gitea_admin/sientia-dataops-scouter_temporal
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

SIENTIAPDE-1316

Browse Source 
Update .gitignore and refactor metrics.py, activities.py, and gates.py for improved clarity and consistency. Added coverage.xml and cache directories to .gitignore. Standardized string formatting and parameter handling in metrics and activities classes, enhancing code readability. Removed the deprecated faker.py file and adjusted related tests accordingly.
This commit is contained in:
vitor-aignosi
2025-10-16 13:31:12 -03:00
parent 8bdbf049b8
commit 97eb5bc904
27 changed files with 1101 additions and 1336 deletions
Download Patch File Download Diff File Expand all files Collapse all files

45
tests/utils/quality/test_filters.py
View File

@@ -1,8 +1,9 @@
import pytest
import pandas as pd
import numpy as np
import pandas as pd
import pytest
from pandas.testing import assert_frame_equal
from scouter.utils.quality.filters import check_data_range, out_of_bounds_filter, null_values_filter
from scouter.utils.quality.filters import check_data_range, null_values_filter, out_of_bounds_filter
# Fixtures
@@ -10,12 +11,14 @@ from scouter.utils.quality.filters import check_data_range, out_of_bounds_filter
@pytest.fixture
def sample_dataframe():
"""Fixture providing a sample DataFrame for testing."""
return pd.DataFrame({
'tag': ['temp', 'temp', 'pressure', 'pressure', 'humidity', 'wind_speed'],
'name': ['temp', 'temp', 'pressure', 'pressure', 'humidity', 'wind_speed'],
'value': [25, 35, 95, 105, 60, None],
'timestamp': pd.date_range(start='2023-01-01', periods=6)
})
return pd.DataFrame(
{
'tag': ['temp', 'temp', 'pressure', 'pressure', 'humidity', 'wind_speed'],
'name': ['temp', 'temp', 'pressure', 'pressure', 'humidity', 'wind_speed'],
'value': [25, 35, 95, 105, 60, None],
'timestamp': pd.date_range(start='2023-01-01', periods=6),
}
)
@pytest.fixture
@@ -25,7 +28,7 @@ def nodes_data_range():
'temp': {'data_range': [10, 30]},
'pressure': {'data_range': [90, 100]},
'humidity': {'data_range': [40, 80]},
'wind_speed': {'data_range': [0, 50]}
'wind_speed': {'data_range': [0, 50]},
}
@@ -59,6 +62,7 @@ def test_check_data_range(value, val_range, expected):
else:
assert result == expected
# Tests for out_of_bounds_filter
@@ -72,8 +76,8 @@ def test_out_of_bounds_filter(sample_dataframe, nodes_data_range):
'timestamp': [
pd.Timestamp('2023-01-02'),
pd.Timestamp('2023-01-04'),
pd.Timestamp('2023-01-06')
]
pd.Timestamp('2023-01-06'),
],
}
expected_df = pd.DataFrame(expected_data)
@@ -91,6 +95,7 @@ def test_out_of_bounds_filter_empty_df(nodes_data_range):
assert result.empty
assert list(result.columns) == ['tag', 'name', 'value', 'timestamp']
# Tests for null_values_filter
@@ -100,7 +105,7 @@ def test_null_values_filter(sample_dataframe, nodes_data_range):
'tag': ['wind_speed'],
'name': ['wind_speed'],
'value': [None],
'timestamp': [pd.Timestamp('2023-01-06')]
'timestamp': [pd.Timestamp('2023-01-06')],
}
expected_df = pd.DataFrame(expected_data)
@@ -113,12 +118,14 @@ def test_null_values_filter(sample_dataframe, nodes_data_range):
def test_null_values_filter_no_nulls(nodes_data_range):
"""Test with a DataFrame containing no null values."""
df = pd.DataFrame({
'tag': ['temp', 'pressure'],
'name': ['temp', 'pressure'],
'value': [25, 100],
'timestamp': pd.date_range(start='2023-01-01', periods=2)
})
df = pd.DataFrame(
{
'tag': ['temp', 'pressure'],
'name': ['temp', 'pressure'],
'value': [25, 100],
'timestamp': pd.date_range(start='2023-01-01', periods=2),
}
)
result = null_values_filter(df, nodes_data_range)
assert result.empty
assert list(result.columns) == ['tag', 'name', 'value', 'timestamp']