temp/features/steps/auto_debug_coverage_boost_steps.py

"""Step definitions for auto_debug_coverage_boost.feature.

These steps target specific uncovered lines in
src/cleveragents/agents/graphs/auto_debug.py:

- Line 124: _analyze_error when content == "Mock LLM response" (true branch)
- Lines 128-130: _analyze_error exception handler
- Lines 200-203: _generate_fix when content == "Mock LLM response" (true branch)
- Lines 209-213: _generate_fix JSONDecodeError fallback
- Lines 215-220: _generate_fix exception handler
- Line 266: _validate_fix when content == "Mock LLM response"
- Lines 269-270: _validate_fix JSON parse with is_valid
- Lines 271-276: _validate_fix JSONDecodeError keyword scan
- Lines 277-279: _validate_fix exception handler
- Lines 283-286: _validate_fix invalid → append to attempted_fixes
- Lines 290-295: _should_retry_fix routing logic
- Lines 297-305: _finalize result construction
"""

import json
from typing import Any

from behave import given, then, when

from cleveragents.agents.graphs.auto_debug import AutoDebugAgent, AutoDebugState


# ---------------------------------------------------------------------------
# Helpers: mock LLM classes
# ---------------------------------------------------------------------------
class _MockResponse:
    """Minimal response object with a content attribute."""

    def __init__(self, content: str) -> None:
        self.content = content


class _MockLLM:
    """Stub LLM that returns a fixed content string."""

    def __init__(self, content: str = "Mock LLM response") -> None:
        self._content = content

    def invoke(self, messages: Any) -> _MockResponse:
        return _MockResponse(self._content)


class _FailingLLM:
    """Stub LLM whose invoke always raises an exception."""

    def invoke(self, messages: Any) -> None:
        raise RuntimeError("LLM service unavailable for testing")


def _base_state(**overrides: Any) -> AutoDebugState:
    """Create a minimal valid AutoDebugState with optional overrides."""
    state: AutoDebugState = {
        "messages": [],
        "context": {},
        "result": None,
        "error": None,
        "metadata": {},
        "error_message": "NameError: name 'x' is not defined",
        "code_context": "print(x)",
        "attempted_fixes": [],
        "current_fix": {},
        "fix_validated": False,
    }
    state.update(overrides)  # type: ignore[typeddict-item]
    return state


# ---------------------------------------------------------------------------
# Background
# ---------------------------------------------------------------------------
@given("the auto debug graph module is imported")
def step_auto_debug_module_imported(context):
    """Verify the auto_debug module is importable."""
    assert AutoDebugAgent is not None
    assert AutoDebugState is not None


# ---------------------------------------------------------------------------
# Agent construction helpers
# ---------------------------------------------------------------------------
@given('an auto debug agent with a mock LLM returning "{content}"')
def step_create_agent_mock_llm(context, content):
    """Create an AutoDebugAgent with a mock LLM returning the given content."""
    context.ad_agent = AutoDebugAgent(llm=_MockLLM(content), max_fix_attempts=3)


@given(
    'an auto debug agent with a mock LLM returning "{content}"'
    " and max {n:d} fix attempts"
)
def step_create_agent_mock_llm_max_attempts(context, content, n):
    """Create an AutoDebugAgent with a mock LLM and custom max_fix_attempts."""
    context.ad_agent = AutoDebugAgent(llm=_MockLLM(content), max_fix_attempts=n)


@given("an auto debug agent with a failing LLM")
def step_create_agent_failing_llm(context):
    """Create an AutoDebugAgent with an LLM that always raises."""
    context.ad_agent = AutoDebugAgent(llm=_FailingLLM(), max_fix_attempts=3)


@given("an auto debug agent with a mock LLM returning valid fix JSON")
def step_create_agent_valid_fix_json(context):
    """Create an agent whose LLM returns well-formed fix JSON."""
    fix_json = json.dumps(
        {
            "description": "Parsed JSON fix",
            "code": "print('fixed')",
            "files_to_modify": ["main.py"],
        }
    )
    context.ad_agent = AutoDebugAgent(llm=_MockLLM(fix_json), max_fix_attempts=3)


@given("an auto debug agent with a mock LLM returning valid validation JSON true")
def step_create_agent_valid_validation_json_true(context):
    """Create an agent whose LLM returns validation JSON with is_valid=true."""
    validation_json = json.dumps(
        {
            "is_valid": True,
            "reasoning": "Fix resolves the error",
            "issues": [],
        }
    )
    context.ad_agent = AutoDebugAgent(llm=_MockLLM(validation_json), max_fix_attempts=3)


@given("an auto debug agent with a mock LLM returning valid validation JSON false")
def step_create_agent_valid_validation_json_false(context):
    """Create an agent whose LLM returns validation JSON with is_valid=false."""
    validation_json = json.dumps(
        {
            "is_valid": False,
            "reasoning": "Fix does not resolve the error",
            "issues": ["error persists"],
        }
    )
    context.ad_agent = AutoDebugAgent(llm=_MockLLM(validation_json), max_fix_attempts=3)


# ---------------------------------------------------------------------------
# State preparation helpers
# ---------------------------------------------------------------------------
@given("a state with an existing error analysis message")
def step_state_with_analysis(context):
    """Prepare a state that already has an error_analysis message."""
    context.ad_state = _base_state(
        messages=[
            {
                "role": "assistant",
                "content": "Detected a NameError due to undefined variable",
                "type": "error_analysis",
            }
        ]
    )


@given("a state with two previous attempted fixes")
def step_state_with_previous_attempts(context):
    """Prepare a state with two prior attempted fixes."""
    context.ad_state = _base_state(
        messages=[
            {
                "role": "assistant",
                "content": "Analysis result",
                "type": "error_analysis",
            }
        ],
        attempted_fixes=[
            {"description": "First attempt", "code": "# v1", "files_to_modify": []},
            {"description": "Second attempt", "code": "# v2", "files_to_modify": []},
        ],
    )


@given("a state with a current fix to validate")
def step_state_with_current_fix(context):
    """Prepare a state that has a current_fix ready for validation."""
    context.ad_state = _base_state(
        current_fix={
            "description": "Proposed fix",
            "code": "x = 42\nprint(x)",
            "files_to_modify": ["main.py"],
        }
    )


@given("a finalize state with fix_validated {validated} and {n:d} attempted fixes")
def step_finalize_state(context, validated, n):
    """Prepare a state for _finalize with the given validation status."""
    is_valid = validated.lower() == "true"
    fixes = [
        {"description": f"Attempt {i + 1}", "code": f"# v{i + 1}"} for i in range(n)
    ]
    context.ad_state = _base_state(
        fix_validated=is_valid,
        attempted_fixes=fixes,
        current_fix={"description": "Latest fix", "code": "# latest"},
    )


# ---------------------------------------------------------------------------
# When: call individual node methods
# ---------------------------------------------------------------------------
@when("I call _analyze_error with an error state")
def step_call_analyze_error(context):
    """Invoke _analyze_error directly on the agent."""
    state = _base_state()
    context.ad_result_state = context.ad_agent._analyze_error(state)


@when("I call _generate_fix with the prepared state")
def step_call_generate_fix(context):
    """Invoke _generate_fix directly on the agent."""
    context.ad_result_state = context.ad_agent._generate_fix(context.ad_state)


@when("I call _validate_fix with the prepared state")
def step_call_validate_fix(context):
    """Invoke _validate_fix directly on the agent."""
    context.ad_result_state = context.ad_agent._validate_fix(context.ad_state)


@when("I call _finalize with the prepared state")
def step_call_finalize(context):
    """Invoke _finalize directly on the agent."""
    context.ad_result_state = context.ad_agent._finalize(context.ad_state)


@when("I check should_retry_fix with validated {validated} and {n:d} attempt")
def step_check_should_retry_singular(context, validated, n):
    """Invoke _should_retry_fix with a crafted state (singular form)."""
    is_valid = validated.lower() == "true"
    fixes = [{"description": f"fix {i + 1}"} for i in range(n)]
    state = _base_state(fix_validated=is_valid, attempted_fixes=fixes)
    context.routing_decision = context.ad_agent._should_retry_fix(state)


@when("I check should_retry_fix with validated {validated} and {n:d} attempts")
def step_check_should_retry_plural(context, validated, n):
    """Invoke _should_retry_fix with a crafted state (plural form)."""
    is_valid = validated.lower() == "true"
    fixes = [{"description": f"fix {i + 1}"} for i in range(n)]
    state = _base_state(fix_validated=is_valid, attempted_fixes=fixes)
    context.routing_decision = context.ad_agent._should_retry_fix(state)


@when("I try to create an AutoDebugAgent with no LLM")
def step_create_agent_no_llm(context):
    """Attempt to create an AutoDebugAgent with llm=None."""
    try:
        AutoDebugAgent(llm=None)
        context.constructor_error = None
    except ValueError as exc:
        context.constructor_error = exc


# ---------------------------------------------------------------------------
# Then: assertions
# ---------------------------------------------------------------------------
@then('the analysis message should be "{expected}"')
def step_verify_analysis_message(context, expected):
    """Verify the last error_analysis message content."""
    messages = context.ad_result_state.get("messages", [])
    analysis_msgs = [m for m in messages if m.get("type") == "error_analysis"]
    assert analysis_msgs, "No error_analysis message found in state"
    actual = analysis_msgs[-1]["content"]
    assert actual == expected, f"Expected '{expected}', got '{actual}'"


@then('the current fix description should be "{expected}"')
def step_verify_fix_description(context, expected):
    """Verify the current_fix description."""
    fix = context.ad_result_state.get("current_fix", {})
    actual = fix.get("description", "")
    assert actual == expected, f"Expected '{expected}', got '{actual}'"


@then('the current fix code should be "{expected}"')
def step_verify_fix_code(context, expected):
    """Verify the current_fix code."""
    fix = context.ad_result_state.get("current_fix", {})
    actual = fix.get("code", "")
    assert actual == expected, f"Expected '{expected}', got '{actual}'"


@then("the fix should be marked as validated")
def step_fix_validated_true(context):
    """Verify fix_validated is True."""
    assert context.ad_result_state["fix_validated"] is True


@then("the fix should not be marked as validated")
def step_fix_validated_false(context):
    """Verify fix_validated is False."""
    assert context.ad_result_state["fix_validated"] is False


@then("the current fix should be appended to attempted fixes")
def step_fix_appended_to_attempted(context):
    """Verify the current fix was moved to attempted_fixes."""
    attempted = context.ad_result_state.get("attempted_fixes", [])
    assert len(attempted) > 0, "Expected at least one attempted fix"
    last = attempted[-1]
    assert last.get("description") == "Proposed fix", (
        f"Expected 'Proposed fix', got '{last.get('description')}'"
    )


@then('the routing decision should be "{expected}"')
def step_verify_routing(context, expected):
    """Verify the _should_retry_fix routing result."""
    assert context.routing_decision == expected, (
        f"Expected '{expected}', got '{context.routing_decision}'"
    )


@then("the result should indicate success {expected}")
def step_verify_result_success(context, expected):
    """Verify the result dict success field."""
    is_success = expected.lower() == "true"
    result = context.ad_result_state.get("result")
    assert result is not None, "No result dict found in state"
    assert result["success"] is is_success, (
        f"Expected success={is_success}, got {result['success']}"
    )


@then("the result attempts count should be {n:d}")
def step_verify_result_attempts(context, n):
    """Verify the result dict attempts count."""
    result = context.ad_result_state.get("result")
    assert result is not None, "No result dict found in state"
    assert result["attempts"] == n, f"Expected {n} attempts, got {result['attempts']}"


@then("a ValueError should be raised with the missing provider message")
def step_verify_constructor_valueerror(context):
    """Verify the constructor raised a ValueError about missing LLM."""
    assert context.constructor_error is not None, "Expected ValueError but none raised"
    assert isinstance(context.constructor_error, ValueError)
    assert "No LLM provider configured" in str(context.constructor_error)