# Author: Mainak Jas <mainak.jas@telecom-paristech.fr>
#         Mikolaj Magnuski <mmagnuski@swps.edu.pl>
#         Stefan Appelhoff <stefan.appelhoff@mailbox.org>
#
# License: BSD-3-Clause

from copy import deepcopy
import os.path as op
import shutil

import numpy as np
from numpy.testing import (assert_array_equal, assert_array_almost_equal,
                           assert_equal, assert_allclose)
import pytest
from scipy import io

import mne
from mne import write_events, read_epochs_eeglab
from mne.channels import read_custom_montage
from mne.io import read_raw_eeglab
from mne.io.eeglab.eeglab import _get_montage_information, _dol_to_lod
from mne.io.eeglab._eeglab import _readmat
from mne.io.tests.test_raw import _test_raw_reader
from mne.datasets import testing
from mne.utils import Bunch, _check_pymatreader_installed
from mne.annotations import events_from_annotations, read_annotations

base_dir = op.join(testing.data_path(download=False), 'EEGLAB')

raw_fname_mat = op.join(base_dir, 'test_raw.set')
raw_fname_onefile_mat = op.join(base_dir, 'test_raw_onefile.set')
raw_fname_event_duration = op.join(base_dir, 'test_raw_event_duration.set')
epochs_fname_mat = op.join(base_dir, 'test_epochs.set')
epochs_fname_onefile_mat = op.join(base_dir, 'test_epochs_onefile.set')
raw_mat_fnames = [raw_fname_mat, raw_fname_onefile_mat]
epochs_mat_fnames = [epochs_fname_mat, epochs_fname_onefile_mat]
raw_fname_chanloc = op.join(base_dir, 'test_raw_chanloc.set')
raw_fname_chanloc_fids = op.join(base_dir, 'test_raw_chanloc_fids.set')
raw_fname_2021 = op.join(base_dir, 'test_raw_2021.set')
raw_fname_h5 = op.join(base_dir, 'test_raw_h5.set')
raw_fname_onefile_h5 = op.join(base_dir, 'test_raw_onefile_h5.set')
epochs_fname_h5 = op.join(base_dir, 'test_epochs_h5.set')
epochs_fname_onefile_h5 = op.join(base_dir, 'test_epochs_onefile_h5.set')
raw_h5_fnames = [raw_fname_h5, raw_fname_onefile_h5]
epochs_h5_fnames = [epochs_fname_h5, epochs_fname_onefile_h5]

montage_path = op.join(base_dir, 'test_chans.locs')


@testing.requires_testing_data
@pytest.mark.parametrize('fname', [
    raw_fname_mat,
    pytest.param(
        raw_fname_h5,
        marks=[
            pytest.mark.skipif(
                not _check_pymatreader_installed(strict=False),
                reason='pymatreader not installed'
            )
        ]
    ),
    raw_fname_chanloc,
], ids=op.basename)
def test_io_set_raw(fname):
    """Test importing EEGLAB .set files."""
    montage = read_custom_montage(montage_path)
    montage.ch_names = [
        'EEG {0:03d}'.format(ii) for ii in range(len(montage.ch_names))
    ]

    kws = dict(reader=read_raw_eeglab, input_fname=fname)
    if fname.endswith('test_raw_chanloc.set'):
        with pytest.warns(RuntimeWarning,
                          match="The data contains 'boundary' events"):
            raw0 = _test_raw_reader(**kws)
    elif '_h5' in fname:  # should be safe enough, and much faster
        raw0 = read_raw_eeglab(fname, preload=True)
    else:
        raw0 = _test_raw_reader(**kws)

    # test that preloading works
    if fname.endswith('test_raw_chanloc.set'):
        raw0.set_montage(montage, on_missing='ignore')
        # crop to check if the data has been properly preloaded; we cannot
        # filter as the snippet of raw data is very short
        raw0.crop(0, 1)
    else:
        raw0.set_montage(montage)
        raw0.filter(1, None, l_trans_bandwidth='auto', filter_length='auto',
                    phase='zero')

    # test that using uint16_codec does not break stuff
    read_raw_kws = dict(input_fname=fname, preload=False, uint16_codec='ascii')
    if fname.endswith('test_raw_chanloc.set'):
        with pytest.warns(RuntimeWarning,
                          match="The data contains 'boundary' events"):
            raw0 = read_raw_eeglab(**read_raw_kws)
            raw0.set_montage(montage, on_missing='ignore')
    else:
        raw0 = read_raw_eeglab(**read_raw_kws)
        raw0.set_montage(montage)

    # Annotations
    if fname != raw_fname_chanloc:
        assert len(raw0.annotations) == 154
        assert set(raw0.annotations.description) == {'rt', 'square'}
        assert_array_equal(raw0.annotations.duration, 0.)


@testing.requires_testing_data
def test_io_set_raw_more(tmp_path):
    """Test importing EEGLAB .set files."""
    tmp_path = str(tmp_path)
    eeg = io.loadmat(raw_fname_mat, struct_as_record=False,
                     squeeze_me=True)['EEG']

    # test reading file with one event (read old version)
    negative_latency_fname = op.join(tmp_path, 'test_negative_latency.set')
    evnts = deepcopy(eeg.event[0])
    evnts.latency = 0
    io.savemat(negative_latency_fname,
               {'EEG': {'trials': eeg.trials, 'srate': eeg.srate,
                        'nbchan': eeg.nbchan,
                        'data': 'test_negative_latency.fdt',
                        'epoch': eeg.epoch, 'event': evnts,
                        'chanlocs': eeg.chanlocs, 'pnts': eeg.pnts}},
               appendmat=False, oned_as='row')
    shutil.copyfile(op.join(base_dir, 'test_raw.fdt'),
                    negative_latency_fname.replace('.set', '.fdt'))
    with pytest.warns(RuntimeWarning, match="has a sample index of -1."):
        read_raw_eeglab(input_fname=negative_latency_fname, preload=True)

    # test negative event latencies
    evnts.latency = -1
    io.savemat(negative_latency_fname,
               {'EEG': {'trials': eeg.trials, 'srate': eeg.srate,
                        'nbchan': eeg.nbchan,
                        'data': 'test_negative_latency.fdt',
                        'epoch': eeg.epoch, 'event': evnts,
                        'chanlocs': eeg.chanlocs, 'pnts': eeg.pnts}},
               appendmat=False, oned_as='row')
    with pytest.raises(ValueError, match='event sample index is negative'):
        with pytest.warns(RuntimeWarning, match="has a sample index of -1."):
            read_raw_eeglab(input_fname=negative_latency_fname, preload=True)

    # test overlapping events
    overlap_fname = op.join(tmp_path, 'test_overlap_event.set')
    io.savemat(overlap_fname,
               {'EEG': {'trials': eeg.trials, 'srate': eeg.srate,
                        'nbchan': eeg.nbchan, 'data': 'test_overlap_event.fdt',
                        'epoch': eeg.epoch,
                        'event': [eeg.event[0], eeg.event[0]],
                        'chanlocs': eeg.chanlocs, 'pnts': eeg.pnts}},
               appendmat=False, oned_as='row')
    shutil.copyfile(op.join(base_dir, 'test_raw.fdt'),
                    overlap_fname.replace('.set', '.fdt'))
    read_raw_eeglab(input_fname=overlap_fname, preload=True)

    # test reading file with empty event durations
    empty_dur_fname = op.join(tmp_path, 'test_empty_durations.set')
    evnts = deepcopy(eeg.event)
    for ev in evnts:
        ev.duration = np.array([], dtype='float')

    io.savemat(empty_dur_fname,
               {'EEG': {'trials': eeg.trials, 'srate': eeg.srate,
                        'nbchan': eeg.nbchan,
                        'data': 'test_negative_latency.fdt',
                        'epoch': eeg.epoch, 'event': evnts,
                        'chanlocs': eeg.chanlocs, 'pnts': eeg.pnts}},
               appendmat=False, oned_as='row')
    shutil.copyfile(op.join(base_dir, 'test_raw.fdt'),
                    empty_dur_fname.replace('.set', '.fdt'))
    raw = read_raw_eeglab(input_fname=empty_dur_fname, preload=True)
    assert (raw.annotations.duration == 0).all()

    # test reading file when the EEG.data name is wrong
    io.savemat(overlap_fname,
               {'EEG': {'trials': eeg.trials, 'srate': eeg.srate,
                        'nbchan': eeg.nbchan, 'data': 'test_overla_event.fdt',
                        'epoch': eeg.epoch,
                        'event': [eeg.event[0], eeg.event[0]],
                        'chanlocs': eeg.chanlocs, 'pnts': eeg.pnts}},
               appendmat=False, oned_as='row')
    with pytest.warns(RuntimeWarning, match="must have changed on disk"):
        read_raw_eeglab(input_fname=overlap_fname, preload=True)

    # raise error when both EEG.data and fdt name from set are wrong
    overlap_fname = op.join(tmp_path, 'test_ovrlap_event.set')
    io.savemat(overlap_fname,
               {'EEG': {'trials': eeg.trials, 'srate': eeg.srate,
                        'nbchan': eeg.nbchan, 'data': 'test_overla_event.fdt',
                        'epoch': eeg.epoch,
                        'event': [eeg.event[0], eeg.event[0]],
                        'chanlocs': eeg.chanlocs, 'pnts': eeg.pnts}},
               appendmat=False, oned_as='row')
    with pytest.raises(FileNotFoundError, match="not find the .fdt data file"):
        read_raw_eeglab(input_fname=overlap_fname, preload=True)

    # test reading file with one channel
    one_chan_fname = op.join(tmp_path, 'test_one_channel.set')
    io.savemat(one_chan_fname,
               {'EEG': {'trials': eeg.trials, 'srate': eeg.srate,
                        'nbchan': 1, 'data': np.random.random((1, 3)),
                        'epoch': eeg.epoch, 'event': eeg.epoch,
                        'chanlocs': {'labels': 'E1', 'Y': -6.6069,
                                     'X': 6.3023, 'Z': -2.9423},
                        'times': eeg.times[:3], 'pnts': 3}},
               appendmat=False, oned_as='row')
    read_raw_eeglab(input_fname=one_chan_fname, preload=True)

    # test reading file with 3 channels - one without position information
    # first, create chanlocs structured array
    ch_names = ['F3', 'unknown', 'FPz']
    x, y, z = [1., 2., np.nan], [4., 5., np.nan], [7., 8., np.nan]
    dt = [('labels', 'S10'), ('X', 'f8'), ('Y', 'f8'), ('Z', 'f8')]
    nopos_dt = [('labels', 'S10'), ('Z', 'f8')]
    chanlocs = np.zeros((3,), dtype=dt)
    nopos_chanlocs = np.zeros((3,), dtype=nopos_dt)
    for ind, vals in enumerate(zip(ch_names, x, y, z)):
        for fld in range(4):
            chanlocs[ind][dt[fld][0]] = vals[fld]
            if fld in (0, 3):
                nopos_chanlocs[ind][dt[fld][0]] = vals[fld]
    # In theory this should work and be simpler, but there is an obscure
    # SciPy writing bug that pops up sometimes:
    # nopos_chanlocs = np.array(chanlocs[['labels', 'Z']])

    # test reading channel names but not positions when there is no X (only Z)
    # field in the EEG.chanlocs structure
    nopos_fname = op.join(tmp_path, 'test_no_chanpos.set')
    io.savemat(nopos_fname,
               {'EEG': {'trials': eeg.trials, 'srate': eeg.srate, 'nbchan': 3,
                        'data': np.random.random((3, 2)), 'epoch': eeg.epoch,
                        'event': eeg.epoch, 'chanlocs': nopos_chanlocs,
                        'times': eeg.times[:2], 'pnts': 2}},
               appendmat=False, oned_as='row')
    # load the file
    raw = read_raw_eeglab(input_fname=nopos_fname, preload=True)

    # test that channel names have been loaded but not channel positions
    for i in range(3):
        assert_equal(raw.info['chs'][i]['ch_name'], ch_names[i])
        assert_array_equal(raw.info['chs'][i]['loc'][:3],
                           np.array([np.nan, np.nan, np.nan]))


@pytest.mark.timeout(60)  # ~60 sec on Travis OSX
@testing.requires_testing_data
@pytest.mark.parametrize('fnames', [
    epochs_mat_fnames,
    pytest.param(
        epochs_h5_fnames,
        marks=[
            pytest.mark.slowtest,
            pytest.mark.skipif(
                not _check_pymatreader_installed(strict=False),
                reason='pymatreader not installed'
            )
        ]
    )
])
def test_io_set_epochs(fnames):
    """Test importing EEGLAB .set epochs files."""
    epochs_fname, epochs_fname_onefile = fnames
    with pytest.warns(RuntimeWarning, match='multiple events'):
        epochs = read_epochs_eeglab(epochs_fname)
    with pytest.warns(RuntimeWarning, match='multiple events'):
        epochs2 = read_epochs_eeglab(epochs_fname_onefile)
    # one warning for each read_epochs_eeglab because both files have epochs
    # associated with multiple events
    assert_array_equal(epochs.get_data(), epochs2.get_data())


@testing.requires_testing_data
def test_io_set_epochs_events(tmp_path):
    """Test different combinations of events and event_ids."""
    tmp_path = str(tmp_path)
    out_fname = op.join(tmp_path, 'test-eve.fif')
    events = np.array([[4, 0, 1], [12, 0, 2], [20, 0, 3], [26, 0, 3]])
    write_events(out_fname, events)
    event_id = {'S255/S8': 1, 'S8': 2, 'S255/S9': 3}
    out_fname = op.join(tmp_path, 'test-eve.fif')
    epochs = read_epochs_eeglab(epochs_fname_mat, events, event_id)
    assert_equal(len(epochs.events), 4)
    assert epochs.preload
    assert epochs._bad_dropped
    epochs = read_epochs_eeglab(epochs_fname_mat, out_fname, event_id)
    pytest.raises(ValueError, read_epochs_eeglab, epochs_fname_mat,
                  None, event_id)
    pytest.raises(ValueError, read_epochs_eeglab, epochs_fname_mat,
                  epochs.events, None)


@testing.requires_testing_data
def test_degenerate(tmp_path):
    """Test some degenerate conditions."""
    # test if .dat file raises an error
    tmp_path = str(tmp_path)
    eeg = io.loadmat(epochs_fname_mat, struct_as_record=False,
                     squeeze_me=True)['EEG']
    eeg.data = 'epochs_fname.dat'
    bad_epochs_fname = op.join(tmp_path, 'test_epochs.set')
    io.savemat(bad_epochs_fname,
               {'EEG': {'trials': eeg.trials, 'srate': eeg.srate,
                        'nbchan': eeg.nbchan, 'data': eeg.data,
                        'epoch': eeg.epoch, 'event': eeg.event,
                        'chanlocs': eeg.chanlocs, 'pnts': eeg.pnts}},
               appendmat=False, oned_as='row')
    shutil.copyfile(op.join(base_dir, 'test_epochs.fdt'),
                    op.join(tmp_path, 'test_epochs.dat'))
    with pytest.warns(RuntimeWarning, match='multiple events'):
        pytest.raises(NotImplementedError, read_epochs_eeglab,
                      bad_epochs_fname)


@pytest.mark.parametrize("fname", [
    raw_fname_mat,
    raw_fname_onefile_mat,
    # We don't test the h5 varaints here because they are implicitly tested
    # in test_io_set_raw
])
@pytest.mark.filterwarnings('ignore: Complex objects')
@testing.requires_testing_data
def test_eeglab_annotations(fname):
    """Test reading annotations in EEGLAB files."""
    annotations = read_annotations(fname)
    assert len(annotations) == 154
    assert set(annotations.description) == {'rt', 'square'}
    assert np.all(annotations.duration == 0.)


@testing.requires_testing_data
def test_eeglab_read_annotations():
    """Test annotations onsets are timestamps (+ validate some)."""
    annotations = read_annotations(raw_fname_mat)
    validation_samples = [0, 1, 2, 3, 5, 7, 11, 13, 17, 19, 23, 29, 31]
    expected_onset = np.array([1.00, 1.69, 2.08, 4.70, 7.71, 11.30, 17.18,
                               20.20, 26.12, 29.14, 35.25, 44.30, 47.15])
    assert annotations.orig_time is None
    assert_array_almost_equal(annotations.onset[validation_samples],
                              expected_onset, decimal=2)

    # test if event durations are imported correctly
    raw = read_raw_eeglab(raw_fname_event_duration, preload=True)
    # file contains 3 annotations with 0.5 s (64 samples) duration each
    assert_allclose(raw.annotations.duration, np.ones(3) * 0.5)


@testing.requires_testing_data
def test_eeglab_event_from_annot():
    """Test all forms of obtaining annotations."""
    raw_fname_mat = op.join(base_dir, 'test_raw.set')
    raw_fname = raw_fname_mat
    event_id = {'rt': 1, 'square': 2}
    raw1 = read_raw_eeglab(input_fname=raw_fname, preload=False)

    annotations = read_annotations(raw_fname)
    assert len(raw1.annotations) == 154
    raw1.set_annotations(annotations)
    events_b, _ = events_from_annotations(raw1, event_id=event_id)
    assert len(events_b) == 154


def _assert_array_allclose_nan(left, right):
    assert_array_equal(np.isnan(left), np.isnan(right))
    assert_allclose(left[~np.isnan(left)], right[~np.isnan(left)], atol=1e-8)


@pytest.fixture(scope='session')
def one_chanpos_fname(tmp_path_factory):
    """Test file with 3 channels to exercise EEGLAB reader.

    File characteristics
       - ch_names: 'F3', 'unknown', 'FPz'
       - 'FPz' has no position information.
       - the rest is aleatory

    Notes from when this code was factorized:
    # test reading file with one event (read old version)
    """
    fname = str(tmp_path_factory.mktemp('data') / 'test_chanpos.set')
    file_conent = dict(EEG={
        'trials': 1, 'nbchan': 3, 'pnts': 3, 'epoch': [], 'event': [],
        'srate': 128, 'times': np.array([0., 0.1, 0.2]),
        'data': np.empty([3, 3]),
        'chanlocs': np.array(
            [(b'F3', 1., 4., 7.),
             (b'unknown', np.nan, np.nan, np.nan),
             (b'FPz', 2., 5., 8.)],
            dtype=[('labels', 'S10'), ('X', 'f8'), ('Y', 'f8'), ('Z', 'f8')]
        )
    })

    io.savemat(file_name=fname, mdict=file_conent, appendmat=False,
               oned_as='row')

    return fname


@testing.requires_testing_data
def test_position_information(one_chanpos_fname):
    """Test reading file with 3 channels - one without position information."""
    nan = np.nan
    EXPECTED_LOCATIONS_FROM_FILE = np.array([
        [-4.,  1.,  7.,  0.,  0.,  0., nan, nan, nan, nan, nan, nan],
        [nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan],
        [-5.,  2.,  8.,  0.,  0.,  0., nan, nan, nan, nan, nan, nan],
    ])

    EXPECTED_LOCATIONS_FROM_MONTAGE = np.array([
        [nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan],
        [nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan],
        [nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan, nan],
    ])

    raw = read_raw_eeglab(input_fname=one_chanpos_fname, preload=True)
    assert_array_equal(np.array([ch['loc'] for ch in raw.info['chs']]),
                       EXPECTED_LOCATIONS_FROM_FILE)

    # To accommodate the new behavior so that:
    # read_raw_eeglab(.. montage=montage) and raw.set_montage(montage)
    # behaves the same we need to flush the montage. otherwise we get
    # a mix of what is in montage and in the file
    raw = read_raw_eeglab(
        input_fname=one_chanpos_fname,
        preload=True,
    ).set_montage(None)  # Flush the montage builtin within input_fname

    _assert_array_allclose_nan(np.array([ch['loc'] for ch in raw.info['chs']]),
                               EXPECTED_LOCATIONS_FROM_MONTAGE)


@testing.requires_testing_data
def test_io_set_raw_2021():
    """Test reading new default file format (no EEG struct)."""
    assert "EEG" not in io.loadmat(raw_fname_2021)
    _test_raw_reader(reader=read_raw_eeglab, input_fname=raw_fname_2021,
                     test_preloading=False, preload=True)


@testing.requires_testing_data
def test_read_single_epoch():
    """Test reading raw set file as an Epochs instance."""
    with pytest.raises(ValueError, match='trials less than 2'):
        read_epochs_eeglab(raw_fname_mat)


@testing.requires_testing_data
def test_get_montage_info_with_ch_type():
    """Test that the channel types are properly returned."""
    mat = _readmat(raw_fname_onefile_mat)
    n = len(mat['EEG']['chanlocs']['labels'])
    mat['EEG']['chanlocs']['type'] = ['eeg'] * (n - 2) + ['eog'] + ['stim']
    mat['EEG']['chanlocs'] = _dol_to_lod(mat['EEG']['chanlocs'])
    mat['EEG'] = Bunch(**mat['EEG'])
    ch_names, ch_types, montage = _get_montage_information(mat['EEG'], False)
    assert len(ch_names) == len(ch_types) == n
    assert ch_types == ['eeg'] * (n - 2) + ['eog'] + ['stim']
    assert montage is None

    # test unknown type warning
    mat = _readmat(raw_fname_onefile_mat)
    n = len(mat['EEG']['chanlocs']['labels'])
    mat['EEG']['chanlocs']['type'] = ['eeg'] * (n - 2) + ['eog'] + ['unknown']
    mat['EEG']['chanlocs'] = _dol_to_lod(mat['EEG']['chanlocs'])
    mat['EEG'] = Bunch(**mat['EEG'])
    with pytest.warns(RuntimeWarning, match='Unknown types found'):
        ch_names, ch_types, montage = \
            _get_montage_information(mat['EEG'], False)


@testing.requires_testing_data
@pytest.mark.parametrize('has_type', (True, False))
def test_fidsposition_information(monkeypatch, has_type):
    """Test reading file with 3 fiducial locations."""
    if not has_type:
        def get_bad_information(eeg, get_pos):
            del eeg.chaninfo['nodatchans']['type']
            return _get_montage_information(eeg, get_pos)

        monkeypatch.setattr(mne.io.eeglab.eeglab, '_get_montage_information',
                            get_bad_information)
    raw = read_raw_eeglab(raw_fname_chanloc_fids)
    montage = raw.get_montage()
    pos = montage.get_positions()
    n_eeg = 129
    if not has_type:
        assert pos['nasion'] is None
        assert pos['lpa'] is None
        assert pos['rpa'] is None
        assert len(raw.info['dig']) == n_eeg
        return
    assert pos['nasion'] is not None
    assert pos['lpa'] is not None
    assert pos['rpa'] is not None
    assert len(pos['nasion']) == 3
    assert len(pos['lpa']) == 3
    assert len(pos['rpa']) == 3
    assert len(raw.info['dig']) == n_eeg + 3