@Test public void testSeparators() { Matcher m; m = PatternConst.PUNCT_SEPARATORS.matcher("-*=~,`'"); assertFalse(m.find()); String[] s = {"--", "***", "==", "~~~", ",,", "```", "''"}; m = PatternConst.PUNCT_SEPARATORS.matcher(Joiner.join(s, "")); int i, size = s.length; for (i = 0; i < size; i++) { m.find(); assertEquals(s[i], m.group()); } }
@Override /* This is wrong! Relaxed String Match = Dropping relative clauses, PP, and participial modifiers */ protected String getWordSequence(AbstractMention mention) { if (mention.getSubTreeNodes() != null) { List<DEPNode> l_subNodes = new ArrayList<>(mention.getSubTreeNodes()); DEPNode node = mention.getNode(); removePunctuations(l_subNodes, node); removeRelativeClause(l_subNodes, node); removePrepositionalMod(l_subNodes, node); // removeParticipialMod(l_subNodes, node); return Joiner.join( l_subNodes.stream().map(n -> n.getWordForm()).collect(Collectors.toList()), " "); } return null; }