pyclts: Enable re-use of characters

[Anaphory]

For Sounds, both transcription systems (bipa=pyclts.TranscriptionSystem('bipa')) and sound class models allow the equivalent of

bipa[bipa['e']]

but this breaks for other symbols. Changing this requires (a) exporting Symbol, and replacing Sound with Symbol in

https://github.com/cldf/clts/blob/e65d4aabf6d1b739e7d1f203564dbb66571ddf22/src/pyclts/transcriptionsystem.py#L279

and in the equivalent place in soundclasses.py.

Is there any reason not to do this?

[LinguList]

I wonder why this would be needed? Sorry, it's too early in the morning here, so I have problems seeing the point.

[xrotwang]

Hm. If this doesn't lead to anything weird with UnknownSound, I think there's no particular reason for this limitation.

[Anaphory]

My use case for this is writing code that is ignorant about the data type it gets. The code should work on sequences of strings as well as on the sequences of Symbols that fall out of bipa(segment_string).

If the segment string contains unknown segments or even just markers, I get

>>> asjp[bipa['+']]
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "/vol/home/kaipingga/devel/clts/src/pyclts/models.py", line 47, in __getitem__
    return self.resolve_sound(sound)
  File "/vol/home/kaipingga/devel/clts/src/pyclts/soundclasses.py", line 46, in resolve_sound
    sound = sound if isinstance(sound, Sound) else self.system[sound]
  File "/vol/home/kaipingga/devel/clts/src/pyclts/models.py", line 47, in __getitem__
    return self.resolve_sound(sound)
  File "/vol/home/kaipingga/devel/clts/src/pyclts/transcriptionsystem.py", line 278, in resolve_sound
    if set(string.split(' ')).intersection(set(list(self.sound_classes)+['diphthong',
AttributeError: 'Marker' object has no attribute 'split'

where the desired behaviour would be to output the same as for

>>> asjp['+']
'+'

[LinguList]

okay, that looks convincing enough. Can you propose a PR directly on that?

[Anaphory]

Yep, about to.