HWPF: Improve reading of auto-saved ("complex") document

git-svn-id: https://svn.apache.org/repos/asf/poi/trunk@960587 13f79535-47bb-0310-9956-ffa450edef68

diff --git a/src/documentation/content/xdocs/status.xml b/src/documentation/content/xdocs/status.xml
index acea33a3af608f03aa6769eae9b5df2696197d65..8e5199d0ae526a20074d8d642ca2e660a3cf11d5 100644 (file)
--- a/src/documentation/content/xdocs/status.xml
+++ b/src/documentation/content/xdocs/status.xml
@@ -39,6 +39,7 @@
            <action dev="POI-DEVELOPERS" type="add">49508 - Allow the addition of paragraphs to XWPF Table Cells</action>
            <action dev="POI-DEVELOPERS" type="fix">49446 - Don't consider 17.16.23 field codes as properly part of the paragraph's text</action>
            <action dev="POI-DEVELOPERS" type="fix">XSLFSlideShow shouldn't break on .thmx (theme) files. Support for them is still very limited though</action>
+           <action dev="POI-DEVELOPERS" type="fix">HWPF: Improve reading of auto-saved ("complex") documents</action>
         </release>
         <release version="3.7-beta1" date="2010-06-20">
            <action dev="POI-DEVELOPERS" type="fix">49432 - Lazy caching of XSSFComment CTComment objects by reference, to make repeated comment searching faster</action>

diff --git a/src/scratchpad/src/org/apache/poi/hwpf/HWPFOldDocument.java b/src/scratchpad/src/org/apache/poi/hwpf/HWPFOldDocument.java
index 211dc9a6b7aeb2e613491a67ae0f8ed4d1f427c8..4226c7f492de777cb4884f51c3edcef56444fdf3 100644 (file)
--- a/src/scratchpad/src/org/apache/poi/hwpf/HWPFOldDocument.java
+++ b/src/scratchpad/src/org/apache/poi/hwpf/HWPFOldDocument.java
@@ -86,7 +86,7 @@ public class HWPFOldDocument extends HWPFDocumentCore {
             TextPiece tp = new TextPiece(
                     0, textData.length, textData, pd, 0
             );
-            tpt.getTextPieces().add(tp);
+            tpt.add(tp);
             
             text.append(tp.getStringBuffer());
         }

diff --git a/src/scratchpad/src/org/apache/poi/hwpf/model/TextPieceTable.java b/src/scratchpad/src/org/apache/poi/hwpf/model/TextPieceTable.java
index 81849c6fb46c1744f67dca0371dc24e698abbf58..fae08e03f5e903b0e347d703ecc37cfb1bb78fe5 100644 (file)
--- a/src/scratchpad/src/org/apache/poi/hwpf/model/TextPieceTable.java
+++ b/src/scratchpad/src/org/apache/poi/hwpf/model/TextPieceTable.java
@@ -17,14 +17,15 @@
 
 package org.apache.poi.hwpf.model;
 
+import org.apache.poi.hwpf.model.io.HWPFOutputStream;
+import org.apache.poi.poifs.common.POIFSConstants;
+
 import java.io.IOException;
 import java.util.ArrayList;
-import java.util.Arrays;
+import java.util.Collections;
+import java.util.Comparator;
 import java.util.List;
 
-import org.apache.poi.hwpf.model.io.HWPFOutputStream;
-import org.apache.poi.poifs.common.POIFSConstants;
-
 /**
  * The piece table for matching up character positions to bits of text. This
  * mostly works in bytes, but the TextPieces themselves work in characters. This
@@ -34,6 +35,7 @@ import org.apache.poi.poifs.common.POIFSConstants;
  */
 public final class TextPieceTable implements CharIndexTranslator {
        protected ArrayList<TextPiece> _textPieces = new ArrayList<TextPiece>();
+    protected ArrayList<TextPiece> _textPiecesFCOrder = new ArrayList<TextPiece>();
        // int _multiple;
        int _cpMin;
 
@@ -96,11 +98,9 @@ public final class TextPieceTable implements CharIndexTranslator {
 
                // In the interest of our sanity, now sort the text pieces
                // into order, if they're not already
-               TextPiece[] tp = _textPieces.toArray(new TextPiece[_textPieces.size()]);
-               Arrays.sort(tp);
-               for (int i = 0; i < tp.length; i++) {
-                       _textPieces.set(i, tp[i]);
-               }
+        Collections.sort(_textPieces);
+        _textPiecesFCOrder = new ArrayList<TextPiece>(_textPieces);
+        Collections.sort(_textPiecesFCOrder, new FCComparator());
        }
 
        public int getCpMin() {
@@ -111,6 +111,13 @@ public final class TextPieceTable implements CharIndexTranslator {
                return _textPieces;
        }
 
+    public void add(TextPiece piece) {
+        _textPieces.add(piece);
+        _textPiecesFCOrder.add(piece);
+        Collections.sort(_textPieces);
+        Collections.sort(_textPiecesFCOrder, new FCComparator());
+    }
+
        /**
         * Is the text at the given Character offset unicode, or plain old ascii? In
         * a very evil fashion, you have to actually know this to make sense of
@@ -238,7 +245,7 @@ public final class TextPieceTable implements CharIndexTranslator {
        public int getCharIndex(int bytePos) {
                int charCount = 0;
 
-        for(TextPiece tp : _textPieces) {
+        for(TextPiece tp : _textPiecesFCOrder) {
                        int pieceStart = tp.getPieceDescriptor().getFilePosition();
                        if (pieceStart >= bytePos) {
                                break;
@@ -259,4 +266,15 @@ public final class TextPieceTable implements CharIndexTranslator {
                return charCount;
        }
 
+    private static class FCComparator implements Comparator<TextPiece> {
+        public int compare(TextPiece textPiece, TextPiece textPiece1) {
+            if (textPiece.getPieceDescriptor().fc>textPiece1.getPieceDescriptor().fc) {
+                return 1;
+            } else if (textPiece.getPieceDescriptor().fc<textPiece1.getPieceDescriptor().fc) {
+                return -1;
+            } else {
+                return 0;
+            }
+        }
+    }
 }