from icosagon.data import Data, \
    _equal
from icosagon.trainprep import prepare_training, \
    TrainValTest
from icosagon.model import Model
from icosagon.trainloop import TrainLoop
import torch
import pytest
import pdb
import time


def test_train_loop_01():
    d = Data()
    d.add_node_type('Dummy', 10)
    fam = d.add_relation_family('Dummy-Dummy', 0, 0, False)
    fam.add_relation_type('Dummy Rel', torch.rand(10, 10).round())

    prep_d = prepare_training(d, TrainValTest(.8, .1, .1))

    m = Model(prep_d)

    loop = TrainLoop(m)

    assert loop.model == m
    assert loop.lr == 0.001
    assert loop.loss == torch.nn.functional.binary_cross_entropy_with_logits
    assert loop.batch_size == 100


def test_train_loop_02():
    d = Data()
    d.add_node_type('Dummy', 10)
    fam = d.add_relation_family('Dummy-Dummy', 0, 0, False)
    fam.add_relation_type('Dummy Rel', torch.rand(10, 10).round())

    prep_d = prepare_training(d, TrainValTest(.8, .1, .1))

    m = Model(prep_d)

    for prm in m.parameters():
        print(prm.shape, prm.is_leaf, prm.requires_grad)

    loop = TrainLoop(m)

    loop.run_epoch()

    for prm in m.parameters():
        print(prm.shape, prm.is_leaf, prm.requires_grad)


def test_train_loop_03():
    # pdb.set_trace()
    if torch.cuda.device_count() == 0:
        pytest.skip('CUDA required for this test')

    adj_mat = torch.rand(10, 10).round()
    dev = torch.device('cuda:0')
    adj_mat = adj_mat.to(dev)

    d = Data()
    d.add_node_type('Dummy', 10)
    fam = d.add_relation_family('Dummy-Dummy', 0, 0, False)
    fam.add_relation_type('Dummy Rel', adj_mat)

    prep_d = prepare_training(d, TrainValTest(.8, .1, .1))
    # pdb.set_trace()

    m = Model(prep_d)
    m = m.to(dev)

    print(list(m.parameters()))

    for prm in m.parameters():
        assert prm.device == dev

    loop = TrainLoop(m)

    loop.run_epoch()


def test_train_loop_04():
    adj_mat = torch.rand(10, 10).round()

    d = Data()
    d.add_node_type('Dummy', 10)
    fam = d.add_relation_family('Dummy-Dummy', 0, 0, False)
    fam.add_relation_type('Dummy Rel', adj_mat)

    prep_d = prepare_training(d, TrainValTest(.8, .1, .1))

    m = Model(prep_d)

    old_values = []
    for prm in m.parameters():
        old_values.append(prm.clone().detach())

    loop = TrainLoop(m)

    loop.run_epoch()

    for i, prm in enumerate(m.parameters()):
        assert not prm.requires_grad or \
            not torch.all(_equal(prm, old_values[i]))


def test_train_loop_05():
    adj_mat = torch.rand(10, 10).round().to_sparse()

    d = Data()
    d.add_node_type('Dummy', 10)
    fam = d.add_relation_family('Dummy-Dummy', 0, 0, False)
    fam.add_relation_type('Dummy Rel', adj_mat)

    prep_d = prepare_training(d, TrainValTest(.8, .1, .1))

    m = Model(prep_d)

    old_values = []
    for prm in m.parameters():
        old_values.append(prm.clone().detach())

    loop = TrainLoop(m)

    loop.run_epoch()

    for i, prm in enumerate(m.parameters()):
        assert not prm.requires_grad or \
            not torch.all(_equal(prm, old_values[i]))


def test_timing_01():
    adj_mat = (torch.rand(2000, 2000) < .001).to(torch.float32).to_sparse()
    rep = torch.eye(2000).requires_grad_(True)
    t = time.time()
    for _ in range(1300):
        _ = torch.sparse.mm(adj_mat, rep)
    print('Elapsed:', time.time() - t)


def test_timing_02():
    adj_mat = (torch.rand(2000, 2000) < .001).to(torch.float32)
    adj_mat_batch = [adj_mat.view(1, 2000, 2000)] * 1300
    adj_mat_batch = torch.cat(adj_mat_batch)
    rep = torch.eye(2000).requires_grad_(True)
    t = time.time()
    res = torch.matmul(adj_mat_batch, rep)
    print('Elapsed:', time.time() - t)
    print('res.shape:', res.shape)


def test_timing_03():
    adj_mat = (torch.rand(2000, 2000) < .001).to(torch.float32)
    adj_mat_batch = [adj_mat.view(1, 2000, 2000).to_sparse()] * 1300
    adj_mat_batch = torch.cat(adj_mat_batch)
    rep = torch.eye(2000).requires_grad_(True)
    rep_batch = [rep.view(1, 2000, 2000)] * 1300
    rep_batch = torch.cat(rep_batch)
    t = time.time()
    with pytest.raises(RuntimeError):
        _ = torch.bmm(adj_mat_batch, rep)
    print('Elapsed:', time.time() - t)


def test_timing_04():
    adj_mat = (torch.rand(2000, 2000) < .0001).to(torch.float32).to_sparse()
    rep = torch.eye(2000).requires_grad_(True)
    t = time.time()
    for _ in range(1300):
        _ = torch.sparse.mm(adj_mat, rep)
    print('Elapsed:', time.time() - t)


def test_timing_05():
    if torch.cuda.device_count() == 0:
        pytest.skip('Test requires CUDA')
    dev = torch.device('cuda:0')
    adj_mat = (torch.rand(2000, 2000) < .001).to(torch.float32).to_sparse().to(dev)
    rep = torch.eye(2000).requires_grad_(True).to(dev)
    t = time.time()
    for _ in range(1300):
        _ = torch.sparse.mm(adj_mat, rep)
    torch.cuda.synchronize()
    print('Elapsed:', time.time() - t)