public DocumentAffiliation(String id, String index, String rawText) { this.id = id; this.index = ContentCleaner.clean(index); if (rawText.matches(".+ \\([^\\(\\)]*\\)$")) { rawText = StringUtils.reverse(rawText).replaceFirst("\\)", "").replaceFirst("\\( ", " ,"); rawText = StringUtils.reverse(rawText); } this.rawText = MetadataTools.cleanAndNormalize(rawText); this.tokens = new ArrayList<Token<AffiliationLabel>>(); }
@Override public void clean() { index = ContentCleaner.clean(index); rawText = ContentCleaner.cleanAllAndBreaks(rawText); }