placeholder/features/steps/context_analysis_graph_coverage_steps.py

"""Step definitions for context analysis graph coverage scenarios.

These steps specifically target uncovered lines in
src/cleveragents/agents/graphs/context_analysis.py as identified
by the build/coverage.xml report.
"""

from __future__ import annotations

import asyncio
import shutil
import tempfile
from pathlib import Path
from typing import Any

from behave import given, then, when
from langchain_community.llms import FakeListLLM
from langchain_core.documents import Document

from cleveragents.agents.graphs.context_analysis import (
    ContextAnalysisAgent,
    ContextAnalysisState,
)


def _default_context_test_llm() -> FakeListLLM:
    """Create a FakeListLLM for context analysis test purposes."""
    return FakeListLLM(
        responses=[
            "Dependencies: ['os', 'sys', 'pathlib']",
            "Relevance: High - contains core functionality",
            "Summary: Python module implementing core business logic",
        ]
    )


# Enough responses so every LLM call across the full workflow is satisfied.
_DEFAULT_RESPONSES = [
    "Dependencies: ['os', 'sys', 'pathlib']",
    "Relevance: 0.8 - core module",
    "Summary: Generated context overview",
] * 30


def _make_state(**overrides: Any) -> ContextAnalysisState:
    """Build a minimal valid ContextAnalysisState, with optional overrides."""
    state: ContextAnalysisState = {
        "file_paths": [],
        "documents": [],
        "dependencies": {},
        "summary": "",
        "relevance_scores": {},
        "chunks": [],
        "error": None,
    }
    state.update(overrides)  # type: ignore[typeddict-item]
    return state


def _ensure_temp_dir(context: Any) -> Path:
    if not hasattr(context, "graph_temp_dir") or context.graph_temp_dir is None:
        context.graph_temp_dir = Path(tempfile.mkdtemp(prefix="ctx-graph-cov-"))
    return context.graph_temp_dir


# ---------------------------------------------------------------------------
# Background steps
# ---------------------------------------------------------------------------


@given("the context analysis graph module is importable")
def step_module_importable(context: Any) -> None:
    ContextAnalysisAgent  # noqa: B018
    ContextAnalysisState  # noqa: B018


@given("I have a fake LLM configured for context analysis graph tests")
def step_configure_fake_llm(context: Any) -> None:
    context.graph_fake_llm = FakeListLLM(responses=list(_DEFAULT_RESPONSES))


# ---------------------------------------------------------------------------
# Scenario: Agent initialises with an explicitly provided LLM
# Targets lines 115, 117, 125 (the else branch of if llm is None)
# ---------------------------------------------------------------------------


@when("I create a ContextAnalysisAgent with a custom LLM")
def step_create_agent_custom_llm(context: Any) -> None:
    custom_llm = FakeListLLM(responses=list(_DEFAULT_RESPONSES))
    context.custom_llm_ref = custom_llm
    context.graph_agent = ContextAnalysisAgent(llm=custom_llm)


@then("the agent should use the provided LLM instance")
def step_agent_uses_provided_llm(context: Any) -> None:
    assert context.graph_agent.llm is context.custom_llm_ref


@then("the agent should be initialised successfully")
def step_agent_initialised(context: Any) -> None:
    assert isinstance(context.graph_agent, ContextAnalysisAgent)


# ---------------------------------------------------------------------------
# Helper: ensure an agent on the context
# ---------------------------------------------------------------------------


def _ensure_agent(context: Any, **kwargs: Any) -> ContextAnalysisAgent:
    if hasattr(context, "graph_agent") and context.graph_agent is not None:
        return context.graph_agent
    llm = kwargs.pop("llm", getattr(context, "graph_fake_llm", None))
    if llm is None:
        llm = FakeListLLM(responses=list(_DEFAULT_RESPONSES))
    context.graph_agent = ContextAnalysisAgent(llm=llm, **kwargs)
    return context.graph_agent


@given("I have a ContextAnalysisAgent for graph coverage")
def step_have_agent(context: Any) -> None:
    context.graph_agent = None  # force fresh agent
    _ensure_agent(context)


@given(
    "I have a ContextAnalysisAgent for graph coverage "
    "with chunk_size {chunk_size:d} and chunk_overlap {overlap:d}"
)
def step_have_agent_custom_chunks(context: Any, chunk_size: int, overlap: int) -> None:
    context.graph_agent = None
    _ensure_agent(context, chunk_size=chunk_size, chunk_overlap=overlap)


# ---------------------------------------------------------------------------
# Scenario: Load files discovers missing file paths
# Targets lines 228-236, 248-252
# ---------------------------------------------------------------------------


@when("I call load_files with a nonexistent file path")
def step_load_files_missing(context: Any) -> None:
    state = _make_state(file_paths=["/tmp/does_not_exist_xyz.py"])
    context.load_result = context.graph_agent._load_files(state)


@then("the returned documents list should be empty")
def step_returned_docs_empty(context: Any) -> None:
    assert context.load_result["documents"] == []


@then('the returned error should contain "{text}"')
def step_returned_error_contains(context: Any, text: str) -> None:
    error = context.load_result.get("error")
    assert error is not None and text in error, (
        f"Expected '{text}' in error, got: {error!r}"
    )


# ---------------------------------------------------------------------------
# Scenario: Load files rejects directory paths
# Targets lines 238-240
# ---------------------------------------------------------------------------


@given('I have a temporary directory called "{dirname}"')
def step_create_temp_directory(context: Any, dirname: str) -> None:
    temp_dir = _ensure_temp_dir(context)
    dir_path = temp_dir / dirname
    dir_path.mkdir(parents=True, exist_ok=True)
    context.graph_dir_path = dir_path


@when("I call load_files with the directory path")
def step_load_files_with_dir(context: Any) -> None:
    state = _make_state(file_paths=[str(context.graph_dir_path)])
    context.load_result = context.graph_agent._load_files(state)


# ---------------------------------------------------------------------------
# Scenario: Load files reads a real file from disk
# Targets lines 242-244 (TextLoader path)
# ---------------------------------------------------------------------------


@given('I have a temporary file "{filename}" containing "{content}"')
def step_create_temp_file(context: Any, filename: str, content: str) -> None:
    temp_dir = _ensure_temp_dir(context)
    file_path = temp_dir / filename
    file_path.parent.mkdir(parents=True, exist_ok=True)
    file_path.write_text(content, encoding="utf-8")
    context.graph_file_path = file_path


@when("I call load_files with that file path")
def step_load_files_real_file(context: Any) -> None:
    state = _make_state(file_paths=[str(context.graph_file_path)])
    context.load_result = context.graph_agent._load_files(state)


@then("the returned documents list should have {count:d} entry")
def step_returned_docs_count(context: Any, count: int) -> None:
    assert len(context.load_result["documents"]) == count


@then("the returned error should be empty")
def step_returned_error_empty(context: Any) -> None:
    assert context.load_result.get("error") is None


# ---------------------------------------------------------------------------
# Scenario: Load files combines missing and directory errors
# Targets lines 231-250 comprehensively
# ---------------------------------------------------------------------------


@when("I call load_files with both a missing file and the directory path")
def step_load_files_mixed_errors(context: Any) -> None:
    state = _make_state(
        file_paths=[
            "/tmp/definitely_missing_abc.py",
            str(context.graph_dir_path),
        ]
    )
    context.load_result = context.graph_agent._load_files(state)


# ---------------------------------------------------------------------------
# Scenario: Chunking splits documents exceeding chunk size
# Targets lines 317-326 (the step/chunking loop)
# ---------------------------------------------------------------------------


@given("I have a state with a single document of {size:d} characters")
def step_state_with_large_doc(context: Any, size: int) -> None:
    content = "a" * size
    doc = Document(page_content=content, metadata={"source": "large.py"})
    context.graph_state = _make_state(documents=[doc])


@when("I call chunk_documents on the state")
def step_call_chunk_documents(context: Any) -> None:
    result = context.graph_agent._chunk_documents(context.graph_state)
    context.graph_state.update(result)


@then("the resulting chunks list should have at least {count:d} entries")
def step_chunks_at_least(context: Any, count: int) -> None:
    assert len(context.graph_state["chunks"]) >= count


@then("every chunk should carry a chunk_index in its metadata")
def step_chunks_have_metadata(context: Any) -> None:
    for chunk in context.graph_state["chunks"]:
        assert "chunk_index" in chunk.metadata


# ---------------------------------------------------------------------------
# Scenario: Relevance scoring skips duplicate file chunks
# Targets line 345 (if file_path in files_seen: continue)
# ---------------------------------------------------------------------------


@given('I have a state with two chunks from the same source file "{source}"')
def step_state_dup_chunks(context: Any, source: str) -> None:
    chunks = [
        Document(
            page_content="chunk 0 content",
            metadata={"source": source, "chunk_index": 0},
        ),
        Document(
            page_content="chunk 1 content",
            metadata={"source": source, "chunk_index": 1},
        ),
    ]
    context.graph_state = _make_state(chunks=chunks)


@when("I call score_relevance on the state")
def step_call_score_relevance(context: Any) -> None:
    result = context.graph_agent._score_relevance(context.graph_state)
    context.graph_state.update(result)


@then("the relevance scores dictionary should contain exactly {count:d} entry")
def step_scores_count(context: Any, count: int) -> None:
    assert len(context.graph_state["relevance_scores"]) == count


# ---------------------------------------------------------------------------
# Scenario: Relevance parser returns 0.3 for low keyword
# Targets line 388 (if "low" in llm_output.lower(): return 0.3)
# ---------------------------------------------------------------------------


@when('I parse relevance from the text "{text}"')
def step_parse_relevance(context: Any, text: str) -> None:
    context.parsed_relevance = context.graph_agent._parse_relevance_score(text)


@then("the parsed relevance score should be {expected:f}")
def step_parsed_relevance_value(context: Any, expected: float) -> None:
    assert abs(context.parsed_relevance - expected) < 1e-9, (
        f"Expected {expected}, got {context.parsed_relevance}"
    )


# ---------------------------------------------------------------------------
# Scenario: Async invocation completes the full workflow
# Targets lines 444-446 (ainvoke method)
# ---------------------------------------------------------------------------


@when("I run the workflow asynchronously via ainvoke")
def step_ainvoke_workflow(context: Any) -> None:
    state = _make_state(file_paths=[str(context.graph_file_path)])
    config = {"configurable": {"thread_id": "graph-cov-async"}}

    async def _run() -> ContextAnalysisState:
        return await context.graph_agent.ainvoke(state, config=config)

    context.async_result = asyncio.run(_run())


@then("the async result should contain a summary")
def step_async_result_summary(context: Any) -> None:
    assert "summary" in context.async_result
    assert isinstance(context.async_result["summary"], str)
    assert context.async_result["summary"] != ""


@then("the async result should contain documents")
def step_async_result_documents(context: Any) -> None:
    assert "documents" in context.async_result


# ---------------------------------------------------------------------------
# Scenario: Sync streaming produces node-level events
# Targets lines 452-458 (stream method)
# ---------------------------------------------------------------------------


@when("I stream the workflow synchronously")
def step_sync_stream(context: Any) -> None:
    state = _make_state(file_paths=[str(context.graph_file_path)])
    config = {"configurable": {"thread_id": "graph-cov-stream"}}
    context.stream_events = list(context.graph_agent.stream(state, config=config))


@then("I should receive at least {count:d} stream event")
def step_stream_event_count(context: Any, count: int) -> None:
    assert len(context.stream_events) >= count


@then("each stream event should be a dictionary with a node key")
def step_stream_events_are_dicts(context: Any) -> None:
    expected_nodes = {
        "load_files",
        "analyze_dependencies",
        "chunk_documents",
        "score_relevance",
        "summarize_context",
    }
    for event in context.stream_events:
        assert isinstance(event, dict)
        assert any(key in expected_nodes for key in event)


# ---------------------------------------------------------------------------
# Scenario: Async streaming produces node-level events
# Targets lines 464-471 (astream method + async for yield)
# ---------------------------------------------------------------------------


@when("I stream the workflow asynchronously")
def step_async_stream(context: Any) -> None:
    state = _make_state(file_paths=[str(context.graph_file_path)])
    config = {"configurable": {"thread_id": "graph-cov-astream"}}

    async def _run() -> list[dict[str, Any]]:
        events: list[dict[str, Any]] = []
        async for event in context.graph_agent.astream(state, config=config):
            events.append(event)
        return events

    context.astream_events = asyncio.run(_run())


@then("I should receive at least {count:d} async stream event")
def step_astream_event_count(context: Any, count: int) -> None:
    assert len(context.astream_events) >= count


@then("each async stream event should be a dictionary with a node key")
def step_astream_events_are_dicts(context: Any) -> None:
    expected_nodes = {
        "load_files",
        "analyze_dependencies",
        "chunk_documents",
        "score_relevance",
        "summarize_context",
    }
    for event in context.astream_events:
        assert isinstance(event, dict)
        assert any(key in expected_nodes for key in event)


# ---------------------------------------------------------------------------
# Scenario: Agent initialises with default FakeListLLM when no LLM provided
# Targets lines 114-117 (if llm is None: FakeListLLM)
# ---------------------------------------------------------------------------


@when("I create a ContextAnalysisAgent without providing an LLM")
def step_create_agent_no_llm(context: Any) -> None:
    context.graph_agent = ContextAnalysisAgent(llm=_default_context_test_llm())


@then("the agent LLM should be a FakeListLLM")
def step_agent_llm_is_fake(context: Any) -> None:
    assert type(context.graph_agent.llm).__name__ == "FakeListLLM"


# ---------------------------------------------------------------------------
# Scenario: Sync invoke runs the complete workflow end to end
# Targets lines 436-438 (invoke method)
# ---------------------------------------------------------------------------


@when("I invoke the workflow synchronously")
def step_sync_invoke(context: Any) -> None:
    state = _make_state(file_paths=[str(context.graph_file_path)])
    config = {"configurable": {"thread_id": "graph-cov-invoke"}}
    context.sync_result = context.graph_agent.invoke(state, config=config)


@then("the sync result should contain a summary")
def step_sync_result_summary(context: Any) -> None:
    assert "summary" in context.sync_result
    assert isinstance(context.sync_result["summary"], str)
    assert context.sync_result["summary"] != ""


@then("the sync result should contain documents")
def step_sync_result_documents(context: Any) -> None:
    assert "documents" in context.sync_result


# ---------------------------------------------------------------------------
# Scenario: Load files returns preloaded documents unchanged
# Targets lines 221-226 (if preloaded_docs: return early)
# ---------------------------------------------------------------------------


@given("I have a state with preloaded documents")
def step_state_with_preloaded_docs(context: Any) -> None:
    context.preloaded_docs = [
        Document(page_content="preloaded content", metadata={"source": "pre.py"}),
    ]
    context.preloaded_state = _make_state(documents=context.preloaded_docs)


@when("I call load_files on the preloaded state")
def step_load_files_preloaded(context: Any) -> None:
    context.load_result = context.graph_agent._load_files(context.preloaded_state)


@then("the returned documents should equal the preloaded documents")
def step_returned_docs_equal_preloaded(context: Any) -> None:
    assert context.load_result["documents"] == context.preloaded_docs


# ---------------------------------------------------------------------------
# Cleanup
# ---------------------------------------------------------------------------


def after_scenario(context: Any, _scenario: Any) -> None:
    """Clean up temporary files created during graph coverage scenarios."""
    temp_dir = getattr(context, "graph_temp_dir", None)
    if temp_dir is not None and Path(temp_dir).exists():
        shutil.rmtree(temp_dir, ignore_errors=True)
    for attr in (
        "graph_agent",
        "graph_temp_dir",
        "graph_file_path",
        "graph_dir_path",
        "graph_state",
        "load_result",
        "custom_llm_ref",
        "parsed_relevance",
        "async_result",
        "stream_events",
        "astream_events",
    ):
        if hasattr(context, attr):
            setattr(context, attr, None)


# ---------------------------------------------------------------------------
# Scenario: Agent raises ValueError when no LLM provided
# ---------------------------------------------------------------------------


@when("I create a context analysis agent with no LLM expecting an error")
def step_create_agent_no_llm_error(context: Any) -> None:
    context.raised_error = None
    try:
        from cleveragents.agents.graphs.context_analysis import (
            ContextAnalysisAgent as _CA,
        )

        _CA(llm=None)
    except ValueError as exc:
        context.raised_error = exc


@then("a ValueError should have been raised about missing LLM")
def step_check_value_error_missing_llm(context: Any) -> None:
    assert context.raised_error is not None, (
        "Expected ValueError but no error was raised"
    )
    assert "No LLM provider configured" in str(context.raised_error)