GSW_AI_LAB/tempest-py/tests/utility/test_shortestpaths.py

import stormpy
import stormpy.logic
from stormpy.storage import BitVector
from stormpy.utility import ShortestPathsGenerator
from stormpy.utility import MatrixFormat
from helpers.helper import get_example_path

import pytest
import math


# this is admittedly slightly overengineered

class ModelWithKnownShortestPaths:
    """Knuth's die model with reference kSP methods"""

    def __init__(self):
        self.target_label = "one"

        program_path = get_example_path("dtmc", "die.pm")
        raw_formula = "P=? [ F \"" + self.target_label + "\" ]"
        program = stormpy.parse_prism_program(program_path)
        formulas = stormpy.parse_properties_for_prism_program(raw_formula, program)

        self.model = stormpy.build_model(program, formulas)

    def probability(self, k):
        return (1 / 2) ** ((2 * k) + 1)

    def state_set(self, k):
        return BitVector(self.model.nr_states, [0, 1, 3, 7])

    def path(self, k):
        path = [0] + k * [1, 3] + [7]
        return list(reversed(path))  # SPG returns traversal from back


@pytest.fixture(scope="module", params=[1, 2, 3, 3000, 42])
def index(request):
    return request.param


@pytest.fixture(scope="module")
def model_with_known_shortest_paths():
    return ModelWithKnownShortestPaths()


@pytest.fixture(scope="module")
def model(model_with_known_shortest_paths):
    return model_with_known_shortest_paths.model


@pytest.fixture(scope="module")
def expected_distance(model_with_known_shortest_paths):
    return model_with_known_shortest_paths.probability


@pytest.fixture(scope="module")
def expected_state_set(model_with_known_shortest_paths):
    return model_with_known_shortest_paths.state_set


@pytest.fixture(scope="module")
def expected_path(model_with_known_shortest_paths):
    return model_with_known_shortest_paths.path


@pytest.fixture(scope="module")
def target_label(model_with_known_shortest_paths):
    return model_with_known_shortest_paths.target_label


@pytest.fixture
def state(model):
    some_state = 7
    assert model.nr_states > some_state, "test model too small"
    return some_state


@pytest.fixture
def state_list(model):
    some_state_list = [4, 5, 7]
    assert model.nr_states > max(some_state_list), "test model too small"
    return some_state_list


@pytest.fixture
def state_bitvector(model, state_list):
    return BitVector(length=model.nr_states, set_entries=state_list)


@pytest.fixture
def transition_matrix(model):
    return model.transition_matrix


@pytest.fixture
def target_prob_map(model, state_list):
    return {i: (1.0 if i in state_list else 0.0) for i in range(model.nr_states)}


@pytest.fixture
def target_prob_list(target_prob_map):
    return [target_prob_map[i] for i in range(max(target_prob_map.keys()))]


@pytest.fixture
def initial_states(model):
    return BitVector(model.nr_states, model.initial_states)


@pytest.fixture
def matrix_format():
    return MatrixFormat.Straight


class TestShortestPaths:
    def test_spg_ctor_bitvector_target(self, model, state_bitvector):
        _ = ShortestPathsGenerator(model, state_bitvector)

    def test_spg_ctor_single_state_target(self, model, state):
        _ = ShortestPathsGenerator(model, state)

    def test_spg_ctor_state_list_target(self, model, state_list):
        _ = ShortestPathsGenerator(model, state_list)

    def test_spg_ctor_label_target(self, model, target_label):
        _ = ShortestPathsGenerator(model, target_label)

    def test_spg_ctor_matrix_vector(self, transition_matrix, target_prob_list, initial_states, matrix_format):
        _ = ShortestPathsGenerator(transition_matrix, target_prob_list, initial_states, matrix_format)

    def test_spg_ctor_matrix_map(self, transition_matrix, target_prob_map, initial_states, matrix_format):
        _ = ShortestPathsGenerator(transition_matrix, target_prob_map, initial_states, matrix_format)

    def test_spg_distance(self, model, target_label, index, expected_distance):
        spg = ShortestPathsGenerator(model, target_label)
        assert math.isclose(spg.get_distance(index), expected_distance(index))

    def test_spg_state_set(self, model, target_label, index, expected_state_set):
        spg = ShortestPathsGenerator(model, target_label)
        assert spg.get_states(index) == expected_state_set(index)

    def test_spg_state_list(self, model, target_label, index, expected_path):
        spg = ShortestPathsGenerator(model, target_label)
        assert spg.get_path_as_list(index) == expected_path(index)