Fix for bug #48245 - tweak HWPF table cell detection to work across more files

git-svn-id: https://svn.apache.org/repos/asf/poi/trunk@953694 13f79535-47bb-0310-9956-ffa450edef68

src/documentation/content/xdocs/status.xml

@@ -34,6 +34,7 @@
 
     <changes>
         <release version="3.7-SNAPSHOT" date="2010-??-??">
+           <action dev="POI-DEVELOPERS" type="add">48245 - tweak HWPF table cell detection to work across more files</action>
            <action dev="POI-DEVELOPERS" type="add">48996 - initial support for External Name References in HSSF formula evaluation</action>
            <action dev="POI-DEVELOPERS" type="fix">46664 - fix up Tab IDs when adding new sheets, so that print areas don't end up invalid</action>
            <action dev="POI-DEVELOPERS" type="fix">45269 - improve replaceText on HWPF ranges</action>

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

@@ -17,6 +17,7 @@
 
 package org.apache.poi.hwpf.usermodel;
 
+import org.apache.poi.hwpf.model.PropertyNode;
 import org.apache.poi.hwpf.sprm.TableSprmUncompressor;
 
 public final class TableRow
@@ -57,10 +58,19 @@ public final class TableRow
         p = getParagraph(end);
         s = p.text();
       }
-      _cells[cellIndex] = new TableCell(start, end+1, this, levelNum,
+
+      // Create it for the correct paragraph range
+      _cells[cellIndex] = new TableCell(start, end, this, levelNum,
                                         _tprops.getRgtc()[cellIndex],
                                         _tprops.getRgdxaCenter()[cellIndex],
                                         _tprops.getRgdxaCenter()[cellIndex+1]-_tprops.getRgdxaCenter()[cellIndex]);
+      // Now we've decided where everything is, tweak the
+      //  record of the paragraph end so that the
+      //  paragraph level counts work
+      // This is a bit hacky, we really need a better fix...
+      _cells[cellIndex]._parEnd++;
+      
+      // Next!
       end++;
       start = end;
     }

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

@@ -80,6 +80,18 @@ public final class TestProblems extends HWPFTestCase {
       assertEquals(0, p.getStartOffset());
       assertEquals(20, p.getEndOffset());
 
+      // Check a few bits of the table directly
+      assertEquals("One paragraph is ok\7", r.getParagraph(0).text());
+      assertEquals("First para is ok\r", r.getParagraph(1).text());
+      assertEquals("Second paragraph is skipped\7", r.getParagraph(2).text());
+      assertEquals("One paragraph is ok\7", r.getParagraph(3).text());
+      assertEquals("\7", r.getParagraph(4).text());
+      assertEquals("\r", r.getParagraph(5).text());
+      for(int i=0; i<=5; i++) {
+         assertFalse(r.getParagraph(i).usesUnicode());
+      }
+
+
       // Get the table
       Table t = r.getTable(p);
 
@@ -87,11 +99,24 @@ public final class TestProblems extends HWPFTestCase {
       assertEquals(1, t.numRows());
       TableRow row = t.getRow(0);
 
+      // sanity check our row
+      assertEquals(5, row.numParagraphs());
+      assertEquals(0, row._parStart);
+      assertEquals(5, row._parEnd);
+      assertEquals(0, row.getStartOffset());
+      assertEquals(87, row.getEndOffset());
+
+
       // get the first cell
       TableCell cell = row.getCell(0);
       // First cell should have one paragraph
       assertEquals(1, cell.numParagraphs());
       assertEquals("One paragraph is ok\7", cell.getParagraph(0).text());
+      assertEquals(0, cell._parStart);
+      assertEquals(1, cell._parEnd);
+      assertEquals(0, cell.getStartOffset());
+      assertEquals(20, cell.getEndOffset());
+
 
       // get the second
       cell = row.getCell(1);
@@ -99,12 +124,21 @@ public final class TestProblems extends HWPFTestCase {
       assertEquals(2, cell.numParagraphs());
       assertEquals("First para is ok\r", cell.getParagraph(0).text());
       assertEquals("Second paragraph is skipped\7", cell.getParagraph(1).text());
+      assertEquals(1, cell._parStart);
+      assertEquals(3, cell._parEnd);
+      assertEquals(20, cell.getStartOffset());
+      assertEquals(65, cell.getEndOffset());
+
 
       // get the last cell
       cell = row.getCell(2);
       // Last cell should have one paragraph
       assertEquals(1, cell.numParagraphs());
       assertEquals("One paragraph is ok\7", cell.getParagraph(0).text());
+      assertEquals(3, cell._parStart);
+      assertEquals(4, cell._parEnd);
+      assertEquals(65, cell.getStartOffset());
+      assertEquals(85, cell.getEndOffset());
    }
 
    public void testRangeDelete() {
@@ -198,4 +232,154 @@ public final class TestProblems extends HWPFTestCase {
          }
       }
    }
+
+   /**
+    * Bug #48245 - don't include the text from the
+    *  next cell in the current one
+    */
+   public void testTableIterator() throws Exception {
+      HWPFDocument doc = HWPFTestDataSamples.openSampleFile("simple-table2.doc");
+      Range r = doc.getRange();
+
+      // Check the text is as we'd expect
+      assertEquals(13, r.numParagraphs());
+      assertEquals("Row 1/Cell 1\u0007", r.getParagraph(0).text());
+      assertEquals("Row 1/Cell 2\u0007", r.getParagraph(1).text());
+      assertEquals("Row 1/Cell 3\u0007", r.getParagraph(2).text());
+      assertEquals("\u0007", r.getParagraph(3).text());
+      assertEquals("Row 2/Cell 1\u0007", r.getParagraph(4).text());
+      assertEquals("Row 2/Cell 2\u0007", r.getParagraph(5).text());
+      assertEquals("Row 2/Cell 3\u0007", r.getParagraph(6).text());
+      assertEquals("\u0007", r.getParagraph(7).text());
+      assertEquals("Row 3/Cell 1\u0007", r.getParagraph(8).text());
+      assertEquals("Row 3/Cell 2\u0007", r.getParagraph(9).text());
+      assertEquals("Row 3/Cell 3\u0007", r.getParagraph(10).text());
+      assertEquals("\u0007", r.getParagraph(11).text());
+      assertEquals("\r", r.getParagraph(12).text());
+      for(int i=0; i<=12; i++) {
+         assertFalse(r.getParagraph(i).usesUnicode());
+      }
+
+      Paragraph p;
+
+      // Take a look in detail at the first couple of
+      //  paragraphs
+      p = r.getParagraph(0);
+      assertEquals(1,  p.numParagraphs());
+      assertEquals(0,  p.getStartOffset());
+      assertEquals(13, p.getEndOffset());
+      assertEquals(0,  p._parStart);
+      assertEquals(1,  p._parEnd);
+
+      p = r.getParagraph(1);
+      assertEquals(1,  p.numParagraphs());
+      assertEquals(13, p.getStartOffset());
+      assertEquals(26, p.getEndOffset());
+      assertEquals(1,  p._parStart);
+      assertEquals(2,  p._parEnd);
+
+      p = r.getParagraph(2);
+      assertEquals(1,  p.numParagraphs());
+      assertEquals(26, p.getStartOffset());
+      assertEquals(39, p.getEndOffset());
+      assertEquals(2,  p._parStart);
+      assertEquals(3,  p._parEnd);
+
+
+      // Now look at the table
+      Table table = r.getTable(r.getParagraph(0));
+      assertEquals(3, table.numRows());
+
+      TableRow row;
+      TableCell cell;
+
+
+      row = table.getRow(0);
+      assertEquals(0, row._parStart);
+      assertEquals(4, row._parEnd);
+
+      cell = row.getCell(0);
+      assertEquals(1, cell.numParagraphs());
+      assertEquals(0, cell._parStart);
+      assertEquals(1, cell._parEnd);
+      assertEquals(0, cell.getStartOffset());
+      assertEquals(13, cell.getEndOffset());
+      assertEquals("Row 1/Cell 1\u0007", cell.text());
+
+      cell = row.getCell(1);
+      assertEquals(1, cell.numParagraphs());
+      assertEquals(1, cell._parStart);
+      assertEquals(2, cell._parEnd);
+      assertEquals(13, cell.getStartOffset());
+      assertEquals(26, cell.getEndOffset());
+      assertEquals("Row 1/Cell 2\u0007", cell.text());
+
+      cell = row.getCell(2);
+      assertEquals(1, cell.numParagraphs());
+      assertEquals(2, cell._parStart);
+      assertEquals(3, cell._parEnd);
+      assertEquals(26, cell.getStartOffset());
+      assertEquals(39, cell.getEndOffset());
+      assertEquals("Row 1/Cell 3\u0007", cell.text());
+
+
+      // Onto row #2
+      row = table.getRow(1);
+      assertEquals(4, row._parStart);
+      assertEquals(8, row._parEnd);
+
+      cell = row.getCell(0);
+      assertEquals(1, cell.numParagraphs());
+      assertEquals(4, cell._parStart);
+      assertEquals(5, cell._parEnd);
+      assertEquals(40, cell.getStartOffset());
+      assertEquals(53, cell.getEndOffset());
+      assertEquals("Row 2/Cell 1\u0007", cell.text());
+
+      cell = row.getCell(1);
+      assertEquals(1, cell.numParagraphs());
+      assertEquals(5, cell._parStart);
+      assertEquals(6, cell._parEnd);
+      assertEquals(53, cell.getStartOffset());
+      assertEquals(66, cell.getEndOffset());
+      assertEquals("Row 2/Cell 2\u0007", cell.text());
+
+      cell = row.getCell(2);
+      assertEquals(1, cell.numParagraphs());
+      assertEquals(6, cell._parStart);
+      assertEquals(7, cell._parEnd);
+      assertEquals(66, cell.getStartOffset());
+      assertEquals(79, cell.getEndOffset());
+      assertEquals("Row 2/Cell 3\u0007", cell.text());
+
+
+      // Finally row 3
+      row = table.getRow(2);
+      assertEquals(8, row._parStart);
+      assertEquals(12, row._parEnd);
+
+      cell = row.getCell(0);
+      assertEquals(1, cell.numParagraphs());
+      assertEquals(8, cell._parStart);
+      assertEquals(9, cell._parEnd);
+      assertEquals(80, cell.getStartOffset());
+      assertEquals(93, cell.getEndOffset());
+      assertEquals("Row 3/Cell 1\u0007", cell.text());
+
+      cell = row.getCell(1);
+      assertEquals(1, cell.numParagraphs());
+      assertEquals(9, cell._parStart);
+      assertEquals(10, cell._parEnd);
+      assertEquals(93, cell.getStartOffset());
+      assertEquals(106, cell.getEndOffset());
+      assertEquals("Row 3/Cell 2\u0007", cell.text());
+
+      cell = row.getCell(2);
+      assertEquals(1, cell.numParagraphs());
+      assertEquals(10, cell._parStart);
+      assertEquals(11, cell._parEnd);
+      assertEquals(106, cell.getStartOffset());
+      assertEquals(119, cell.getEndOffset());
+      assertEquals("Row 3/Cell 3\u0007", cell.text());
+   }
 }