Problem with nested list bullets in XHTMLImporter

[Michael]

I have an application in which I'm writing a HTML string to a docx file using XHTMLImporter.
I'm using the convert method from XHTMLImporter and everything works fine, but in the case of a nested list, the nested bullets have the same style (dot) and not the default styles from Word. In the document.xml file, nested paragraphs have the <w:ilvl w:val="X"/> element properly incremented and are indented correctly.
Can I have a tip on how to make nested bullet points change the bullet style?

[jason]

You have 2 basic options:
1. apply pre-existing Word styles (sounds like this is what you want to do?)
2. rely on the XHTML only

Do you have control over the format of the incoming XHTML?

Treating "2. rely on the XHTML only" first :-

Try setting @list-style-type on the relevant ul|ol; see further See further https://developer.mozilla.org/en-US/doc ... style-type

This triggers getLvlTextFromCSSListStyleType or getNumberFormatFromCSSListStyleType at https://github.com/plutext/docx4j-Impor ... .java#L243

Syntax: [ Download ] [ Hide ]

Syntax: [ Download ] [ Show ]

Using java Syntax Highlighting

        private String getLvlTextFromCSSListStyleType(String listStyleType, int level) {

                if ( listStyleType.equals("disc")) {
                        return "";
                }
                if ( listStyleType.equals("circle")) {
                        return "o";
                }
                if ( listStyleType.equals("square")) {
                        return "";
                }

                return "%"+level+".";
        }
 

Parsed in 0.015 seconds, using GeSHi 1.0.8.4

There is scope to enhance this method to support other options.

Turning now to "1. apply pre-existing Word styles ":-

At https://github.com/plutext/docx4j-Impor ... java#L1276

Syntax: [ Download ] [ Hide ]

Syntax: [ Download ] [ Show ]

Using java Syntax Highlighting

} else if (isListItem(blockBox.getElement())
                                /*  <li class="ListParagraph Normal DocDefaults " style="display: list-item;text-align: justify;"><span style="">Level 1</span></li>

                                 *   <li>
                                            <p>Item 2</p>
                                            DON"T TRIGGER THIS LINE
                                        </li>
                                 */
                                && !(blockBox instanceof com.openhtmltopdf.render.AnonymousBlockBox)) {

                    // Paragraph level styling
                //P currentP = this.getCurrentParagraph(true);
               
                        // You'll get an NPE here if you have li which isn't in ol|ul
                listHelper.peekListItemStateStack().init();
               
                PPr pPr =  Context.getWmlObjectFactory().createPPr();
                this.getCurrentParagraph(true).setPPr(pPr);
               
                if (paragraphFormatting.equals(FormattingOption.IGNORE_CLASS)) {
                       
                        listHelper.addNumbering(this.getCurrentParagraph(true), blockBox.getElement(), cssMap);                        
                        addParagraphProperties(pPr, blockBox, cssMap );
                                               
                } else {
                        // CLASS_TO_STYLE_ONLY or CLASS_PLUS_OTHER
                        if (listHelper.peekListStack().getElement()!=null
                                        && listHelper.peekListStack().getElement().getAttribute("class")!=null) {
                                // NB Currently, you need to put this @class on the ol|ul at each level of nesting,
                                // if you want to use the list style.
                                // If you only put it on some levels, well, new list(s) will be created for the others,
                                // with imperfect results...
                               
                                String cssClass = listHelper.peekListStack().getElement().getAttribute("class").trim();
                                log.debug(cssClass);
                                if (cssClass.equals("")) {
                                        // What to do? same thing as if no @class specified
                                        if (paragraphFormatting.equals(FormattingOption.CLASS_PLUS_OTHER)) {
                                        listHelper.addNumbering(this.getCurrentParagraph(true), blockBox.getElement(), cssMap);
                                                addParagraphProperties(pPr, blockBox, cssMap );
                                        }
                                        // else its CLASS_TO_STYLE_ONLY,
                                        // but since we have no @class, do nothing
                                       
                                } else {
                                        // Usual case...
                                       
                                        // Our XHTML export gives a space separated list of class names,
                                        // reflecting the style hierarchy.  Here, we just want the first one.
                                        // TODO, replace this with a configurable stylenamehandler.
                                        int pos = cssClass.indexOf(" ");
                                        if (pos>-1) {
                                                cssClass = cssClass.substring(0,  pos);
                                        }
                                       
                                        // if the docx contains this stylename, set it
                                        Style s = this.stylesByID.get(cssClass);
                                        if (s==null) {
                                                log.debug("No docx style for @class='" + cssClass + "'");
                                               
                                                if (paragraphFormatting.equals(FormattingOption.CLASS_PLUS_OTHER)) {
                                                listHelper.addNumbering(this.getCurrentParagraph(true), blockBox.getElement(), cssMap);
                                                        addParagraphProperties(pPr, blockBox, cssMap );
                                                }
                                                // else, can't number
                                               
                                        } else if (s.getType()!=null && s.getType().equals("numbering")) {
                                                log.debug("Using list style from @class='" + cssClass + "'");
                                               
                                                /* it should contain something like:
                                                 *
                                                 *     <w:pPr>
                                                                              <w:numPr>
                                                                                <w:numId w:val="1"/>
                                                                              </w:numPr>
                                                                            </w:pPr>
                                                 *
                                                 * Use this...
                                                 *
                                                 * We don't actually set the numbering style on the
                                                 * paragraph, because numbering styles aren't used that way
                                                 * in Word.
                                                 */
                                                BigInteger numId = s.getPPr().getNumPr().getNumId().getVal();
                                                listHelper.setNumbering(pPr, numId);  
                                                        // TODO: @start is ignored in this case (it is handled in addNumbering)
                                               
                                                // Note that we just use the numbering it points to;
                                                // we don't follow it to its abstract num (which is in fact
                                                // where the w:styleLink matching our @class should be found).
                                               
                                                // TODO: if this list is being used a second time, we should
                                                // restart numbering??  Is it restarted in the HTML?

                                                // OK, we've applied @class
                                                if (paragraphFormatting.equals(FormattingOption.CLASS_PLUS_OTHER)) {
                                                        // now apply ad hoc formatting
                                                        addParagraphProperties(pPr, blockBox, cssMap );
                                                }                                              
                                               
                                        } else {
                                                log.debug("For docx style for @class='" + cssClass + "', but its not a numbering style ");
                                               
                                                if (paragraphFormatting.equals(FormattingOption.CLASS_PLUS_OTHER)) {
                                                       
                                                listHelper.addNumbering(this.getCurrentParagraph(true), blockBox.getElement(), cssMap);
                                               
                                                // SPECIAL CASE
                                                if (Docx4jProperties.getProperty("docx4j.model.datastorage.BindingTraverser.XHTML.Block.rStyle.Adopt", false)
                                                                        && s.getType()!=null && s.getType().equals("paragraph")) {      
                                                       
                                                        log.debug(".. using " + s.getStyleId() );
                                                       
                                                        PStyle pStyle = Context.getWmlObjectFactory().createPPrBasePStyle();
                                                        pStyle.setVal(s.getStyleId());
                                                        this.getCurrentParagraph(false).getPPr().setPStyle(pStyle);                                                            
                                                }
                                                        addParagraphProperties(pPr, blockBox, cssMap );
                                                }                                                      
                                               
                                        }
                                       
                                }
                        } else {
                                // No @class
                                if (paragraphFormatting.equals(FormattingOption.CLASS_PLUS_OTHER)) {
                                        addParagraphProperties(pPr, blockBox, cssMap );
                                }
                                // else its CLASS_TO_STYLE_ONLY,
                                // but since we have no @class, do nothing
                        }
                       
                }
 

Parsed in 0.032 seconds, using GeSHi 1.0.8.4

So to re-use an existing Word style on a list-item, you need:
1. to trigger isListItem, which using <li> does
2. to set paragraphFormatting to CLASS_TO_STYLE_ONLY or CLASS_PLUS_OTHER prior to starting the importer
3. set an appropriate li/@class in your XHTML <----------------------

[Michael]

Thank you for the extensive explanation. I've analyzed your suggestions, but I'm a beginner and some things are unclear to me.
In my application I read HTML data from another source and then create a docx. The data that comes in is in simple HTML form, e.g.

Code: Select all

<ul>
    <li>
        Text1
    </li>
    <li>
        <ul>
            <li>
                Text2
            </li>
        </ul>
    </li>
</ul>


Theoretically, I could manually set a separate @list-style-type for each nesting, but it would be cumbersome to manually analyze the nesting level and set the bullet style each time. So I chose your option number "1".

What I managed to do was set:

Code: Select all

xHTMLImporter.setParagraphFormatting(FormattingOption.CLASS_PLUS_OTHER);


Code: Select all

PPr xhtmlPPr = xhtmlPar.getPPr();

if (xhtmlPPr.getNumPr() != null) {
    PStyle pStyle = Context.getWmlObjectFactory().createPPrBasePStyle();
    pStyle.setVal("SpecifiedStyleID");
    xhtmlPPr.setPStyle(pStyle);
}


Unfortunately, setting a specific style ID from the document's styles.xml file does not change the bullets. I would like the subsequent levels of the list to have the same nested bullets as when I simply manually write such a list in this document in Word. I have tried many styles from the styles.xml file, but I don't know how to check which of these styles Word uses by default when I edit a document.

[jason]

Your approach of iterating through the paragraphs and setting the style after importing the XHTML should work.

To understand why not, we'd need to see your code and the resulting docx.

My approach "1. apply pre-existing Word styles " requires you to set an appropriate li/@class in your XHTML (or on the parent ol|ul).

If you aren't able to modify the incoming XHTML source to suit, you could always apply an XSLT to add in those class attributes. In your example, it'd be <ul class="SpecifiedStyleID">.

Before you try to write a suitable XSLT, I'd suggest you experiment with a static XHTML file to satisfy yourself that this approach works for you.

[Michael]

After setting in code the style in the importing XHTML, in the way I described above, in generated docx I get an example <p> tag:

Code: Select all

<w:p>
    <w:pPr>
        <w:pStyle w:val="SpecifiedStyleID"/>
        <w:numPr>
            <w:ilvl w:val="1"/>
            <w:numId w:val="31"/>
        </w:numPr>
        <w:spacing w:before="0" w:after="0"/>
        <w:jc w:val="left"/>
    </w:pPr>
    <w:r>
        <w:rPr>
            <w:rFonts w:ascii="Roboto Condensed Light" w:hAnsi="Roboto Condensed Light"/>
            <w:b w:val="false"/>
            <w:i w:val="false"/>
            <w:color w:val="000000"/>
            <w:sz w:val="24"/>
        </w:rPr>
        <w:t xml:space="preserve">Text</w:t>
    </w:r>
</w:p>


For the SpecifiedStyleID, I used all available styles from the styles.xml file that apply to lists, but entering these styles does not change the bullet styles in the generated document.
How can I find out what style Word is using when I manually edit a generated document in Word and create a new list in it?

[Michael]

How can I find out what style Word is using when I manually edit a generated document in Word and create a new list in it?

I created a Word document with a manually typed nested list and read the style in document.xml. When using this style when converting via XHTMLImpotrter according to the above method, the style of the nested bullets does not change.
Is there any additional information needed to resolve this issue? Unless I use XSLT, but that will be new to me.

[Michael]

In the above example, I think my generator adds the bullet style correctly.
I would like to ask whether the display of the appropriate bullet style depends only on the <w:pStyle w:val="SpecifiedStyleID"/> entry? If I use a different style name in the document generator code, even an incorrect one, which does not exist at all in the styles.xml file, the document always displays the same default style at all nesting levels (black dot).
Only when I write text by hand in MS Word, the bullet styles change accordingly.

[Michael]

Sorry for such a large number of posts, but I would like to add that maybe the reason is setting a specific font for each RPr before conversion?

Code: Select all

runRPr.getRFonts().setAscii("Roboto Condensed Light");
runRPr.getRFonts().setHAnsi("Roboto Condensed Light");


[Michael]

I am attaching a document with a simple example in which there are two nested lists: the first one generated by XHTMLImporter and the second one written by hand in MS Word using the same style.
The XHTML code for both lists seems to look the same, but in the list generated by the XHTML Importer, the nested bullet in Level 1 has the same "Disc" style as in Level 0. In turn, in the handwritten list, the bullets in Level 0 and Level 1 change correctly (Level 1 has the "Circle" style).

Another observation is that in the list generated by XHTMLImporter, I can manually create only as many nests (using the TAB key) as were generated. If I create a new list manually, I can use all nestings available in MS Word, including those defined in a given style.

Another observation is that when you manually copy a list generated by XHTMLImporter to another document, the bullet style restrictions are carried over to the other document.

[Michael]

Your approach of iterating through the paragraphs and setting the style after importing the XHTML should work.
To understand why not, we'd need to see your code and the resulting docx.

I would like to ask if the attached document and part of the code is not enough to determine why the bullets are not working correctly according to the style?

I also tried replacing each <ul> tag with <ul style="list-style-type: circle"> before the conversion to see if DISC would change to CIRCLE at all, but nothing changed.
As a test, I managed to change the style of numbered points from numbers to bullets using the code:

Code: Select all

Numbering numbering = getNumberingDefinitionsPart().getJaxbElement();
for (Numbering.AbstractNum abstractNumNode : numbering.getAbstractNum() ) {
    for (Lvl lvl : abstractNumNode.getLvl()) {
        NumFmt numFmt = new NumFmt();
        numFmt.setVal(NumberFormat.BULLET);
        lvl.setNumFmt(numFmt);
    }
}


Maybe I could also force the bullets for individual levels to change in a similar way (but I don't know how)?

[jason]

You simply need a paragraph style definition which contains numPr settings.

NOT a style of @type="numbering": http://webapp.docx4java.org/OnlineDemo/ ... tyles.html

For example:

Syntax: [ Download ] [ Hide ]

Syntax: [ Download ] [ Show ]

Using xml Syntax Highlighting

  <w:style w:type="paragraph" w:styleId="ListParagraph">
    <w:name w:val="List Paragraph"/>
    <w:basedOn w:val="Normal"/>
    <w:uiPriority w:val="34"/>
    <w:qFormat/>
    <w:pPr>
      <w:ind w:left="720"/>
      <w:contextualSpacing/>
    </w:pPr>
  </w:style>
 
  <w:style w:type="paragraph" w:customStyle="1" w:styleId="ListTop">
    <w:name w:val="List top level"/>
    <w:basedOn w:val="ListParagraph"/>
    <w:qFormat/>
    <w:pPr>
      <w:numPr>
        <w:ilvl w:val="0"/>
        <w:numId w:val="1"/>
      </w:numPr>
    </w:pPr>
  </w:style>
  <w:style w:type="paragraph" w:customStyle="1" w:styleId="ListChildLevel">
    <w:name w:val="List child level"/>
    <w:basedOn w:val="ListParagraph"/>
    <w:qFormat/>
    <w:pPr>
      <w:numPr>
        <w:ilvl w:val="1"/>
        <w:numId w:val="1"/>
      </w:numPr>
    </w:pPr>
  </w:style>

 

Parsed in 0.006 seconds, using GeSHi 1.0.8.4

I created the above in Word but manually added w:ilvl/@w:val="0" for clarity. By default Word based my list styles on ListParagraph, but it doesn't have to be that way. You could base them on Normal, or any other paragraph style, or nothing at all.

In your sample docx, the numbering is directly applied in document.xml; it doesn't come from styles.xml

[Michael]

Thanks for your answer, but I don't understand one thing:

In your sample docx, the numbering is directly applied in document.xml; it doesn't come from styles.xml

I did not find a definition of the used "Akapitzlist" style in document.xml, but it is defined in styles.xml:

Syntax: [ Download ] [ Hide ]

Syntax: [ Download ] [ Show ]

Using xml Syntax Highlighting

<w:style w:type="paragraph" w:styleId="Akapitzlist">
    <w:name w:val="List Paragraph"/>
    <w:basedOn w:val="Normalny"/>
    <w:autoRedefine/>
    <w:uiPriority w:val="34"/>
    <w:qFormat/>
    <w:rsid w:val="00463463"/>
    <w:pPr>
        <w:ind w:left="720"/>
        <w:contextualSpacing/>
    </w:pPr>
</w:style>
 

Parsed in 0.002 seconds, using GeSHi 1.0.8.4

So how does it work?
Why also do the Level 1 bullets below have different styles (the first from XHTMLImporter and the second handwritten in Word)?

Syntax: [ Download ] [ Hide ]

Syntax: [ Download ] [ Show ]

Using xml Syntax Highlighting

<w:p w14:paraId="2BAC69A2" w14:textId="77777777" w:rsidR="00AC04D3" w:rsidRDefault="00AC04D3" w:rsidP="00AC04D3">
    <w:pPr>
        < w:pStyle w:val="Akapitzlist"/>
        <w:numPr>
            <w:ilvl w:val="1"/>
            <w:numId w:val="3"/>
       </w:numPr>
    </w:pPr>
    <w:r>
        <w:t>Level 1 XHTML Importer</w:t>
    </w:r>
</w:p>
 

Parsed in 0.002 seconds, using GeSHi 1.0.8.4

Syntax: [ Download ] [ Hide ]

Syntax: [ Download ] [ Show ]

Using xml Syntax Highlighting

<w:p w14:paraId="49788138" w14:textId="77777777" w:rsidR="00AC04D3" w:rsidRDefault="00AC04D3" w:rsidP="00AC04D3">
    <w:pPr>
        <w:pStyle w:val="Akapitzlist"/>
        <w:numPr>
            <w:ilvl w:val="1"/>
            <w:numId w:val="4"/>
        </w:numPr>
    </w:pPr>
    <w:r>
        <w:t>Level 1 MS Word</w:t>
    </w:r>
</w:p>
 

Parsed in 0.002 seconds, using GeSHi 1.0.8.4

[jason]

Hi Michael

I thought I'd create a sample illustrating how to do this.

In doing so, I discovered a couple of things:
1. the @class approach (ie to adopt an existing Word style) only worked with styles of type numbering. Now fixed to work with paragraph styles: https://github.com/plutext/docx4j-Impor ... d7827bc953
2. the @list-style-type approach doesn't work because openhtmltopdf isn't passing on the value of that attribute correctly

So for now only the @class approach is feasible, and https://github.com/plutext/docx4j-Impor ... tyles.java is an example of it. It should work if your styles are of type numbering. For it to work with the usual paragraph styles, you'll need the above commit (or wait for a new release).

[Michael]

Thanks a lot for update with working with paragraph styles. For me it will be easier to wait for a new release. I have to work with Java 8 so I will wait for the next distribution for Java8.

[jason]

It is in docx-java-f6/docx4j-importxhtml-8-3-10-released-t3118.html

[Michael]

I would like to check the changes made with nested bullets. For this purpose I downloaded the latest docx4j-community-8.3.11 and docx4j-ImportXHTML-8.3.11 and installed them with all dependencies. Previously I used docx4j-community-8.3.4 and docx4j-ImportXHTML-8.3.2.
After updating the plugins I get the error:

Code: Select all

java.lang.NoSuchMethodError: org.apache.commons.compress.archivers.zip.ZipArchiveInputStream.getNextEntry()Lorg/apache/commons/compress/archivers/zip/ZipArchivEntry;

What could be causing the problem?

[jason]

Does that error come from attempting to open a docx file?

docx4j 8.3.11 uses commons-compress 1.26.1

Code: Select all

[INFO] --- dependency:3.7.0:tree (default-cli) @ docx4j-core ---
[INFO] org.docx4j:docx4j-core:jar:8.3.11
[INFO] +- org.apache.commons:commons-compress:jar:1.26.1:compile

Please check you have that on your classpath, and no other conflicting version.

If the problem persists, please supply a small maven project which reproduces the isse.

[Michael]

I have found a potential cause of the problem.
The commons-compress error appears when loading a docx file. If I create a simple application with only loading a docx file, everything works correctly.
On the other hand, the application in which I ultimately use docx4j is a plugin in MagicDraw, in which I indicate in the plugin.xml file which dependencies my plugin should use, including the appropriate version of docx4j and other required dependencies. However, it turns out that despite indicating the location of the commons-compress-1.26.1.jar plugin in the plugin.xml in MagicDraw, docx4j in the MagicDraw plugin does not use this library, but uses commons-compress-1.3.jar from the MagicDraw lib folder (from ProgramFiles).
I was unable to force the MagicDraw plugin to use the specified commons-compress-1.26.1.jar. Or maybe the reason is that docx4j-core-8.3.11 is not able to use commons-compress-1.26.1.jar which is in the same directory in the MagicDraw plugin folder, but chooses commons-compress-1.3.jar from MagicDraw libraries?

[jason]

https://docs.nomagic.com/display/MD190/ ... ss+loading suggests you might try using your own classloader and setting property class-lookup="LocalFirst"?

Maybe easier - if you haven't tried it already - you could try replacing commons-compress-1.26.1.jar with commons-compress-1.3.jar in your MagicDraw plugin

[Michael]

Thanks for the tips. I did the first try with "own classloader" and "class-lookup=LocalFirst" before and I had another MagicDraw error then, but this is already a problem for me and maybe I will go in this direction to improve it.
I used commons-compress-1.3.jar instead of commons-compress-1.26.1.jar in this test project and I checked that with docx4j-core-8.3.11 and commons-compress-1.3.jar there is the same "NoSuchMethodError" error.
Since MagicDraw only has access to commons-compress-1.26.1.jar in the plugin and to its own commons-compress-1.3.jar, I suppose that MagicDraw itself somehow imposes the use of its commons-compress-1.3.jar. I have verified that the earlier version docx4j-core-8.3.4 works correctly with commons-compress-1.3.jar and therefore I had no problem with it.

The problem is probably on the MagicDraw side and I will continue to solve it.

[Michael]

I have implemented the correct use of "own classloader" and "class-lookup=LocalFirst" in MagicDraw by forcing the use of the local dependency commons-compress-1.26.1.jar and everything works fine.

I have also tested that using the own bullet list style "<ul class='styleId'> works fine. Thanks for your help.

[jason]

Good stuff Michael.

[jason]

Just an aside to note that commons-compress-1.26.1.jar is a fairly recent release (May 2024), but 1.3 used by MagicDraw is ancient (dating from 2011)!

[Michael]

but 1.3 used by MagicDraw is ancient (dating from 2011)!

Yes, that's right, but I'm using the old version of MagicDraw "19.0".