JsonObject writeTextAnnotation(TextAnnotation ta, boolean doWriteTokenOffsets) {
// get rid of the views that are empty
Set<String> viewNames = new HashSet<>(ta.getAvailableViews());
for (String vu : viewNames) {
if (ta.getView(vu) == null) {
logger.warn("View " + vu + " is null");
ta.removeView(vu);
}
}
JsonObject json = new JsonObject();
writeString("corpusId", ta.getCorpusId(), json);
writeString("id", ta.getId(), json);
writeString("text", ta.getText(), json);
writeStringArray("tokens", ta.getTokens(), json);
if (doWriteTokenOffsets) writeTokenOffsets(TOKENOFFSETS, ta.getView(ViewNames.TOKENS), json);
writeSentences(ta, json);
JsonArray views = new JsonArray();
for (String viewName : Sorters.sortSet(ta.getAvailableViews())) {
if (viewName.equals(ViewNames.SENTENCE)) continue;
JsonObject view = new JsonObject();
writeString("viewName", viewName, view);
views.add(view);
JsonArray viewData = new JsonArray();
List<View> topKViews = ta.getTopKViews(viewName);
for (View topKView : topKViews) {
JsonObject kView = new JsonObject();
writeView(topKView, kView);
viewData.add(kView);
}
view.add("viewData", viewData);
}
json.add("views", views);
writeAttributes(ta, json);
return json;
}
protected String getNERString() {
List<Constituent> constituents = new ArrayList<>(view.getConstituents());
Collections.sort(constituents, TextAnnotationUtilities.constituentStartComparator);
StringBuilder sb = new StringBuilder();
String text = textAnnotation.getText();
int where = 0;
for (Constituent c : constituents) {
int start = c.getStartCharOffset();
String startstring = text.substring(where, start);
sb.append(startstring)
.append("[")
.append(c.getLabel())
.append(" ")
.append(c.getTokenizedSurfaceForm())
.append(" ] ");
where = c.getEndCharOffset();
}
return sb.toString();
}
