"""
Utility tools for writing unit tests for packages which rely on `sentinelhub-py`
"""
import os
from typing import Any, Callable, Dict, Optional, Tuple, Union
import numpy as np
from pytest import approx
from .exceptions import deprecated_function
[docs]def get_output_folder(current_file: str) -> str:
"""Use fixtures if possible. This is meant only for test cases"""
return os.path.join(os.path.dirname(os.path.realpath(current_file)), "TestOutputs")
[docs]@deprecated_function(message_suffix="Use `assert_statistics_match` instead.")
def test_numpy_data(
data: Optional[np.ndarray] = None,
exp_shape: Optional[Tuple[int, ...]] = None,
exp_dtype: Union[None, type, np.dtype] = None,
exp_min: Optional[float] = None,
exp_max: Optional[float] = None,
exp_mean: Optional[float] = None,
exp_median: Optional[float] = None,
exp_std: Optional[float] = None,
delta: float = 1e-4,
) -> None:
"""Deprecated version of `assert_statistics_match`."""
if data is None:
return
assert_statistics_match(
data, exp_shape, exp_dtype, exp_min, exp_max, exp_mean, exp_median, exp_std, rel_delta=delta
)
[docs]def assert_statistics_match(
data: np.ndarray,
exp_shape: Optional[Tuple[int, ...]] = None,
exp_dtype: Union[None, type, np.dtype] = None,
exp_min: Optional[float] = None,
exp_max: Optional[float] = None,
exp_mean: Optional[float] = None,
exp_median: Optional[float] = None,
exp_std: Optional[float] = None,
rel_delta: Optional[float] = None,
abs_delta: Optional[float] = None,
) -> None:
"""Validates basic statistics of data array
:param data: Data array
:param exp_shape: Expected shape
:param exp_dtype: Expected dtype
:param exp_min: Expected minimal value
:param exp_max: Expected maximal value
:param exp_mean: Expected mean value
:param exp_median: Expected median value
:param exp_std: Expected standard deviation value
:param rel_delta: Precision of validation (relative)
:param abs_delta: Precision of validation (absolute)
"""
stats_suite: Dict[str, Tuple[Callable[[np.ndarray], Any], Any]] = {
"shape": (lambda array: array.shape, exp_shape),
"dtype": (lambda array: array.dtype, exp_dtype),
"min": (np.nanmin, exp_min),
"max": (np.nanmax, exp_max),
"mean": (np.nanmean, exp_mean),
"median": (np.nanmedian, exp_median),
"std": (np.nanstd, exp_std),
}
is_precise = {"shape", "dtype"}
data_stats: Dict[str, Any] = {}
exp_stats: Dict[str, Any] = {}
for name, (func, expected) in stats_suite.items():
if expected is not None:
data_stats[name] = func(data)
exp_stats[name] = expected if name in is_precise else approx(expected, rel=rel_delta, abs=abs_delta)
assert data_stats == exp_stats, "Statistics differ from expected values"