Java Code Examples for net.htmlparser.jericho.Source#fullSequentialParse()

public void setHTMLText(String htmlText) {
    try {
        this.text = htmlText;
        if (formatt) {
            Source segment = new Source(text);
            segment.fullSequentialParse();
            SourceFormatter formatter = new SourceFormatter(
                    segment);
            htmlText = formatter.toString();
        }

        editorPane.read(new StringReader(htmlText), null);
    } catch (IOException e) {
        e.printStackTrace();
    }
}
 

protected void locateView ()
{
    if ( !locate )
    {
        return;
    }

    final WebSyntaxArea syntaxArea = editors.get ( editorTabs.getSelectedIndex () );
    final String xml = syntaxArea.getText ();

    final Source xmlSource = new Source ( xml );
    xmlSource.setLogger ( null );
    xmlSource.fullSequentialParse ();

    final StartTag tag = xmlSource.getPreviousStartTag ( syntaxArea.getCaretPosition (), "style" );
    if ( tag != null )
    {
        // todo Won't work with new scheme, have to go all the way up and gather all style identifiers
        final String type = tag.getAttributeValue ( ComponentStyleConverter.COMPONENT_TYPE_ATTRIBUTE );
        final String id = tag.getAttributeValue ( ComponentStyleConverter.STYLE_ID_ATTRIBUTE );
        locateView ( previewPanel, type, id );
    }
}
 

protected void loadFirstResource ( final List<ClassResource> resources, final List<String> xmlContent, final List<String> xmlNames,
                                   final List<ClassResource> xmlFiles ) throws IOException
{
    final ClassResource rf = resources.get ( 0 );
    final Source xmlSource = new Source ( ReflectUtils.getClassSafely ( rf.getClassName () ).getResource ( rf.getPath () ) );
    xmlSource.setLogger ( null );
    xmlSource.fullSequentialParse ();

    final Element baseClassTag = xmlSource.getFirstElement ( SkinInfoConverter.CLASS_NODE );
    final String baseClass = baseClassTag != null ? baseClassTag.getContent ().toString () : null;

    for ( final Element includeTag : xmlSource.getAllElements ( SkinInfoConverter.INCLUDE_NODE ) )
    {
        final String includeClass = includeTag.getAttributeValue ( SkinInfoConverter.NEAR_CLASS_ATTRIBUTE );
        final String finalClass = includeClass != null ? includeClass : baseClass;
        final String src = includeTag.getContent ().toString ();
        resources.add ( new ClassResource ( finalClass, src ) );
    }

    xmlContent.add ( xmlSource.toString () );
    xmlNames.add ( new File ( rf.getPath () ).getName () );
    xmlFiles.add ( rf );

    resources.remove ( 0 );
}
 

/**
 * Returns plain text extracted from the specified HTML.
 *
 * @param html          HTML
 * @param lineSeparator line separator
 * @return plain text
 */
@NotNull
public static String getPlainText ( @NotNull final String html, @NotNull final String lineSeparator )
{
    final String plain;
    final Source source = new Source ( html );
    final Tag[] tags = source.fullSequentialParse ();
    if ( tags.length > 0 )
    {
        final Renderer renderer = source.getRenderer ();
        renderer.setIncludeHyperlinkURLs ( false );
        renderer.setIncludeAlternateText ( false );
        renderer.setDecorateFontStyles ( false );
        renderer.setMaxLineLength ( Integer.MAX_VALUE );
        renderer.setHRLineLength ( 40 );
        renderer.setBlockIndentSize ( 4 );
        renderer.setConvertNonBreakingSpaces ( false );
        renderer.setNewLine ( lineSeparator );
        plain = renderer.toString ();
    }
    else
    {
        plain = html;
    }
    return plain;
}
 

public String convert() {
    Source source = new Source(code);
    source.fullSequentialParse();

    int from = 0;

    StringBuffer result = new StringBuffer();

    for (StartTag tag : source.getAllStartTags()) {
        if (tag.getName().startsWith("%")) {

            addLines(result, source, from, tag.getBegin());

            from = tag.getEnd();

            String script = tag.toString();
            if (script.length() > 4) {
                if (script.charAt(2) == '=') {
                    result.append("doc.print(");
                    result.append(script.substring(3, script.length() - 2));
                    result.append(");");
                } else {
                    result.append(script.substring(2, script.length() - 2));
                }
            }
        }
    }

    addLines(result, source, from, source.getEnd());

    return result.toString();
}
 

public Source getReport(String name, Query query) {
    Element element = engine.getElement(name, ReportPlugin
            .getReportsQualifier(engine).getId());
    if (element == null)
        throw new DataException("Error.reportNotFound", "Report " + name
                + " not found", name);
    HashMap<String, Object> map = new HashMap<String, Object>();
    if (query != null)
        map.put("query", query);
    String htmlReport = reportQuery.getHTMLReport(element, map);
    Source source = new Source(htmlReport);
    source.fullSequentialParse();
    return source;
}
 

public void printHTMLPage(HTMLPage page) throws IOException {
    byte[] data = page.getData();
    if (data == null)
        return;
    Source source = new Source(new ByteArrayInputStream(data));
    source.fullSequentialParse();
    printHTMLPage(source);
}
 

public void realWrite(boolean printHeadBodyTags) throws IOException {
    flush();
    ByteArrayOutputStream stream = (ByteArrayOutputStream) this.out;
    this.out = outputStream;
    if (!printHeadBodyTags) {
        stream.writeTo(outputStream);
        return;
    }
    println("<html>");
    println("<head>");
    println("<style>");
    for (Style style : styles)
        println(style);
    println("</style>");
    println("</head>");

    String htmlText = new String(stream.toByteArray(), "UTF-8");
    Source source = new Source(htmlText);
    source.fullSequentialParse();

    List<StartTag> startTags = source.getAllStartTags("body");
    if (startTags.size() == 0) {
        println("<body>");
        println(htmlText);
        println("</body>");
    } else {
        println(new StringBuffer(startTags.get(0).getElement()));
    }

    println("</html>");
}
 

@Override
public void actionPerformed(ActionEvent e) {
    String text = getText();
    Source segment = new Source(text);
    segment.fullSequentialParse();
    SourceFormatter formatter = new SourceFormatter(segment);
    text = formatter.toString();
    editorPane.selectAll();
    editorPane.replaceSelection(text);
}
 

public void realWriteWithHTMLUpdate() throws IOException {
    flush();
    ByteArrayOutputStream out = (ByteArrayOutputStream) this.out;
    Source source = new Source(new String(out.toByteArray(), "UTF-8"));
    source.fullSequentialParse();
    List<StartTag> list = source.getAllStartTags("html");
    if (list.size() == 0) {
        realWrite();
        return;
    }

    this.out = outputStream;

    OutputStreamWriter writer = new OutputStreamWriter(this.out, "UTF-8");
    OutputDocument document = new OutputDocument(source);

    StringBuffer style = new StringBuffer();
    if (this.styles.size() > 0) {
        for (Style style2 : this.styles)
            style.append(style2.toString());
    }

    List<StartTag> h = source.getAllStartTags("style");
    if (h.size() > 0) {
        document.insert(h.get(0).getElement().getEndTag().getBegin(), style);
    } else {

        style.insert(0, "\n<style>\n");
        style.append("</style>\n");

        h = source.getAllStartTags("head");
        if (h.size() > 0) {
            document.insert(h.get(0).getElement().getEndTag().getBegin(),
                    style);
        } else {
            style.insert(0, "\n<head>\n");
            style.append("</head>\n");
            document.insert(h.get(0).getElement().getEndTag().getBegin(),
                    style);
        }
    }
    document.writeTo(writer);
    writer.flush();

}