Anchoring normalization results in empty character range (from==to)

[danielhers]

When normalizing graph 20010012 for EDS, node 5 is originally anchored to character 19 (from=19, to=20), but after normalization, the anchor sequence is empty (from=to=20). This seems to be a bug in the anchoring normalization code:

mtool/graph.py

Lines 85 to 112 in eda24cf

	def union(anchors):
	characters = set();
	for anchor in anchors:
	if "from" in anchor and "to" in anchor:
	for i in range(anchor["from"], anchor["to"]):
	characters.add(i);
	result = [];
	last = start = None;
	for i in sorted(characters):
	if start is None: start = i;
	if last is None:
	last = i;
	continue;
	elif i == last + 1 \
	or all(c in score.core.SPACE for c in input[last:i]):
	last = i;
	continue;
	else:
	result.append({"from": start, "to": last + 1});
	last = start = i;
	if len(characters) > 0:
	result.append({"from": start, "to": i + 1});
	if anchors != result:
	old = [anchor for anchor in anchors if anchor not in result];
	new = [anchor for anchor in result if anchor not in anchors];
	print("{} ==> {} [{}]".format(old, new, input),
	file = sys.stderr);
	return result;

The print in line 110 shows when running it, printing:
[{'from': 20, 'to': 20}] ==> [] [Then, in the guests' honor, the speedway hauled out four drivers, crews and even the official Indianapolis 500 announcer for a 10-lap exhibition race.]

I think it might be because the apostrophe character, which comprises the whole anchor, is considered a space character.