WHAT

Lexicon is a simple collection of Python dict subclasses providing extra
power:

AliasDict, a dictionary supporting both simple and complex key aliasing:
- Alias a single key to another key, so that e.g. mydict['bar'] points
  to mydict['foo'], for both reads and writes.
- Alias a single key to a list of other keys, for writing only, e.g. with
  active_groups = AliasDict({'ops': True, 'biz': True, 'dev': True, 'product': True}) one can make an alias 'tech' mapping to ('ops', 'dev') and then e.g. active_groups['tech'] = False.
- Aliasing is recursive: an alias pointing to another alias will behave as
  if it points to the other alias' target.
AttributeDict, supporting attribute read & write access, e.g. mydict = AttributeDict({'foo': 'bar'}) exhibits mydict.foo and mydict.foo = 'new value'.
Lexicon, a subclass of both of the above which exhibits both sets of
behavior.

HOW

Regular use:

Development/hacking/test suite:

Install dev dependencies via any of the following:
- pip install --group dev (assumes pip >= 25.1)
- uv sync
- anything else that understands PEP 735 dependency groups
inv test (or inv --list to see other dev tasks)

In all examples, 'myalias' is the alias and 'realkey' is the "real",
unaliased key.

alias(from_'myalias', to='realkey'): Alias myalias to realkey so
d['myalias'] behaves exactly like d['realkey'] for both reads and
writes.
- from_ is the first keyword argument, but typically it can be omitted
  and still reads fine. See below examples for this usage. See below for
  details on how an alias affects other dict operations.
alias('myalias', to=('realkey', 'otherrealkey')): Alias myalias to
both realkey and otherrealkey. As you might expect, this only works
well for writes, as there is never any guarantee that all targets of the
alias will contain the same value.
unalias('myalias'): Removes the myalias alias; any subsequent
reads/writes to myalias will behave as normal for a regular dict.
'myalias' in d (aka __contains__): Returns True when given an alias,
so if myalias is an alias to some other key, dictionary membership tests
will behave as if myalias is set.
del d['myalias'] (aka __delitem__): This effectively becomes del d['realkey'] -- to remove the alias itself, use unalias().
del d['realkey']: Deletes the real key/value pair (i.e. it calls
dict.__del__) but doesn't touch any aliases pointing to realkey.
- As a result, "dangling" aliases pointing to nonexistent keys will raise
  KeyError on access, but will continue working if the target key is
  repopulated later.

Caveats:

Because of the single-key/multi-key duality, AliasDict is incapable of
honoring non-string-type keys when aliasing (it must test isinstance(key, basestring) to tell strings apart from non-string iterables).
- AliasDict instances may still use non-string keys, of course -- it
  just can't use them as alias targets.

d.key = 'value' (aka __setattr__): Maps directly to d['key'] = 'value'.
d.key (aka __getattr__): Maps directly to d['key'].
del d.key (aka __delattr__): Maps directly to del d['key'].
Collisions between "real" or pre-existing attributes, and
attributes-as-dict-keys, always results in the real attribute winning. Thus
it isn't possible to use attribute access to access e.g. d['get'].

Lexicon subclasses from AttributeDict first, then AliasDict, with the
end result that attribute access will honor aliases. E.g.:

d = Lexicon()
d.alias('myalias', to='realkey')
d.myalias = 'foo'
print d.realkey # prints 'foo'