data("gam_example")
set.seed(42L)

test_that("rmwcs solver can be constructed with specific parameters", {
    solver <- rmwcs_solver()
    expect_equal(solver$separation, "strong")
    timelimit(solver) <- 20
    expect_equal(solver$timelimit, 20)
})


test_that("one vertex best solution is returned", {
    solver <- rmwcs_solver()
    g <- igraph::make_graph(c("A", "B"), directed = FALSE)
    V(g)$weight <- c(5, -2)
    solution <- solve_mwcsp(solver, g)
    expect_equal(solution$weight, 5)
})

test_that("rmwcs solver works on specific test", {
    solver <- rmwcs_solver()
    g <- make_ring(5) %>% set.vertex.attribute("weight", value = 2:-2)
    s <- solve_mwcsp(solver, g)
    expect_equal(s$weight, 3)
    expect_true(s$solved_to_optimality)
})

test_that("rmwcs solver doesn't crash on simple graphs", {
    solver <- rmwcs_solver()
    size <- 10
    card <- 3

    test_graph <- function(make) {
        scores <- runif(size) - 0.5;
        g <- make(size) %>% set.vertex.attribute("weight", value = scores)
        solve_mwcsp(solver, g)

        solution <- solve_mwcsp(solver, g, max_cardinality = card)
        expect_lte(length(V(solution$graph)), 3)

        g <- set.vertex.attribute(g, name = "budget_cost", value = runif(size))
        V(g)$weight <- 1
        solution <- solve_mwcsp(solver, g, budget = card)

        expect_lte(sum(V(solution$graph)$budget_cost), card)
    }

    test_graph(make_ring)
    test_graph(function(x) make_star(x, mode = "undirected"))
    test_graph(make_full_graph)
})

test_that("rmwcs solver builds connected solutions on a GAM instance", {
    solver <- rmwcs_solver()
    sol <- solve_mwcsp(solver, gam_example)
    expect_true(is.connected(sol$graph))
})

test_that("rmwcs solver works with empty solutions", {
    solver <- rmwcs_solver()
    g <- graph.ring(10)
    V(g)$weight <- -1
    sol <- solve_mwcsp(solver, g)
    expect_true(vcount(sol$graph) == 0)
})