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

Example 1 with TrpTranscriptStatistics

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

the class PageXmlUtils method extractStats.

public static TrpTranscriptStatistics extractStats(PcGtsType page) {
    TrpTranscriptStatistics s = new TrpTranscriptStatistics();
    int nrOfRegions, nrOfTranscribedRegions, nrOfWordsInRegions, nrOfLines, nrOfTranscribedLines, nrOfWordsInLines, nrOfWords, nrOfTranscribedWords;
    nrOfRegions = nrOfTranscribedRegions = nrOfWordsInRegions = nrOfLines = nrOfTranscribedLines = nrOfWordsInLines = nrOfWords = nrOfTranscribedWords = 0;
    List<TextRegionType> regs = PageXmlUtils.getTextRegions(page);
    nrOfRegions = regs.size();
    for (TextRegionType r : regs) {
        if (r.getTextEquiv() != null && r.getTextEquiv().getUnicode() != null && !r.getTextEquiv().getUnicode().trim().isEmpty()) {
            nrOfTranscribedRegions += 1;
            // TODO use tokenizer here
            nrOfWordsInRegions += r.getTextEquiv().getUnicode().split(" ").length;
        }
        List<TextLineType> lines = r.getTextLine();
        nrOfLines += lines.size();
        for (TextLineType l : lines) {
            if (l.getTextEquiv() != null && l.getTextEquiv().getUnicode() != null && !l.getTextEquiv().getUnicode().trim().isEmpty()) {
                nrOfTranscribedLines += 1;
                // TODO use tokenizer here
                nrOfWordsInLines += l.getTextEquiv().getUnicode().split(" ").length;
            }
            List<WordType> words = l.getWord();
            nrOfWords += words.size();
            for (WordType w : words) {
                if (w.getTextEquiv() != null && w.getTextEquiv().getUnicode() != null && !w.getTextEquiv().getUnicode().trim().isEmpty()) {
                    nrOfTranscribedWords += 1;
                }
            }
        }
    }
    s.setNrOfLines(nrOfLines);
    s.setNrOfRegions(nrOfRegions);
    s.setNrOfTranscribedLines(nrOfTranscribedLines);
    s.setNrOfTranscribedWords(nrOfTranscribedWords);
    s.setNrOfTranscribedRegions(nrOfTranscribedRegions);
    s.setNrOfWords(nrOfWords);
    s.setNrOfWordsInLines(nrOfWordsInLines);
    s.setNrOfWordsInRegions(nrOfWordsInRegions);
    return s;
}

Example 2 with TrpTranscriptStatistics

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

the class DocStatisticsBuilder method compute.

public TrpTranscriptStatistics compute(TrpDoc doc) throws JAXBException {
    if (doc == null) {
        throw new IllegalArgumentException("TrpDoc is null!");
    }
    TrpTranscriptStatistics stats = new TrpTranscriptStatistics();
    List<TrpPage> pages = doc.getPages();
    for (TrpPage p : pages) {
        final String msg = "Computing stats: page " + p.getPageNr() + "/" + pages.size();
        logger.debug(msg);
        notifyObservers(msg);
        setChanged();
        URL xmlUrl = p.getCurrentTranscript().getUrl();
        PcGtsType pc = PageXmlUtils.unmarshal(xmlUrl);
        TrpTranscriptStatistics pageStats = PageXmlUtils.extractStats(pc);
        stats.add(pageStats);
    }
    return stats;
}