mvcisback / dfa Goto Github PK

Hi,
I started using the package and tried to run the following code described at the README:

parity = DFA(
    start=0, inputs={0, 1}, label=lambda s: s,
    transition=lambda s, c: (s + c) & 1,
)

self_composed = parity | parity

assert self_composed.label([(0, 0), (1, 0)]) == (1, 0)

and I'm getting the following error:

TypeError: unsupported operand type(s) for |: 'DFA' and 'DFA'

dfa objects are incompatible with both pickle and dill.

import dfa
import dill

dfa1 = dfa.DFA(start=0,
               inputs = {0,1},
               label=lambda s: (s % 4) == 3,
               transition=lambda s, c: (s + c) % 4,
       )

with open('temp.dill', 'wb') as f:
    dill.dump(dfa1, f)

This code fails with error _pickle.PicklingError: Can't pickle <function _make_lookuper.<locals>.make_lookuper at 0x102a2b5e0>: it's not found as funcy.calc._make_lookuper.<locals>.make_lookuper.

In paths(...) in dfa/utils.py, there is a potential for an infinite loop when there are only a finite number of paths available for a given DFA, but paths is called a greater number of times than that.

@mvcisback and I discovered this issue when iterating on DFAs that were the result of XOR operations between two existing DFAs.

The following code fails, which is slightly unexpected:

dfa1 = DFA(
    start=0,
    inputs={0, 1},
    outputs=None,
    label=lambda s: (s % 4) == 3,
    transition=lambda s, c: (s + c) % 4,
)

I would expect this to be the same as not setting any outputs.
Setting an empty dictionary yields problems with the lstar library.