groupdict() returns only first character of the match if there is a single named capture group

[simonw]

This line:

[m.groupdict() for m in rure.compile('(?P<word>\w+)').finditer("hello there")]

Returns this:

[{'word': 'h'}, {'word': 't'}]

I would expect it to return this:

[{'word': 'hello'}, {'word': 'there'}]

Adding a second named group fixes this issue for some reason:

[m.groupdict() for m in rure.compile('(?P<name>\w+)(?P<nomatch>)').finditer("hello there")]

Returns:

[{'name': 'hello', 'nomatch': ''}, {'name': 'there', 'nomatch': ''}]

[simonw]

I think the bug may be in this code:

rure-python/rure/regex.py

Lines 235 to 251 in 024bf4b

	def groupdict(self, default=None, decode=True):
	group_indices = sorted(self.re.groupindex.values())
	group_data = self.group(*[i for i in group_indices], default=default)
	gdict = {}
	for pos, g_name_index in enumerate(
	sorted(self.re.groupindex.items(),
	# Sort by value
	key=lambda x: x[1])):
	gname = g_name_index[0]
	gval = group_data[pos]
	if decode:
	if isinstance(gname, bytes):
	gname = gname.decode('utf8')
	if isinstance(gval, bytes):
	gval = gval.decode('utf8')
	gdict[gname] = gval
	return gdict

[simonw]

This happens without .finditer() too:

>>> rure.compile('(?P<name>\w+)').match("hello").groupdict()
{'name': 'h'}
>>> rure.compile('(?P<name>\w+)(?P<nomatch>)').match("hello").groupdict()
{'name': 'hello', 'nomatch': ''}