"""
Testing for grid search module (sklearn.grid_search)

"""

from collections import Iterable, Sized
from sklearn.externals.six.moves import cStringIO as StringIO
from sklearn.externals.six.moves import xrange
from itertools import chain, product
import pickle
import sys

import numpy as np
import scipy.sparse as sp

from sklearn.utils.testing import assert_equal
from sklearn.utils.testing import assert_not_equal
from sklearn.utils.testing import assert_raises
from sklearn.utils.testing import assert_warns
from sklearn.utils.testing import assert_raise_message
from sklearn.utils.testing import assert_false, assert_true
from sklearn.utils.testing import assert_array_equal
from sklearn.utils.testing import assert_almost_equal
from sklearn.utils.testing import assert_array_almost_equal
from sklearn.utils.testing import assert_no_warnings
from sklearn.utils.testing import ignore_warnings
from sklearn.utils.mocking import CheckingClassifier, MockDataFrame

from scipy.stats import bernoulli, expon, uniform

from sklearn.externals.six.moves import zip
from sklearn.base import BaseEstimator
from sklearn.datasets import make_classification
from sklearn.datasets import make_blobs
from sklearn.datasets import make_multilabel_classification
from sklearn.grid_search import (GridSearchCV, RandomizedSearchCV,
                                 ParameterGrid, ParameterSampler,
                                 ChangedBehaviorWarning)
from sklearn.svm import LinearSVC, SVC
from sklearn.tree import DecisionTreeRegressor
from sklearn.tree import DecisionTreeClassifier
from sklearn.cluster import KMeans
from sklearn.neighbors import KernelDensity
from sklearn.metrics import f1_score
from sklearn.metrics import make_scorer
from sklearn.metrics import roc_auc_score
from sklearn.cross_validation import KFold, StratifiedKFold, FitFailedWarning
from sklearn.preprocessing import Imputer
from sklearn.pipeline import Pipeline


# Neither of the following two estimators inherit from BaseEstimator,
# to test hyperparameter search on user-defined classifiers.
class MockClassifier(object):
    """Dummy classifier to test the cross-validation"""
    def __init__(self, foo_param=0):
        self.foo_param = foo_param

    def fit(self, X, Y):
        assert_true(len(X) == len(Y))
        return self

    def predict(self, T):
        return T.shape[0]

    predict_proba = predict
    decision_function = predict
    transform = predict

    def score(self, X=None, Y=None):
        if self.foo_param > 1:
            score = 1.
        else:
            score = 0.
        return score

    def get_params(self, deep=False):
        return {'foo_param': self.foo_param}

    def set_params(self, **params):
        self.foo_param = params['foo_param']
        return self


class LinearSVCNoScore(LinearSVC):
    """An LinearSVC classifier that has no score method."""
    @property
    def score(self):
        raise AttributeError

X = np.array([[-1, -1], [-2, -1], [1, 1], [2, 1]])
y = np.array([1, 1, 2, 2])


def assert_grid_iter_equals_getitem(grid):
    assert_equal(list(grid), [grid[i] for i in range(len(grid))])


def test_parameter_grid():
    # Test basic properties of ParameterGrid.
    params1 = {"foo": [1, 2, 3]}
    grid1 = ParameterGrid(params1)
    assert_true(isinstance(grid1, Iterable))
    assert_true(isinstance(grid1, Sized))
    assert_equal(len(grid1), 3)
    assert_grid_iter_equals_getitem(grid1)

    params2 = {"foo": [4, 2],
               "bar": ["ham", "spam", "eggs"]}
    grid2 = ParameterGrid(params2)
    assert_equal(len(grid2), 6)

    # loop to assert we can iterate over the grid multiple times
    for i in xrange(2):
        # tuple + chain transforms {"a": 1, "b": 2} to ("a", 1, "b", 2)
        points = set(tuple(chain(*(sorted(p.items())))) for p in grid2)
        assert_equal(points,
                     set(("bar", x, "foo", y)
                         for x, y in product(params2["bar"], params2["foo"])))

    assert_grid_iter_equals_getitem(grid2)

    # Special case: empty grid (useful to get default estimator settings)
    empty = ParameterGrid({})
    assert_equal(len(empty), 1)
    assert_equal(list(empty), [{}])
    assert_grid_iter_equals_getitem(empty)
    assert_raises(IndexError, lambda: empty[1])

    has_empty = ParameterGrid([{'C': [1, 10]}, {}, {'C': [.5]}])
    assert_equal(len(has_empty), 4)
    assert_equal(list(has_empty), [{'C': 1}, {'C': 10}, {}, {'C': .5}])
    assert_grid_iter_equals_getitem(has_empty)


def test_grid_search():
    # Test that the best estimator contains the right value for foo_param
    clf = MockClassifier()
    grid_search = GridSearchCV(clf, {'foo_param': [1, 2, 3]}, verbose=3)
    # make sure it selects the smallest parameter in case of ties
    old_stdout = sys.stdout
    sys.stdout = StringIO()
    grid_search.fit(X, y)
    sys.stdout = old_stdout
    assert_equal(grid_search.best_estimator_.foo_param, 2)

    for i, foo_i in enumerate([1, 2, 3]):
        assert_true(grid_search.grid_scores_[i][0]
                    == {'foo_param': foo_i})
    # Smoke test the score etc:
    grid_search.score(X, y)
    grid_search.predict_proba(X)
    grid_search.decision_function(X)
    grid_search.transform(X)

    # Test exception handling on scoring
    grid_search.scoring = 'sklearn'
    assert_raises(ValueError, grid_search.fit, X, y)


@ignore_warnings
def test_grid_search_no_score():
    # Test grid-search on classifier that has no score function.
    clf = LinearSVC(random_state=0)
    X, y = make_blobs(random_state=0, centers=2)
    Cs = [.1, 1, 10]
    clf_no_score = LinearSVCNoScore(random_state=0)
    grid_search = GridSearchCV(clf, {'C': Cs}, scoring='accuracy')
    grid_search.fit(X, y)

    grid_search_no_score = GridSearchCV(clf_no_score, {'C': Cs},
                                        scoring='accuracy')
    # smoketest grid search
    grid_search_no_score.fit(X, y)

    # check that best params are equal
    assert_equal(grid_search_no_score.best_params_, grid_search.best_params_)
    # check that we can call score and that it gives the correct result
    assert_equal(grid_search.score(X, y), grid_search_no_score.score(X, y))

    # giving no scoring function raises an error
    grid_search_no_score = GridSearchCV(clf_no_score, {'C': Cs})
    assert_raise_message(TypeError, "no scoring", grid_search_no_score.fit,
                         [[1]])


def test_grid_search_score_method():
    X, y = make_classification(n_samples=100, n_classes=2, flip_y=.2,
                               random_state=0)
    clf = LinearSVC(random_state=0)
    grid = {'C': [.1]}

    search_no_scoring = GridSearchCV(clf, grid, scoring=None).fit(X, y)
    search_accuracy = GridSearchCV(clf, grid, scoring='accuracy').fit(X, y)
    search_no_score_method_auc = GridSearchCV(LinearSVCNoScore(), grid,
                                              scoring='roc_auc').fit(X, y)
    search_auc = GridSearchCV(clf, grid, scoring='roc_auc').fit(X, y)

    # Check warning only occurs in situation where behavior changed:
    # estimator requires score method to compete with scoring parameter
    score_no_scoring = assert_no_warnings(search_no_scoring.score, X, y)
    score_accuracy = assert_warns(ChangedBehaviorWarning,
                                  search_accuracy.score, X, y)
    score_no_score_auc = assert_no_warnings(search_no_score_method_auc.score,
                                            X, y)
    score_auc = assert_warns(ChangedBehaviorWarning,
                             search_auc.score, X, y)
    # ensure the test is sane
    assert_true(score_auc < 1.0)
    assert_true(score_accuracy < 1.0)
    assert_not_equal(score_auc, score_accuracy)

    assert_almost_equal(score_accuracy, score_no_scoring)
    assert_almost_equal(score_auc, score_no_score_auc)


def test_trivial_grid_scores():
    # Test search over a "grid" with only one point.
    # Non-regression test: grid_scores_ wouldn't be set by GridSearchCV.
    clf = MockClassifier()
    grid_search = GridSearchCV(clf, {'foo_param': [1]})
    grid_search.fit(X, y)
    assert_true(hasattr(grid_search, "grid_scores_"))

    random_search = RandomizedSearchCV(clf, {'foo_param': [0]}, n_iter=1)
    random_search.fit(X, y)
    assert_true(hasattr(random_search, "grid_scores_"))


def test_no_refit():
    # Test that grid search can be used for model selection only
    clf = MockClassifier()
    grid_search = GridSearchCV(clf, {'foo_param': [1, 2, 3]}, refit=False)
    grid_search.fit(X, y)
    assert_true(hasattr(grid_search, "best_params_"))


def test_grid_search_error():
    # Test that grid search will capture errors on data with different
    # length
    X_, y_ = make_classification(n_samples=200, n_features=100, random_state=0)

    clf = LinearSVC()
    cv = GridSearchCV(clf, {'C': [0.1, 1.0]})
    assert_raises(ValueError, cv.fit, X_[:180], y_)


def test_grid_search_iid():
    # test the iid parameter
    # noise-free simple 2d-data
    X, y = make_blobs(centers=[[0, 0], [1, 0], [0, 1], [1, 1]], random_state=0,
                      cluster_std=0.1, shuffle=False, n_samples=80)
    # split dataset into two folds that are not iid
    # first one contains data of all 4 blobs, second only from two.
    mask = np.ones(X.shape[0], dtype=np.bool)
    mask[np.where(y == 1)[0][::2]] = 0
    mask[np.where(y == 2)[0][::2]] = 0
    # this leads to perfect classification on one fold and a score of 1/3 on
    # the other
    svm = SVC(kernel='linear')
    # create "cv" for splits
    cv = [[mask, ~mask], [~mask, mask]]
    # once with iid=True (default)
    grid_search = GridSearchCV(svm, param_grid={'C': [1, 10]}, cv=cv)
    grid_search.fit(X, y)
    first = grid_search.grid_scores_[0]
    assert_equal(first.parameters['C'], 1)
    assert_array_almost_equal(first.cv_validation_scores, [1, 1. / 3.])
    # for first split, 1/4 of dataset is in test, for second 3/4.
    # take weighted average
    assert_almost_equal(first.mean_validation_score,
                        1 * 1. / 4. + 1. / 3. * 3. / 4.)

    # once with iid=False
    grid_search = GridSearchCV(svm, param_grid={'C': [1, 10]}, cv=cv,
                               iid=False)
    grid_search.fit(X, y)
    first = grid_search.grid_scores_[0]
    assert_equal(first.parameters['C'], 1)
    # scores are the same as above
    assert_array_almost_equal(first.cv_validation_scores, [1, 1. / 3.])
    # averaged score is just mean of scores
    assert_almost_equal(first.mean_validation_score,
                        np.mean(first.cv_validation_scores))


def test_grid_search_one_grid_point():
    X_, y_ = make_classification(n_samples=200, n_features=100, random_state=0)
    param_dict = {"C": [1.0], "kernel": ["rbf"], "gamma": [0.1]}

    clf = SVC()
    cv = GridSearchCV(clf, param_dict)
    cv.fit(X_, y_)

    clf = SVC(C=1.0, kernel="rbf", gamma=0.1)
    clf.fit(X_, y_)

    assert_array_equal(clf.dual_coef_, cv.best_estimator_.dual_coef_)


def test_grid_search_bad_param_grid():
    param_dict = {"C": 1.0}
    clf = SVC()
    assert_raises(ValueError, GridSearchCV, clf, param_dict)

    param_dict = {"C": []}
    clf = SVC()
    assert_raises(ValueError, GridSearchCV, clf, param_dict)

    param_dict = {"C": np.ones(6).reshape(3, 2)}
    clf = SVC()
    assert_raises(ValueError, GridSearchCV, clf, param_dict)


def test_grid_search_sparse():
    # Test that grid search works with both dense and sparse matrices
    X_, y_ = make_classification(n_samples=200, n_features=100, random_state=0)

    clf = LinearSVC()
    cv = GridSearchCV(clf, {'C': [0.1, 1.0]})
    cv.fit(X_[:180], y_[:180])
    y_pred = cv.predict(X_[180:])
    C = cv.best_estimator_.C

    X_ = sp.csr_matrix(X_)
    clf = LinearSVC()
    cv = GridSearchCV(clf, {'C': [0.1, 1.0]})
    cv.fit(X_[:180].tocoo(), y_[:180])
    y_pred2 = cv.predict(X_[180:])
    C2 = cv.best_estimator_.C

    assert_true(np.mean(y_pred == y_pred2) >= .9)
    assert_equal(C, C2)


def test_grid_search_sparse_scoring():
    X_, y_ = make_classification(n_samples=200, n_features=100, random_state=0)

    clf = LinearSVC()
    cv = GridSearchCV(clf, {'C': [0.1, 1.0]}, scoring="f1")
    cv.fit(X_[:180], y_[:180])
    y_pred = cv.predict(X_[180:])
    C = cv.best_estimator_.C

    X_ = sp.csr_matrix(X_)