Deleting code points encoded as 2 characters in UTF-16

pom.xml

@@ -4,7 +4,7 @@
 
 	<groupId>ca.crim.nlp</groupId>
 	<artifactId>crim-heideltime</artifactId>
-	<version>3.0.7-SNAPSHOT</version>
+	<version>3.0.8-SNAPSHOT</version>
 
 	<name>HeidelTime</name>
 	<description> This version of HeidelTime extends the well-known multilingual cross-domain temporal tagger (com.github.hiedltime) that extracts temporal expressions from documents and normalizes them according to the TIMEX3 annotation standard.</description>

src/main/java/de/unihd/dbs/heideltime/standalone/HeidelTimeStandalone.java

@@ -844,7 +844,25 @@ public static void main(String[] args) {
 			}
 
 			// double-newstring should not be necessary, but without this, it's not running on Windows (?)
-			String input = new String(new String(inArr, encodingType).getBytes("UTF-8"), "UTF-8");
+			String input_raw = new String(new String(inArr, encodingType).getBytes("UTF-8"), "UTF-8");
+			String input ;
+
+			// Elimination of code points encoded as two 16-bit code units (they were counted as 2 characters)
+			if (outputType.equals(OutputType.JSON)) {
+				StringBuilder sb = new StringBuilder();
+				for (int i = 0; i< input_raw.length(); ++i) {
+					if  (Character.isHighSurrogate(input_raw.charAt(i))) {
+						sb.append('?');
+					}
+					else if (Character.isLowSurrogate(input_raw.charAt(i)))
+						continue;
+					else
+					sb.append(input_raw.charAt(i));
+				}
+				input = sb.toString();
+			}
+			else 
+				input = input_raw;
 
 			HeidelTimeStandalone standalone = new HeidelTimeStandalone(language, type, outputType, null, posTagger, doIntervalTagging);
 			String out = standalone.process(input, dct, pos_file, sentence_file);

src/main/java/de/unihd/dbs/heideltime/standalone/components/impl/JSONResultFormatter.java

@@ -31,7 +31,7 @@ public String format(JCas jcas) throws Exception {
 		try {
 			l.log(Level.FINEST, "Preparing to serialize the results in JSON");
 			JsonCasSerializer serializer = new JsonCasSerializer();
-
+			serializer.setPrettyPrint(true);
 			serializer.serialize(jcas.getCas(), output);
 			outText = output.toString();
 			l.log(Level.FINEST, "JSON-serialization finished.");