use of org.apache.tika.parser.microsoft.ooxml.XWPFListManager in project tika by apache.
the class XWPFEventBasedWordExtractor method handleDocumentPart.
private void handleDocumentPart(PackagePart documentPart, StringBuilder sb) throws IOException, SAXException {
//load the numbering/list manager and styles from the main document part
XWPFNumbering numbering = loadNumbering(documentPart);
XWPFListManager xwpfListManager = new XWPFListManager(numbering);
//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, xwpfListManager, sb);
}
}
} catch (InvalidFormatException e) {
LOG.warn("Invalid format", e);
}
//main document
handlePart(documentPart, xwpfListManager, sb);
//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, xwpfListManager, sb);
}
}
} catch (InvalidFormatException e) {
LOG.warn("Invalid format", e);
}
}
}
Aggregations