more progress with WordToFoExtractor, see Bugzilla 51351

git-svn-id: https://svn.apache.org/repos/asf/poi/trunk@1136001 13f79535-47bb-0310-9956-ffa450edef68

src/scratchpad/src/org/apache/poi/hwpf/extractor/WordToFoExtractor.java

@@ -279,8 +279,27 @@ public class WordToFoExtractor {
 	}
     }
 
-    @SuppressWarnings("unused")
+    /**
-    protected void processImage(Element currentBlock, Picture picture) {
+     * This method shall store image bytes in external file and convert it if
+     * necessary. Images shall be stored using PNG format (for bitmap) or SVG
+     * (for vector). Other formats may be not supported by your XSL FO
+     * processor.
+     * <p>
+     * Please note the
+     * {@link WordToFoUtils#setPictureProperties(Picture, Element)} method.
+     * 
+     * @param currentBlock
+     *            currently processed FO element, like <tt>fo:block</tt>. Shall
+     *            be used as parent of newly created
+     *            <tt>fo:external-graphic</tt> or
+     *            <tt>fo:instream-foreign-object</tt>
+     * @param inlined
+     *            if image is inlined
+     * @param picture
+     *            HWPF object, contained picture data and properties
+     */
+    protected void processImage(Element currentBlock, boolean inlined,
+            Picture picture) {
         // no default implementation -- skip
     }
 
@@ -329,11 +348,24 @@ public class WordToFoExtractor {
         for (int c = 0; c < charRuns; c++) {
             CharacterRun characterRun = paragraph.getCharacterRun(c);
 
+            if (hwpfDocument.getPicturesTable().hasPicture(characterRun)) {
+                Picture picture = hwpfDocument.getPicturesTable()
+                        .extractPicture(characterRun, true);
+
+                processImage(block, characterRun.text().charAt(0) == 0x01,
+                        picture);
+                continue;
+            }
+
 	    String text = characterRun.text();
 	    if (text.getBytes().length == 0)
 		continue;
 
             if (text.getBytes()[0] == FIELD_BEGIN_MARK) {
+                /*
+                 * check if we have a field with calculated image as a result.
+                 * MathType equation, for example.
+                 */
                 int skipTo = tryImageWithinField(hwpfDocument, paragraph, c,
                         block);
 
@@ -553,6 +585,7 @@ public class WordToFoExtractor {
             Paragraph paragraph, int beginMark, Element currentBlock) {
         int separatorMark = -1;
         int pictureMark = -1;
+        int pictureChar = Integer.MIN_VALUE;
         int endMark = -1;
         for (int c = beginMark + 1; c < paragraph.numCharacterRuns(); c++) {
             CharacterRun characterRun = paragraph.getCharacterRun(c);
@@ -582,12 +615,13 @@ public class WordToFoExtractor {
             }
 
             if (hwpfDocument.getPicturesTable().hasPicture(characterRun)) {
-		if (pictureMark != -1) {
+                if (c != -1) {
                     // double;
                     return beginMark;
                 }
 
                 pictureMark = c;
+                pictureChar = characterRun.text().charAt(0);
                 continue;
             }
         }
@@ -598,12 +632,12 @@ public class WordToFoExtractor {
         final CharacterRun pictureRun = paragraph.getCharacterRun(pictureMark);
         final Picture picture = hwpfDocument.getPicturesTable().extractPicture(
                 pictureRun, true);
-	processImage(currentBlock, picture);
+
+        processImage(currentBlock, pictureChar == 0x01, picture);
 
         return endMark;
     }
 
-
     /**
      * Java main() interface to interact with WordToFoExtractor
      *

src/scratchpad/src/org/apache/poi/hwpf/extractor/WordToFoUtils.java

@@ -9,6 +9,7 @@ import org.apache.poi.hwpf.usermodel.BorderCode;
 import org.apache.poi.hwpf.usermodel.CharacterProperties;
 import org.apache.poi.hwpf.usermodel.CharacterRun;
 import org.apache.poi.hwpf.usermodel.Paragraph;
+import org.apache.poi.hwpf.usermodel.Picture;
 import org.apache.poi.hwpf.usermodel.Range;
 import org.apache.poi.hwpf.usermodel.Section;
 import org.apache.poi.hwpf.usermodel.SectionProperties;
@@ -179,6 +180,31 @@ public class WordToFoUtils {
 	}
     }
 
+    public static String getJustification(int js) {
+        switch (js) {
+        case 0:
+            return "start";
+        case 1:
+            return "center";
+        case 2:
+            return "end";
+        case 3:
+        case 4:
+            return "justify";
+        case 5:
+            return "center";
+        case 6:
+            return "left";
+        case 7:
+            return "start";
+        case 8:
+            return "end";
+        case 9:
+            return "justify";
+        }
+        return "";
+    }
+
     public static String getListItemNumberLabel(int number, int format) {
 
 	if (format != 0)
@@ -283,6 +309,9 @@ public class WordToFoUtils {
                 textDecorations.append(" ");
             textDecorations.append("underline");
         }
+        if (characterRun.isVanished()) {
+            inline.setAttribute("visibility", "hidden");
+        }
         if (textDecorations.length() > 0) {
             inline.setAttribute("text-decoration", textDecorations.toString());
         }
@@ -336,39 +365,9 @@ public class WordToFoUtils {
 
     public static void setJustification(Paragraph paragraph,
             final Element element) {
-	final int justification = paragraph.getJustification();
+        String justification = getJustification(paragraph.getJustification());
-	switch (justification) {
+        if (isNotEmpty(justification))
-	case 0:
+            element.setAttribute("text-align", justification);
-	    element.setAttribute("text-align", "start");
-	    break;
-	case 1:
-	    element.setAttribute("text-align", "center");
-	    break;
-	case 2:
-	    element.setAttribute("text-align", "end");
-	    break;
-	case 3:
-	    element.setAttribute("text-align", "justify");
-	    break;
-	case 4:
-	    element.setAttribute("text-align", "justify");
-	    break;
-	case 5:
-	    element.setAttribute("text-align", "center");
-	    break;
-	case 6:
-	    element.setAttribute("text-align", "left");
-	    break;
-	case 7:
-	    element.setAttribute("text-align", "start");
-	    break;
-	case 8:
-	    element.setAttribute("text-align", "end");
-	    break;
-	case 9:
-	    element.setAttribute("text-align", "justify");
-	    break;
-	}
     }
 
     public static void setParagraphProperties(Paragraph paragraph, Element block) {
@@ -399,6 +398,53 @@ public class WordToFoUtils {
 	block.setAttribute("white-space-collapse", "false");
     }
 
+    public static void setPictureProperties(Picture picture,
+            Element graphicElement) {
+        final int aspectRatioX = picture.getAspectRatioX();
+        final int aspectRatioY = picture.getAspectRatioY();
+
+        if (aspectRatioX > 0) {
+            graphicElement.setAttribute("content-width", ((picture.getDxaGoal()
+                    * aspectRatioX / 100) / WordToFoUtils.TWIPS_PER_PT)
+                    + "pt");
+        } else
+            graphicElement.setAttribute("content-width",
+                    (picture.getDxaGoal() / WordToFoUtils.TWIPS_PER_PT) + "pt");
+
+        if (aspectRatioY > 0)
+            graphicElement
+                    .setAttribute("content-height", ((picture.getDyaGoal()
+                            * aspectRatioY / 100) / WordToFoUtils.TWIPS_PER_PT)
+                            + "pt");
+        else
+            graphicElement.setAttribute("content-height",
+                    (picture.getDyaGoal() / WordToFoUtils.TWIPS_PER_PT) + "pt");
+
+        if (aspectRatioX <= 0 || aspectRatioY <= 0) {
+            graphicElement.setAttribute("scaling", "uniform");
+        } else {
+            graphicElement.setAttribute("scaling", "non-uniform");
+        }
+
+        graphicElement.setAttribute("vertical-align", "text-bottom");
+
+        if (picture.getDyaCropTop() != 0 || picture.getDxaCropRight() != 0
+                || picture.getDyaCropBottom() != 0
+                || picture.getDxaCropLeft() != 0) {
+            int rectTop = picture.getDyaCropTop() / WordToFoUtils.TWIPS_PER_PT;
+            int rectRight = picture.getDxaCropRight()
+                    / WordToFoUtils.TWIPS_PER_PT;
+            int rectBottom = picture.getDyaCropBottom()
+                    / WordToFoUtils.TWIPS_PER_PT;
+            int rectLeft = picture.getDxaCropLeft()
+                    / WordToFoUtils.TWIPS_PER_PT;
+            graphicElement.setAttribute("clip", "rect(" + rectTop + "pt, "
+                    + rectRight + "pt, " + rectBottom + "pt, " + rectLeft
+                    + "pt)");
+            graphicElement.setAttribute("oveerflow", "hidden");
+        }
+    }
+
     public static void setTableCellProperties(TableRow tableRow,
 	    TableCell tableCell, Element element, boolean toppest,
 	    boolean bottomest, boolean leftest, boolean rightest) {