Examples with TrpTextLineType eu.transkribus.core.model.beans.pagecontent_trp.TrpTextLineType used on opensource projects

Example 21 with TrpTextLineType

use of eu.transkribus.core.model.beans.pagecontent_trp.TrpTextLineType in project TranskribusCore by Transkribus.

the class ExportCache method storeCustomTagMapForDoc.

/**
 * @param doc
 * @param wordBased
 * @param pageIndices
 * @param blackening
 * @return all (custom) tags of the given document
 * @throws JAXBException
 * @throws IOException
 * @throws InterruptedException
 */
public void storeCustomTagMapForDoc(TrpDoc doc, boolean wordBased, Set<Integer> pageIndices, IProgressMonitor monitor, boolean blackening) throws JAXBException, IOException, InterruptedException {
    doBlackening = blackening;
    tags.clear();
    List<TrpPage> pages = doc.getPages();
    int totalPages = pages.size();
    int c = 0;
    for (int i = 0; i < totalPages; ++i) {
        if (pageIndices != null && !pageIndices.contains(i))
            continue;
        if (monitor != null && monitor.isCanceled()) {
            throw new InterruptedException("User canceled the export");
        }
        // pageTranscripts get fetched before the custom tag map is stored - so normally pageTranscripts.get(i) != null
        JAXBPageTranscript tr;
        if (pageTranscripts == null || pageTranscripts.get(i) == null) {
            TrpPage page = pages.get(i);
            TrpTranscriptMetadata md = page.getCurrentTranscript();
            tr = new JAXBPageTranscript(md);
        } else {
            tr = pageTranscripts.get(i);
            tr.getPageData();
        }
        tr.build();
        TrpPageType trpPage = tr.getPage();
        logger.debug("get tags for page " + (i + 1) + "/" + doc.getNPages());
        List<TrpTextRegionType> textRegions = trpPage.getTextRegions(true);
        for (int j = 0; j < textRegions.size(); ++j) {
            TrpTextRegionType r = textRegions.get(j);
            List<TextLineType> lines = r.getTextLine();
            for (int k = 0; k < lines.size(); ++k) {
                TrpTextLineType trpL = (TrpTextLineType) lines.get(k);
                List<WordType> words = trpL.getWord();
                getTagsForShapeElement(trpL);
                if (wordBased) {
                    for (int l = 0; l < words.size(); ++l) {
                        TrpWordType w = (TrpWordType) words.get(l);
                        getTagsForShapeElement(w);
                    }
                }
            // else{
            // getTagsForShapeElement(trpL);
            // }
            }
        }
        if (monitor != null) {
            monitor.setTaskName("Loaded tags for page " + (i + 1));
            monitor.worked(++c);
        }
    }
}

Example 22 with TrpTextLineType

use of eu.transkribus.core.model.beans.pagecontent_trp.TrpTextLineType in project TranskribusCore by Transkribus.

the class CustomTagUtil method getIndexedCustomTagsForLines.

public static <T extends CustomTag> List<T> getIndexedCustomTagsForLines(TrpTextRegionType region, String tagName) {
    List<T> tags = new ArrayList<>();
    CustomTag ct = null;
    for (TextLineType l : region.getTextLine()) {
        TrpTextLineType tl = (TrpTextLineType) l;
        List<T> lineTags = tl.getCustomTagList().getIndexedTags(tagName);
        for (T t : lineTags) {
            if (ct != null && isContinuation(tagName, t)) {
                ct.continuations.add(t);
            } else {
                tags.add(t);
                ct = t.isContinued() ? t : null;
            }
        }
    }
    return tags;
}