vibe_data_ana/tests/test_analysis_planning_properties.py

"""Property-based tests for analysis planning engine."""

import pytest
from hypothesis import given, strategies as st, settings

from src.engines.analysis_planning import (
    plan_analysis,
    validate_task_dependencies,
    _fallback_analysis_planning,
    _has_circular_dependency
)
from src.models.data_profile import DataProfile, ColumnInfo
from src.models.requirement_spec import RequirementSpec, AnalysisObjective
from src.models.analysis_plan import AnalysisTask


# Strategies for generating test data
@st.composite
def column_info_strategy(draw):
    """Generate random ColumnInfo."""
    name = draw(st.text(min_size=1, max_size=20, alphabet=st.characters(whitelist_categories=('L', 'N'))))
    dtype = draw(st.sampled_from(['numeric', 'categorical', 'datetime', 'text']))
    missing_rate = draw(st.floats(min_value=0.0, max_value=1.0))
    unique_count = draw(st.integers(min_value=1, max_value=1000))
    
    return ColumnInfo(
        name=name,
        dtype=dtype,
        missing_rate=missing_rate,
        unique_count=unique_count,
        sample_values=[],
        statistics={}
    )


@st.composite
def data_profile_strategy(draw):
    """Generate random DataProfile."""
    row_count = draw(st.integers(min_value=10, max_value=100000))
    columns = draw(st.lists(column_info_strategy(), min_size=2, max_size=20))
    inferred_type = draw(st.sampled_from(['ticket', 'sales', 'user', 'unknown']))
    quality_score = draw(st.floats(min_value=0.0, max_value=100.0))
    
    return DataProfile(
        file_path='test.csv',
        row_count=row_count,
        column_count=len(columns),
        columns=columns,
        inferred_type=inferred_type,
        key_fields={},
        quality_score=quality_score,
        summary=f"Test data with {len(columns)} columns"
    )


@st.composite
def requirement_spec_strategy(draw):
    """Generate random RequirementSpec."""
    user_input = draw(st.text(min_size=5, max_size=100))
    num_objectives = draw(st.integers(min_value=1, max_value=5))
    
    objectives = []
    for i in range(num_objectives):
        obj = AnalysisObjective(
            name=f"Objective {i+1}",
            description=draw(st.text(min_size=10, max_size=100)),
            metrics=draw(st.lists(st.text(min_size=3, max_size=20), min_size=1, max_size=5)),
            priority=draw(st.integers(min_value=1, max_value=5))
        )
        objectives.append(obj)
    
    return RequirementSpec(
        user_input=user_input,
        objectives=objectives
    )


# Feature: true-ai-agent, Property 6: 动态任务生成
@given(
    data_profile=data_profile_strategy(),
    requirement=requirement_spec_strategy()
)
@settings(max_examples=20, deadline=None)
def test_dynamic_task_generation(data_profile, requirement):
    """
    Property 6: For any data profile and requirement spec, the analysis
    planning engine should be able to generate a non-empty task list, with
    each task containing unique ID, description, priority, and required tools.
    
    Validates: 场景1验收.2, FR-3.1
    """
    # Use fallback to avoid API dependency
    plan = _fallback_analysis_planning(data_profile, requirement)
    
    # Verify: Should have tasks
    assert len(plan.tasks) > 0, "Should generate at least one task"
    
    # Verify: Each task should have required fields
    task_ids = set()
    for task in plan.tasks:
        # Unique ID
        assert task.id not in task_ids, f"Task ID {task.id} is not unique"
        task_ids.add(task.id)
        
        # Required fields
        assert len(task.name) > 0, "Task name should not be empty"
        assert len(task.description) > 0, "Task description should not be empty"
        assert 1 <= task.priority <= 5, f"Task priority {task.priority} should be between 1 and 5"
        assert isinstance(task.required_tools, list), "Required tools should be a list"
        assert isinstance(task.dependencies, list), "Dependencies should be a list"
        assert task.status in ['pending', 'running', 'completed', 'failed', 'skipped'], \
            f"Invalid task status: {task.status}"
    
    # Verify: Plan should have objectives
    assert len(plan.objectives) > 0, "Plan should have objectives"
    
    # Verify: Estimated duration should be non-negative
    assert plan.estimated_duration >= 0, "Estimated duration should be non-negative"


# Feature: true-ai-agent, Property 7: 任务依赖一致性
@given(
    data_profile=data_profile_strategy(),
    requirement=requirement_spec_strategy()
)
@settings(max_examples=20, deadline=None)
def test_task_dependency_consistency(data_profile, requirement):
    """
    Property 7: For any generated analysis plan, all task dependencies should
    form a directed acyclic graph (DAG), with no circular dependencies.
    
    Validates: FR-3.1
    """
    # Use fallback to avoid API dependency
    plan = _fallback_analysis_planning(data_profile, requirement)
    
    # Verify: No circular dependencies
    assert not _has_circular_dependency(plan.tasks), \
        "Task dependencies should not form a cycle"
    
    # Verify: All dependencies exist
    task_ids = {task.id for task in plan.tasks}
    for task in plan.tasks:
        for dep_id in task.dependencies:
            assert dep_id in task_ids, \
                f"Task {task.id} depends on non-existent task {dep_id}"
            assert dep_id != task.id, \
                f"Task {task.id} should not depend on itself"
    
    # Verify: Validation function agrees
    validation = validate_task_dependencies(plan.tasks)
    assert validation['valid'], "Task dependencies should be valid"
    assert validation['forms_dag'], "Task dependencies should form a DAG"
    assert not validation['has_circular_dependency'], "Should not have circular dependencies"
    assert len(validation['missing_dependencies']) == 0, "Should not have missing dependencies"


# Feature: true-ai-agent, Property 6: 动态任务生成 (priority ordering)
@given(
    data_profile=data_profile_strategy(),
    requirement=requirement_spec_strategy()
)
@settings(max_examples=20, deadline=None)
def test_task_priority_ordering(data_profile, requirement):
    """
    Property 6 (extended): Tasks should respect objective priorities.
    High-priority objectives should generate high-priority tasks.
    
    Validates: FR-3.2
    """
    # Use fallback to avoid API dependency
    plan = _fallback_analysis_planning(data_profile, requirement)
    
    # Verify: All tasks have valid priorities
    for task in plan.tasks:
        assert 1 <= task.priority <= 5, \
            f"Task priority {task.priority} should be between 1 and 5"
    
    # Verify: If objectives have high priority, at least some tasks should too
    max_obj_priority = max(obj.priority for obj in plan.objectives)
    if max_obj_priority >= 4:
        # Should have at least one high-priority task
        high_priority_tasks = [t for t in plan.tasks if t.priority >= 4]
        # This is a soft requirement, so we just check structure
        assert all(1 <= t.priority <= 5 for t in plan.tasks)


# Test circular dependency detection
@given(
    num_tasks=st.integers(min_value=2, max_value=10)
)
@settings(max_examples=10, deadline=None)
def test_circular_dependency_detection(num_tasks):
    """
    Test that circular dependency detection works correctly.
    """
    # Create tasks with no dependencies (should be valid)
    tasks = [
        AnalysisTask(
            id=f"task_{i}",
            name=f"Task {i}",
            description=f"Description {i}",
            priority=3,
            dependencies=[]
        )
        for i in range(num_tasks)
    ]
    
    # Should not have circular dependencies
    assert not _has_circular_dependency(tasks)
    
    # Create a simple cycle: task_0 -> task_1 -> task_0
    if num_tasks >= 2:
        tasks_with_cycle = [
            AnalysisTask(
                id="task_0",
                name="Task 0",
                description="Description 0",
                priority=3,
                dependencies=["task_1"]
            ),
            AnalysisTask(
                id="task_1",
                name="Task 1",
                description="Description 1",
                priority=3,
                dependencies=["task_0"]
            )
        ]
        
        # Should detect the cycle
        assert _has_circular_dependency(tasks_with_cycle)


# Test dependency validation
def test_dependency_validation_with_missing_deps():
    """Test validation detects missing dependencies."""
    tasks = [
        AnalysisTask(
            id="task_1",
            name="Task 1",
            description="Description 1",
            priority=3,
            dependencies=["task_2", "task_999"]  # task_999 doesn't exist
        ),
        AnalysisTask(
            id="task_2",
            name="Task 2",
            description="Description 2",
            priority=3,
            dependencies=[]
        )
    ]
    
    validation = validate_task_dependencies(tasks)
    
    # Should not be valid
    assert not validation['valid']
    
    # Should have missing dependencies
    assert len(validation['missing_dependencies']) > 0
    
    # Should identify task_999 as missing
    missing_dep_ids = [md['missing_dep'] for md in validation['missing_dependencies']]
    assert 'task_999' in missing_dep_ids