Examples with XWPFStylesShim org.apache.tika.parser.microsoft.ooxml.xwpf.XWPFStylesShim used on opensource projects

Example 1 with XWPFStylesShim

use of org.apache.tika.parser.microsoft.ooxml.xwpf.XWPFStylesShim in project tika by apache.

the class SXWPFWordExtractorDecorator method handleDocumentPart.

private void handleDocumentPart(PackagePart documentPart, XHTMLContentHandler xhtml) throws IOException, SAXException {
    //load the numbering/list manager and styles from the main document part
    XWPFNumbering numbering = loadNumbering(documentPart);
    XWPFListManager listManager = new XWPFListManager(numbering);
    XWPFStylesShim styles = null;
    try {
        styles = loadStyles(documentPart);
    } catch (Exception e) {
        metadata.add(TikaCoreProperties.TIKA_META_EXCEPTION_WARNING, ExceptionUtils.getStackTrace(e));
    }
    //headers
    try {
        PackageRelationshipCollection headersPRC = documentPart.getRelationshipsByType(XWPFRelation.HEADER.getRelation());
        if (headersPRC != null) {
            for (int i = 0; i < headersPRC.size(); i++) {
                PackagePart header = documentPart.getRelatedPart(headersPRC.getRelationship(i));
                handlePart(header, styles, listManager, xhtml);
            }
        }
    } catch (InvalidFormatException | ZipException e) {
        metadata.add(TikaCoreProperties.TIKA_META_EXCEPTION_WARNING, ExceptionUtils.getStackTrace(e));
    }
    //main document
    try {
        handlePart(documentPart, styles, listManager, xhtml);
    } catch (ZipException e) {
        metadata.add(TikaCoreProperties.TIKA_META_EXCEPTION_WARNING, ExceptionUtils.getStackTrace(e));
    }
    //for now, just dump other components at end
    for (XWPFRelation rel : new XWPFRelation[] { XWPFRelation.FOOTNOTE, XWPFRelation.COMMENT, XWPFRelation.FOOTER, XWPFRelation.ENDNOTE }) {
        try {
            PackageRelationshipCollection prc = documentPart.getRelationshipsByType(rel.getRelation());
            if (prc != null) {
                for (int i = 0; i < prc.size(); i++) {
                    PackagePart packagePart = documentPart.getRelatedPart(prc.getRelationship(i));
                    handlePart(packagePart, styles, listManager, xhtml);
                }
            }
        } catch (InvalidFormatException | ZipException e) {
            metadata.add(TikaCoreProperties.TIKA_META_EXCEPTION_WARNING, ExceptionUtils.getStackTrace(e));
        }
    }
}

Example 2 with XWPFStylesShim

use of org.apache.tika.parser.microsoft.ooxml.xwpf.XWPFStylesShim in project tika by apache.

the class SXWPFWordExtractorDecorator method loadStyles.

private XWPFStylesShim loadStyles(PackagePart packagePart) throws InvalidFormatException, TikaException, IOException, SAXException {
    PackageRelationshipCollection stylesParts = packagePart.getRelationshipsByType(XWPFRelation.STYLES.getRelation());
    if (stylesParts.size() > 0) {
        PackageRelationship stylesRelationShip = stylesParts.getRelationship(0);
        if (stylesRelationShip == null) {
            return null;
        }
        PackagePart stylesPart = packagePart.getRelatedPart(stylesRelationShip);
        if (stylesPart == null) {
            return null;
        }
        return new XWPFStylesShim(stylesPart, context);
    }
    return null;
}