def test_get_logger_only_name_param(self):
log = get_logger('test_stream')
handler = log.handlers[0]
assert isinstance(handler, logging.StreamHandler)
assert logging.DEBUG == log.level
assert logging.DEBUG == handler.level
def test_get_logger_returns_a_logger_which_is_opening_a_file(
self, mock_open):
out_file = 'test_log.log'
log = get_logger('test_file', log_type='DISK', output_file=out_file)
handler = log.handlers[0]
assert isinstance(handler, logging.FileHandler)
assert logging.DEBUG == log.level
assert logging.DEBUG == handler.level
mock_open.assert_called_once()
def test_logger_writes_to_stream(self):
stream_name = 'test_stream'
stream_message = 'test_message'
log = get_logger(stream_name)
test_stream = io.StringIO()
# replace the real stream with a stream we have control over
log.handlers[0].stream = test_stream
log.debug(stream_message)
stream_content = test_stream.getvalue()
test_stream.close()
assert stream_content is not None
assert '' != stream_content
# We are not checkning the entire string due to time dependence
debug_message = f'{stream_name} - DEBUG - {stream_message}\n'
assert debug_message == ' - '.join(stream_content.split(' - ')[1:])
def test_log_type_disk_without_output_param_raises_value_error(self):
with pytest.raises(ValueError):
get_logger('test_file', log_type='DISK')
from abc import ABCMeta, abstractmethod
import numpy as np
import pandas as pd
from mim.util.logs import get_logger
log = get_logger('Cross Validation')
class ClassBalance(metaclass=ABCMeta):
def balance(self, x, indices) -> np.array:
x = x.iloc[indices, :]
positive = x[x['labels'] > 0].loc[:, 'index'].values
negative = np.setdiff1d(indices, positive)
balanced = self.sample(positive, negative)
return balanced.sort()
def sample(self, positive, negative) -> (np.array, np.array):
return np.append(positive, negative)
class NullBalance(ClassBalance):
def balance(self, x, indices):
return indices
class DownSample(ClassBalance):
def sample(self, positive, negative) -> (np.array, np.array):
if len(negative) >= len(positive):
import re
import numpy as np
import pandas as pd
from sklearn.metrics import (confusion_matrix, precision_score, recall_score,
accuracy_score, f1_score)
from mim.util.logs import get_logger
from mim.util.util import ranksort
from mim.experiments.experiments import result_path
from mim.experiments.factory import experiment_from_name
log = get_logger("Presenter")
class Presenter:
def __init__(self, name):
self.results = dict()
log.debug(f'Loading all experiments for {name}')
self.experiments = experiment_from_name(name)
for xp in self.experiments:
path = result_path(xp)
try:
self.results[xp.name] = pd.read_pickle(path)
except FileNotFoundError:
log.debug(f"Test {xp.name} doesn't exist in path {path}")
def describe(self, like='.*'):
results = []
for name, xp in self._results_that_match_pattern(like):
import argparse
from time import time
import numpy as np
import pandas as pd
from tqdm import tqdm
from .factory import experiment_from_name
from mim.util.metadata import Metadata
from mim.util.logs import get_logger
log = get_logger("Run")
def run_experiments(experiments, continue_on_error=False):
"""
Run all experiments in xps and save the results to disk.
:param experiments: List of experiments to run
:param continue_on_error: Whether to continue running experiments even
if one should fail.
"""
for experiment in experiments:
try:
pd.to_pickle(run_one_experiment(experiment),
experiment.result_path)
except Exception as e:
log.error(f'Something went wrong with task {experiment.name}! '
f'Oh no! :(')
if not continue_on_error:
raise e