compact HTML output of WordToHtmlConverter

git-svn-id: https://svn.apache.org/repos/asf/poi/trunk@1148269 13f79535-47bb-0310-9956-ffa450edef68

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

@@ -70,20 +70,20 @@ public class WordToHtmlConverter extends AbstractWordConverter
         float bottomMargin = section.getMarginBottom() / TWIPS_PER_INCH;
 
         String style = "margin: " + topMargin + "in " + rightMargin + "in "
-                + bottomMargin + "in " + leftMargin + "in; ";
+                + bottomMargin + "in " + leftMargin + "in;";
 
         if ( section.getNumColumns() > 1 )
         {
-            style += "column-count: " + ( section.getNumColumns() ) + "; ";
+            style += "column-count: " + ( section.getNumColumns() ) + ";";
             if ( section.isColumnsEvenlySpaced() )
             {
                 float distance = section.getDistanceBetweenColumns()
                         / TWIPS_PER_INCH;
-                style += "column-gap: " + distance + "in; ";
+                style += "column-gap: " + distance + "in;";
             }
             else
             {
-                style += "column-gap: 0.25in; ";
+                style += "column-gap: 0.25in;";
             }
         }
         return style;
@@ -160,6 +160,7 @@ public class WordToHtmlConverter extends AbstractWordConverter
 
     public Document getDocument()
     {
+        htmlDocumentFacade.updateStylesheet();
         return htmlDocumentFacade.getDocument();
     }
 
@@ -178,24 +179,25 @@ public class WordToHtmlConverter extends AbstractWordConverter
                 && !WordToHtmlUtils.equals( triplet.fontName,
                         blockProperies.pFontName ) )
         {
-            style.append( "font-family: " + triplet.fontName + "; " );
+            style.append( "font-family:" + triplet.fontName + ";" );
         }
         if ( characterRun.getFontSize() / 2 != blockProperies.pFontSize )
         {
-            style.append( "font-size: " + characterRun.getFontSize() / 2 + "; " );
+            style.append( "font-size:" + characterRun.getFontSize() / 2 + "pt;" );
         }
         if ( triplet.bold )
         {
-            style.append( "font-weight: bold; " );
+            style.append( "font-weight:bold;" );
         }
         if ( triplet.italic )
         {
-            style.append( "font-style: italic; " );
+            style.append( "font-style:italic;" );
         }
 
         WordToHtmlUtils.addCharactersProperties( characterRun, style );
         if ( style.length() != 0 )
-            span.setAttribute( "style", style.toString() );
+            span.setAttribute( "class", htmlDocumentFacade.getOrCreateCssClass(
+                    span.getTagName(), "s", style.toString() ) );
 
         Text textNode = htmlDocumentFacade.createText( text );
         span.appendChild( textNode );
@@ -312,22 +314,28 @@ public class WordToHtmlConverter extends AbstractWordConverter
             float visibleHeight = Math.max( 0, imageHeight - cropTop
                     - cropBottom );
 
-            root = htmlDocumentFacade.document.createElement( "div" );
+            root = htmlDocumentFacade.createBlock();
-            root.setAttribute( "style", "vertical-align:text-bottom;width:"
+            root.setAttribute( "class", htmlDocumentFacade.getOrCreateCssClass(
-                    + visibleWidth + "in;height:" + visibleHeight + "in;" );
+                    root.getTagName(), "d", "vertical-align:text-bottom;width:"
+                            + visibleWidth + "in;height:" + visibleHeight
+                            + "in;" ) );
 
             // complex
-            Element inner = htmlDocumentFacade.document.createElement( "div" );
+            Element inner = htmlDocumentFacade.createBlock();
-            inner.setAttribute( "style", "position:relative;width:"
+            inner.setAttribute( "class", htmlDocumentFacade
-                    + visibleWidth + "in;height:" + visibleHeight
+                    .getOrCreateCssClass( inner.getTagName(), "d",
-                    + "in;overflow:hidden;" );
+                            "position:relative;width:" + visibleWidth
+                                    + "in;height:" + visibleHeight
+                                    + "in;overflow:hidden;" ) );
             root.appendChild( inner );
 
             Element image = htmlDocumentFacade.document.createElement( "img" );
             image.setAttribute( "src", imageSourcePath );
-            image.setAttribute( "style", "position:absolute;left:-" + cropLeft
+            image.setAttribute( "class", htmlDocumentFacade
-                    + ";top:-" + cropTop + ";width:" + imageWidth
+                    .getOrCreateCssClass( image.getTagName(), "i",
-                    + "in;height:" + imageHeight + "in;" );
+                            "position:absolute;left:-" + cropLeft + ";top:-"
+                                    + cropTop + ";width:" + imageWidth
+                                    + "in;height:" + imageHeight + "in;" ) );
             inner.appendChild( image );
 
             style.append( "overflow:hidden;" );
@@ -414,7 +422,10 @@ public class WordToHtmlConverter extends AbstractWordConverter
         }
 
         if ( style.length() > 0 )
-            pElement.setAttribute( "style", style.toString() );
+            pElement.setAttribute(
+                    "class",
+                    htmlDocumentFacade.getOrCreateCssClass(
+                            pElement.getTagName(), "p", style.toString() ) );
 
         return;
     }
@@ -422,8 +433,9 @@ public class WordToHtmlConverter extends AbstractWordConverter
     protected void processSection( HWPFDocumentCore wordDocument,
             Section section, int sectionCounter )
     {
-        Element div = htmlDocumentFacade.document.createElement( "div" );
+        Element div = htmlDocumentFacade.createBlock();
-        div.setAttribute( "style", getSectionStyle( section ) );
+        div.setAttribute( "class", htmlDocumentFacade.getOrCreateCssClass(
+                div.getTagName(), "d", getSectionStyle( section ) ) );
         htmlDocumentFacade.body.appendChild( div );
 
         processSectionParagraphes( wordDocument, div, section,
@@ -434,8 +446,9 @@ public class WordToHtmlConverter extends AbstractWordConverter
     protected void processSingleSection( HWPFDocumentCore wordDocument,
             Section section )
     {
-        htmlDocumentFacade.body.setAttribute( "style",
+        htmlDocumentFacade.body
-                getSectionStyle( section ) );
+                .setAttribute( "class", htmlDocumentFacade.getOrCreateCssClass(
+                        "body", "b", getSectionStyle( section ) ) );
 
         processSectionParagraphes( wordDocument, htmlDocumentFacade.body,
                 section, Integer.MIN_VALUE );
@@ -538,15 +551,19 @@ public class WordToHtmlConverter extends AbstractWordConverter
                             .createParagraph() );
                 }
                 if ( tableCellStyle.length() > 0 )
-                    tableCellElement.setAttribute( "style",
+                    tableCellElement.setAttribute( "class", htmlDocumentFacade
-                            tableCellStyle.toString() );
+                            .getOrCreateCssClass(
+                                    tableCellElement.getTagName(),
+                                    tableCellElement.getTagName(),
+                                    tableCellStyle.toString() ) );
 
                 tableRowElement.appendChild( tableCellElement );
             }
 
             if ( tableRowStyle.length() > 0 )
-                tableRowElement
+                tableRowElement.setAttribute( "class", htmlDocumentFacade
-                        .setAttribute( "style", tableRowStyle.toString() );
+                        .getOrCreateCssClass( "tr", "r",
+                                tableRowStyle.toString() ) );
 
             if ( tableRow.isTableHeader() )
             {

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

@@ -26,7 +26,7 @@ public class WordToHtmlUtils extends AbstractWordUtils
 {
     public static void addBold( final boolean bold, StringBuilder style )
     {
-        style.append( "font-weight: " + ( bold ? "bold" : "normal" ) + ";" );
+        style.append( "font-weight:" + ( bold ? "bold" : "normal" ) + ";" );
     }
 
     public static void addBorder( BorderCode borderCode, String where,
@@ -37,21 +37,21 @@ public class WordToHtmlUtils extends AbstractWordUtils
 
         if ( isEmpty( where ) )
         {
-            style.append( "border-style: " + getBorderType( borderCode ) + "; " );
+            style.append( "border:" );
-            style.append( "border-color: " + getColor( borderCode.getColor() )
-                    + "; " );
-            style.append( "border-width: " + getBorderWidth( borderCode )
-                    + "; " );
         }
         else
         {
-            style.append( "border-" + where + "-style: "
+            style.append( "border-" );
-                    + getBorderType( borderCode ) + "; " );
+            style.append( where );
-            style.append( "border-" + where + "-color: "
-                    + getColor( borderCode.getColor() ) + "; " );
-            style.append( "border-" + where + "-width: "
-                    + getBorderWidth( borderCode ) + "; " );
         }
+
+        style.append( ":" );
+        style.append( getBorderWidth( borderCode ) );
+        style.append( ' ' );
+        style.append( getBorderType( borderCode ) );
+        style.append( ' ' );
+        style.append( getColor( borderCode.getColor() ) );
+        style.append( ';' );
     }
 
     public static void addCharactersProperties(
@@ -61,43 +61,43 @@ public class WordToHtmlUtils extends AbstractWordUtils
 
         if ( characterRun.isCapitalized() )
         {
-            style.append( "text-transform: uppercase; " );
+            style.append( "text-transform:uppercase;" );
         }
         if ( characterRun.isHighlighted() )
         {
-            style.append( "background-color: "
+            style.append( "background-color:"
-                    + getColor( characterRun.getHighlightedColor() ) + "; " );
+                    + getColor( characterRun.getHighlightedColor() ) + ";" );
         }
         if ( characterRun.isStrikeThrough() )
         {
-            style.append( "text-decoration: line-through; " );
+            style.append( "text-decoration:line-through;" );
         }
         if ( characterRun.isShadowed() )
         {
-            style.append( "text-shadow: " + characterRun.getFontSize() / 24
+            style.append( "text-shadow:" + characterRun.getFontSize() / 24
-                    + "pt; " );
+                    + "pt;" );
         }
         if ( characterRun.isSmallCaps() )
         {
-            style.append( "font-variant: small-caps; " );
+            style.append( "font-variant:small-caps;" );
         }
         if ( characterRun.getSubSuperScriptIndex() == 1 )
         {
-            style.append( "baseline-shift: super; " );
+            style.append( "baseline-shift:super;" );
-            style.append( "font-size: smaller; " );
+            style.append( "font-size:smaller;" );
         }
         if ( characterRun.getSubSuperScriptIndex() == 2 )
         {
-            style.append( "baseline-shift: sub; " );
+            style.append( "baseline-shift:sub;" );
-            style.append( "font-size: smaller; " );
+            style.append( "font-size:smaller;" );
         }
         if ( characterRun.getUnderlineCode() > 0 )
         {
-            style.append( "text-decoration: underline; " );
+            style.append( "text-decoration:underline;" );
         }
         if ( characterRun.isVanished() )
         {
-            style.append( "visibility: hidden; " );
+            style.append( "visibility:hidden;" );
         }
     }
 
@@ -107,12 +107,12 @@ public class WordToHtmlUtils extends AbstractWordUtils
         if ( isEmpty( fontFamily ) )
             return;
 
-        style.append( "font-family: " + fontFamily + "; " );
+        style.append( "font-family:" + fontFamily + ";" );
     }
 
     public static void addFontSize( final int fontSize, StringBuilder style )
     {
-        style.append( "font-size: " + fontSize + "pt; " );
+        style.append( "font-size:" + fontSize + "pt;" );
     }
 
     public static void addIndent( Paragraph paragraph, StringBuilder style )
@@ -130,7 +130,7 @@ public class WordToHtmlUtils extends AbstractWordUtils
         if ( twipsValue == 0 )
             return;
 
-        style.append( cssName + ": " + ( twipsValue / TWIPS_PER_PT ) + "pt; " );
+        style.append( cssName + ":" + ( twipsValue / TWIPS_PER_PT ) + "pt;" );
     }
 
     public static void addJustification( Paragraph paragraph,
@@ -138,7 +138,7 @@ public class WordToHtmlUtils extends AbstractWordUtils
     {
         String justification = getJustification( paragraph.getJustification() );
         if ( isNotEmpty( justification ) )
-            style.append( "text-align: " + justification + "; " );
+            style.append( "text-align:" + justification + ";" );
     }
 
     public static void addParagraphProperties( Paragraph paragraph,
@@ -154,20 +154,20 @@ public class WordToHtmlUtils extends AbstractWordUtils
 
         if ( paragraph.pageBreakBefore() )
         {
-            style.append( "break-before: page; " );
+            style.append( "break-before:page;" );
         }
 
-        style.append( "hyphenate: "
+        style.append( "hyphenate:"
-                + ( paragraph.isAutoHyphenated() ? "auto" : "none" ) + "; " );
+                + ( paragraph.isAutoHyphenated() ? "auto" : "none" ) + ";" );
 
         if ( paragraph.keepOnPage() )
         {
-            style.append( "keep-together.within-page: always; " );
+            style.append( "keep-together.within-page:always;" );
         }
 
         if ( paragraph.keepWithNext() )
         {
-            style.append( "keep-with-next.within-page: always; " );
+            style.append( "keep-with-next.within-page:always;" );
         }
     }
 
@@ -175,12 +175,12 @@ public class WordToHtmlUtils extends AbstractWordUtils
             TableCell tableCell, boolean toppest, boolean bottomest,
             boolean leftest, boolean rightest, StringBuilder style )
     {
-        style.append( "width: " + ( tableCell.getWidth() / TWIPS_PER_INCH )
+        style.append( "width:" + ( tableCell.getWidth() / TWIPS_PER_INCH )
-                + "in; " );
+                + "in;" );
-        style.append( "padding-start: "
+        style.append( "padding-start:"
-                + ( tableRow.getGapHalf() / TWIPS_PER_INCH ) + "in; " );
+                + ( tableRow.getGapHalf() / TWIPS_PER_INCH ) + "in;" );
-        style.append( "padding-end: "
+        style.append( "padding-end:"
-                + ( tableRow.getGapHalf() / TWIPS_PER_INCH ) + "in; " );
+                + ( tableRow.getGapHalf() / TWIPS_PER_INCH ) + "in;" );
 
         BorderCode top = tableCell.getBrcTop() != null
                 && tableCell.getBrcTop().getBorderType() != 0 ? tableCell
@@ -211,12 +211,12 @@ public class WordToHtmlUtils extends AbstractWordUtils
     {
         if ( tableRow.getRowHeight() > 0 )
         {
-            style.append( "height: "
+            style.append( "height:"
-                    + ( tableRow.getRowHeight() / TWIPS_PER_INCH ) + "in; " );
+                    + ( tableRow.getRowHeight() / TWIPS_PER_INCH ) + "in;" );
         }
         if ( !tableRow.cantSplit() )
         {
-            style.append( "keep-together: always; " );
+            style.append( "keep-together:always;" );
         }
     }
 

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

@@ -92,7 +92,7 @@ public class TestWordToHtmlConverter extends TestCase
     public void testAIOOBTap() throws Exception
     {
         String result = getHtmlText( "AIOOB-Tap.doc" );
-        assertContains( result.substring( 0, 2000 ), "<table>" );
+        assertContains( result.substring( 0, 6000 ), "<table>" );
     }
 
     public void testBug33519() throws Exception