def check_grammar_induction_correctness(
self,
expected_grammar: ContextFreeGrammar,
dataset: List[str] = None,
words_per_slot=1,
prune_redundant=True,
minimal_variables=True,
) -> ContextFreeGrammar:
if dataset is None:
dataset = expected_grammar.generate_all_string()
induced_grammar = grammar_induction.induce_grammar_using_template_trees(
dataset,
words_per_slot=words_per_slot,
prune_redundant=prune_redundant,
minimal_variables=minimal_variables,
)
print(induced_grammar)
# Check if same dataset generation
self.check_grammar_expansion(induced_grammar, dataset)
# Check if isomorph grammar
self.assertTrue(expected_grammar.is_isomorphic_with(induced_grammar))
# Check that the grammar is representable as string, without exception
self.assertTrue(len(str(induced_grammar)) > 0)
return induced_grammar
def check_grammar_expansion(self, grammar: ContextFreeGrammar,
expected_expansion: Collection[str]):
""" Check that grammar indeed generates the dataset it learned from """
generated_dataset = grammar.generate_all_string()
self.assertEqual(set(expected_expansion), set(generated_dataset))