tests/test_validate.py - master - nevers/hackINTERPOL

tests/test_validate.py
104 строки3 KB
nevers
upload files
13 янв 2026, 17:14
13 янв 2026, 17:14d476ef2
"""Tests for CSV validation."""

import pandas as pd

from studypulse.io import validate_dataframe, validate_ranges, validate_schema


def test_validate_schema_missing_columns() -> None:
    """Test schema validation catches missing columns."""
    df = pd.DataFrame({"student_id": [1], "student_name": ["Alice"]})
    is_valid, errors = validate_schema(df)
    assert not is_valid
    assert len(errors) > 0
    assert "Missing required columns" in errors[0]


def test_validate_schema_all_columns() -> None:
    """Test schema validation passes with all required columns."""
    df = pd.DataFrame(
        {
            "student_id": ["STU001"],
            "student_name": ["Alice"],
            "assignment_score": [85.0],
            "assignments_submitted": [8],
            "attendance_rate": [0.9],
            "days_since_last_submission": [3],
            "days_to_deadline": [5],
        }
    )
    is_valid, errors = validate_schema(df)
    assert is_valid
    assert len(errors) == 0


def test_validate_ranges_score_out_of_range() -> None:
    """Test range validation catches out-of-range scores."""
    df = pd.DataFrame(
        {
            "student_id": ["STU001"],
            "student_name": ["Alice"],
            "assignment_score": [150.0],  # Out of range
            "assignments_submitted": [8],
            "attendance_rate": [0.9],
            "days_since_last_submission": [3],
            "days_to_deadline": [5],
        }
    )
    is_valid, errors = validate_ranges(df)
    assert not is_valid
    assert any("assignment_score" in err for err in errors)


def test_validate_ranges_attendance_out_of_range() -> None:
    """Test range validation catches out-of-range attendance."""
    df = pd.DataFrame(
        {
            "student_id": ["STU001"],
            "student_name": ["Alice"],
            "assignment_score": [85.0],
            "assignments_submitted": [8],
            "attendance_rate": [1.5],  # Out of range
            "days_since_last_submission": [3],
            "days_to_deadline": [5],
        }
    )
    is_valid, errors = validate_ranges(df)
    assert not is_valid
    assert any("attendance_rate" in err for err in errors)


def test_validate_ranges_negative_submissions() -> None:
    """Test range validation catches negative submissions."""
    df = pd.DataFrame(
        {
            "student_id": ["STU001"],
            "student_name": ["Alice"],
            "assignment_score": [85.0],
            "assignments_submitted": [-1],  # Invalid
            "attendance_rate": [0.9],
            "days_since_last_submission": [3],
            "days_to_deadline": [5],
        }
    )
    is_valid, errors = validate_ranges(df)
    assert not is_valid
    assert any("assignments_submitted" in err for err in errors)


def test_validate_dataframe_valid() -> None:
    """Test full validation passes for valid data."""
    df = pd.DataFrame(
        {
            "student_id": ["STU001", "STU002"],
            "student_name": ["Alice", "Bob"],
            "assignment_score": [85.0, 72.0],
            "assignments_submitted": [8, 7],
            "attendance_rate": [0.9, 0.85],
            "days_since_last_submission": [3, 5],
            "days_to_deadline": [5, 2],
        }
    )
    is_valid, errors = validate_dataframe(df)
    assert is_valid
    assert len(errors) == 0