    + paragraph.getTableLevel()
+                                    + ") than current table level ("
+                                    + currentTableLevel
+                                    + ") as inner table part" );
+
                 Table table = range.getTable( paragraph );
                 processTable( wordDocument, flow, table );
 

src/scratchpad/src/org/apache/poi/hwpf/usermodel/Paragraph.java

@@ -84,9 +84,9 @@ public class Paragraph extends Range implements Cloneable {
   protected ParagraphProperties _props;
   protected SprmBuffer _papx;
 
-  protected Paragraph(int startIdx, int endIdx, Table parent)
+  protected Paragraph(int startIdxInclusive, int endIdxExclusive, Table parent)
   {
-    super(startIdx, endIdx, Range.TYPE_PARAGRAPH, parent);
+    super(startIdxInclusive, endIdxExclusive, Range.TYPE_PARAGRAPH, parent);
     PAPX papx = _paragraphs.get(_parEnd - 1);
     _props = papx.getParagraphProperties(_doc.getStyleSheet());
     _papx = papx.getSprmBuf();

src/scratchpad/src/org/apache/poi/hwpf/usermodel/Range.java

@@ -90,10 +90,10 @@ public class Range { // TODO -instantiable superclass
 	/** All paragraphs that belong to the document this Range belongs to. */
 	protected List<PAPX> _paragraphs;
 
-	/** The start index in the paragraphs list for this Range */
+	/** The start index in the paragraphs list for this Range, inclusive */
 	protected int _parStart;
 
-	/** The end index in the paragraphs list for this Range. */
+	/** The end index in the paragraphs list for this Range, exclusive */
 	protected int _parEnd;
 
 	/** Have we loaded the characterRun indexes yet. */
@@ -178,9 +178,9 @@ public class Range { // TODO -instantiable superclass
 	 * lists.
 	 *
 	 * @param startIdx
-	 *            The starting index in the list.
+	 *            The starting index in the list, inclusive
 	 * @param endIdx
-	 *            The ending index in the list.
+	 *            The ending index in the list, exclusive
 	 * @param idxType
 	 *            The list type.
 	 * @param parent
@@ -199,27 +199,27 @@ public class Range { // TODO -instantiable superclass
 				_parStart = parent._parStart + startIdx;
 				_parEnd = parent._parStart + endIdx;
 				_start = _paragraphs.get(_parStart).getStart();
-				_end = _paragraphs.get(_parEnd).getEnd();
+				_end = _paragraphs.get(_parEnd - 1).getEnd();
 				_parRangeFound = true;
 				break;
 			case TYPE_CHARACTER:
 				_charStart = parent._charStart + startIdx;
 				_charEnd = parent._charStart + endIdx;
-				_start = _characters.get(_charStart).getStart();
+				_start = _characters.get(_charStart - 1).getStart();
 				_end = _characters.get(_charEnd).getEnd();
 				_charRangeFound = true;
 				break;
 			case TYPE_SECTION:
 				_sectionStart = parent._sectionStart + startIdx;
 				_sectionEnd = parent._sectionStart + endIdx;
-				_start = _sections.get(_sectionStart).getStart();
+				_start = _sections.get(_sectionStart - 1).getStart();
 				_end = _sections.get(_sectionEnd).getEnd();
 				_sectionRangeFound = true;
 				break;
 			case TYPE_TEXT:
 				_textStart = parent._textStart + startIdx;
 				_textEnd = parent._textStart + endIdx;
-				_start = _text.get(_textStart).getStart();
+				_start = _text.get(_textStart - 1).getStart();
 				_end = _text.get(_textEnd).getEnd();
 				_textRangeFound = true;
 				break;
@@ -833,6 +833,12 @@ public class Range { // TODO -instantiable superclass
 
 	public Paragraph getParagraph(int index) {
         initParagraphs();
+
+        if ( index + _parStart >= _parEnd )
+            throw new IndexOutOfBoundsException( "Paragraph #" + index + " ("
+                    + (index + _parStart) + ") not in range [" + _parStart
+                    + "; " + _parEnd + ")" );
+
 		PAPX papx = _paragraphs.get(index + _parStart);
 
 		ParagraphProperties props = papx.getParagraphProperties(_doc.getStyleSheet());
@@ -880,7 +886,7 @@ public class Range { // TODO -instantiable superclass
 
 		r.initAll();
 		int tableLevel = paragraph.getTableLevel();
-		int tableEnd = r._parEnd;
+		int tableEndInclusive = r._parEnd ;
 
         if ( r._parStart != 0 )
         {
@@ -895,24 +901,30 @@ public class Range { // TODO -instantiable superclass
             }
         }
 
+        final Range overallrange = getDocument() instanceof HWPFDocument ? ((HWPFDocument) getDocument())
+                .getOverallRange() : getDocument().getRange();
         int limit = _paragraphs.size();
-        for ( ; tableEnd < limit; tableEnd++ )
+        for ( ; tableEndInclusive < limit - 1; tableEndInclusive++ )
         {
-            Paragraph next = new Paragraph( _paragraphs.get( tableEnd ), this );
+            Paragraph next = new Paragraph( _paragraphs.get( tableEndInclusive + 1 ),
+                    overallrange );
             if ( !next.isInTable() || next.getTableLevel() < tableLevel )
                 break;
         }
 
         initAll();
-		if (tableEnd > _parEnd) {
+        if ( tableEndInclusive + 1 > _parEnd )
+        {
             throw new ArrayIndexOutOfBoundsException(
                     "The table's bounds fall outside of this Range" );
         }
-		if (tableEnd < 0) {
+        if ( tableEndInclusive < 0 )
+        {
             throw new ArrayIndexOutOfBoundsException(
                     "The table's end is negative, which isn't allowed!" );
         }
-		return new Table(r._parStart, tableEnd, r._doc.getRange(), paragraph.getTableLevel());
+        return new Table( r._parStart, tableEndInclusive + 1, r._doc.getRange(),
+                paragraph.getTableLevel() );
     }
 
 	/**
@@ -989,6 +1001,10 @@ public class Range { // TODO -instantiable superclass
 	 */
 	private int[] findRange(List<? extends PropertyNode> rpl, int min, int start, int end) {
 		int x = min;
+
+        if ( rpl.size() == min )
+            return new int[] { min, min };
+
         PropertyNode node = rpl.get( x );
 
 		while (node==null || (node.getEnd() <= start && x < rpl.size() - 1)) {

src/scratchpad/src/org/apache/poi/hwpf/usermodel/Table.java

@@ -25,9 +25,9 @@ public final class Table extends Range
 
     private int _tableLevel;
 
-    Table( int startIdx, int endIdx, Range parent, int levelNum )
+    Table( int startIdxInclusive, int endIdxExclusive, Range parent, int levelNum )
     {
-        super( startIdx, endIdx, Range.TYPE_PARAGRAPH, parent );
+        super( startIdxInclusive, endIdxExclusive, Range.TYPE_PARAGRAPH, parent );
         _rows = new ArrayList<TableRow>();
         _tableLevel = levelNum;
 
@@ -41,8 +41,8 @@ public final class Table extends Range
             rowEnd++;
             if ( p.isTableRowEnd() && p.getTableLevel() == levelNum )
             {
-                _rows.add( new TableRow( rowStart, rowEnd, this, levelNum ) );
+                _rows.add( new TableRow( rowStart, rowEnd + 1, this, levelNum ) );
-                rowStart = rowEnd;
+                rowStart = rowEnd + 1;
             }
         }
     }

src/scratchpad/src/org/apache/poi/hwpf/usermodel/TableCell.java

@@ -25,9 +25,11 @@ public final class TableCell
   private int _leftEdge;
   private int _width;
 
-  public TableCell(int startIdx, int endIdx, TableRow parent, int levelNum, TableCellDescriptor tcd, int leftEdge, int width)
+    public TableCell( int startIdxInclusive, int endIdxExclusive,
+            TableRow parent, int levelNum, TableCellDescriptor tcd,
+            int leftEdge, int width )
     {
-    super(startIdx, endIdx, Range.TYPE_PARAGRAPH, parent);
+        super( startIdxInclusive, endIdxExclusive, Range.TYPE_PARAGRAPH, parent );
         _tcd = tcd;
         _leftEdge = leftEdge;
         _width = width;

src/scratchpad/src/org/apache/poi/hwpf/usermodel/TableRow.java

@@ -17,10 +17,14 @@
 
 package org.apache.poi.hwpf.usermodel;
 
-import org.apache.poi.hwpf.sprm.TableSprmUncompressor;
+import java.util.ArrayList;
+import java.util.List;
 
-public final class TableRow
+import org.apache.poi.hwpf.sprm.TableSprmUncompressor;
-  extends Paragraph
+import org.apache.poi.util.POILogFactory;
+import org.apache.poi.util.POILogger;
+
+public final class TableRow extends Paragraph
 {
     private final static char TABLE_CELL_MARK = '\u0007';
 
@@ -30,49 +34,81 @@ public final class TableRow
     private final static short SPRM_FTABLEHEADER = 0x3404;
     private final static short SPRM_DYAROWHEIGHT = (short) 0x9407;
 
+    private static final POILogger logger = POILogFactory
+            .getLogger( TableRow.class );
+
     int _levelNum;
     private TableProperties _tprops;
     private TableCell[] _cells;
 
-  public TableRow(int startIdx, int endIdx, Table parent, int levelNum)
+    public TableRow( int startIdxInclusive, int endIdxExclusive, Table parent,
+            int levelNum )
     {
-    super(startIdx, endIdx, parent);
+        super( startIdxInclusive, endIdxExclusive, parent );
 
         _tprops = TableSprmUncompressor.uncompressTAP( _papx.toByteArray(), 2 );
         _levelNum = levelNum;
-    _cells = new TableCell[_tprops.getItcMac()];
+        final short expectedCellsCount = _tprops.getItcMac();
 
-    int start = 0;
+        int lastCellStart = 0;
-    int end = 0;
+        List<TableCell> cells = new ArrayList<TableCell>( expectedCellsCount );
-
+        for ( int p = 0; p < (endIdxExclusive - startIdxInclusive); p++ )
-    for (int cellIndex = 0; cellIndex < _cells.length; cellIndex++)
         {
-      Paragraph p = getParagraph(start);
+            Paragraph paragraph = getParagraph( p );
-      String s = p.text();
+            String s = paragraph.text();
 
-      while (! ( s.length() > 0 && (s.charAt(s.length() - 1) == TABLE_CELL_MARK) ||
+            if ( s.length() > 0
-                p.isEmbeddedCellMark() && p.getTableLevel() == levelNum) && end < endIdx)
+                    && s.charAt( s.length() - 1 ) == TABLE_CELL_MARK
+                    && paragraph.getTableLevel() == levelNum )
             {
-        end++;
+                TableCellDescriptor tableCellDescriptor = _tprops.getRgtc() != null
-        p = getParagraph(end);
+                        && _tprops.getRgtc().length > cells.size() ? _tprops
-        s = p.text();
+                        .getRgtc()[cells.size()] : new TableCellDescriptor();
+                final short leftEdge = _tprops.getRgdxaCenter() != null
+                        && _tprops.getRgdxaCenter().length > cells.size() ? _tprops
+                        .getRgdxaCenter()[cells.size()] : 0;
+                final short rightEdge = _tprops.getRgdxaCenter() != null
+                        && _tprops.getRgdxaCenter().length > cells.size() + 1 ? _tprops
+                        .getRgdxaCenter()[cells.size() + 1] : 0;
+
+                TableCell tableCell = new TableCell( lastCellStart, p + 1,
+                        this, levelNum, tableCellDescriptor, leftEdge,
+                        rightEdge - leftEdge );
+                cells.add( tableCell );
+                lastCellStart = p + 1;
+            }
         }
 
-      // Create it for the correct paragraph range
+        if ( lastCellStart < (endIdxExclusive - startIdxInclusive - 1) )
-      _cells[cellIndex] = new TableCell(start, end, this, levelNum,
+        {
-                                        _tprops.getRgtc()[cellIndex],
+            TableCellDescriptor tableCellDescriptor = _tprops.getRgtc() != null
-                                        _tprops.getRgdxaCenter()[cellIndex],
+                    && _tprops.getRgtc().length > cells.size() ? _tprops
-                                        _tprops.getRgdxaCenter()[cellIndex+1]-_tprops.getRgdxaCenter()[cellIndex]);
+                    .getRgtc()[cells.size()] : new TableCellDescriptor();
-      // Now we've decided where everything is, tweak the
+            final short leftEdge = _tprops.getRgdxaCenter() != null
-      //  record of the paragraph end so that the
+                    && _tprops.getRgdxaCenter().length > cells.size() ? _tprops
-      //  paragraph level counts work
+                    .getRgdxaCenter()[cells.size()] : 0;
-      // This is a bit hacky, we really need a better fix...
+            final short rightEdge = _tprops.getRgdxaCenter() != null
-      _cells[cellIndex]._parEnd++;
+                    && _tprops.getRgdxaCenter().length > cells.size() + 1 ? _tprops
+                    .getRgdxaCenter()[cells.size() + 1] : 0;
 
-      // Next!
+            TableCell tableCell = new TableCell( lastCellStart,
-      end++;
+                    (endIdxExclusive - startIdxInclusive - 1), this, levelNum,
-      start = end;
+                    tableCellDescriptor, leftEdge, rightEdge - leftEdge );
+            cells.add( tableCell );
         }
+
+        if ( cells.size() != expectedCellsCount )
+        {
+            logger.log( POILogger.WARN,
+                    "Number of found table cells (" + cells.size()
+                            + ") for table row [" + getStartOffset() + "c; "
+                            + getEndOffset()
+                            + "c] not equals to stored property value "
+                            + expectedCellsCount );
+            _tprops.setItcMac( (short) cells.size() );
+        }
+
+        _cells = cells.toArray( new TableCell[cells.size()] );
     }
 
     public int getRowJustification()
@@ -140,31 +176,38 @@ public final class TableRow
         return _cells[index];
     }
 
-  public BorderCode getTopBorder() {
+    public BorderCode getTopBorder()
+    {
         return _tprops.getBrcBottom();
     }
 
-  public BorderCode getBottomBorder() {
+    public BorderCode getBottomBorder()
+    {
         return _tprops.getBrcBottom();
     }
 
-  public BorderCode getLeftBorder() {
+    public BorderCode getLeftBorder()
+    {
         return _tprops.getBrcLeft();
     }
 
-  public BorderCode getRightBorder() {
+    public BorderCode getRightBorder()
+    {
         return _tprops.getBrcRight();
     }
 
-  public BorderCode getHorizontalBorder() {
+    public BorderCode getHorizontalBorder()
+    {
         return _tprops.getBrcHorizontal();
     }
 
-  public BorderCode getVerticalBorder() {
+    public BorderCode getVerticalBorder()
+    {
         return _tprops.getBrcVertical();
     }
 
-  public BorderCode getBarBorder() {
+    public BorderCode getBarBorder()
+    {
         throw new UnsupportedOperationException( "not applicable for TableRow" );
     }
 

src/scratchpad/testcases/org/apache/poi/hwpf/converter/TestWordToConverterSuite.java

@@ -40,7 +40,8 @@ public class TestWordToConverterSuite
     /**
      * YK: a quick hack to exclude failing documents from the suite.
      */
-    private static List<String> failingFiles = Arrays.asList();
+    private static List<String> failingFiles = Arrays
+            .asList( "ProblemExtracting.doc" );
 
     public static Test suite()
     {

src/scratchpad/testcases/org/apache/poi/hwpf/converter/TestWordToHtmlConverter.java

@@ -68,11 +68,6 @@ public class TestWordToHtmlConverter extends TestCase
         assertTrue( result.substring( 0, 2000 ).contains( "<table>" ) );
     }
 
-    public void testBug33519() throws Exception
-    {
-        getHtmlText( "Bug33519.doc" );
-    }
-
     public void testBug46610_2() throws Exception
     {
         String result = getHtmlText( "Bug46610_2.doc" );

src/scratchpad/testcases/org/apache/poi/hwpf/usermodel/TestRange.java

@@ -79,14 +79,14 @@ public final class TestRange extends TestCase
         assertEquals( range.getStartOffset(), 0 );
         assertEquals( range.getEndOffset(), 766 );
 
-        Paragraph lastInMainRange = range.getParagraph( range.numParagraphs() );
+        Paragraph lastInMainRange = range.getParagraph( range.numParagraphs() - 1);
         assertTrue( lastInMainRange.getEndOffset() <= 766 );
 
         Section section = range.getSection( 0 );
         assertTrue( section.getEndOffset() <= 766 );
 
         Paragraph lastInMainSection = section.getParagraph( section
-                .numParagraphs() );
+                .numParagraphs() - 1);
         assertTrue( lastInMainSection.getEndOffset() <= 766 );
     }
 }