dev-puppeteer/my-deepagent/tests/integration/test_instructions.py

"""v0.3 PR #7 — MYDEEPAGENT.md instruction-file hierarchy tests.

Covers:
1. Global file is bootstrapped with template on first call (idempotent).
2. Project file is NEVER auto-created — present iff user wrote it.
3. `resolve_instruction_paths` orders global → project.
4. Resolution is empty if global hasn't been bootstrapped yet.
5. `build_agent` passes the combined list through to `deepagents.create_deep_agent(memory=...)`.
"""

from __future__ import annotations

from pathlib import Path
from typing import Any

import pytest

from my_deepagent.config import load_config
from my_deepagent.instructions import (
    INSTRUCTION_FILENAME,
    ensure_global_instructions_initialized,
    global_instructions_path,
    project_instructions_path,
    resolve_instruction_paths,
)

# ---------------------------------------------------------------------------
# Bootstrap (global only)
# ---------------------------------------------------------------------------


def test_ensure_global_instructions_creates_template(tmp_path: Path) -> None:
    cfg = load_config(workspace_root=tmp_path, data_dir=tmp_path / "data")
    p = ensure_global_instructions_initialized(cfg)
    assert p.is_file()
    assert p.name == INSTRUCTION_FILENAME
    body = p.read_text(encoding="utf-8")
    assert "MYDEEPAGENT.md (global)" in body
    assert "한국어" in body  # template is Korean by default


def test_ensure_global_instructions_idempotent(tmp_path: Path) -> None:
    cfg = load_config(workspace_root=tmp_path, data_dir=tmp_path / "data")
    p = ensure_global_instructions_initialized(cfg)
    p.write_text("custom content", encoding="utf-8")
    # Second call must not overwrite user-edited content.
    p2 = ensure_global_instructions_initialized(cfg)
    assert p2 == p
    assert p.read_text(encoding="utf-8") == "custom content"


# ---------------------------------------------------------------------------
# Project file behaviour
# ---------------------------------------------------------------------------


def test_project_instructions_never_auto_created(tmp_path: Path) -> None:
    cfg = load_config(workspace_root=tmp_path, data_dir=tmp_path / "data")
    repo = tmp_path / "repo"
    repo.mkdir()
    # Bootstrap global — must not touch project file.
    ensure_global_instructions_initialized(cfg)
    assert not project_instructions_path(repo).exists()


# ---------------------------------------------------------------------------
# resolve_instruction_paths
# ---------------------------------------------------------------------------


def test_resolve_paths_includes_only_existing_files(tmp_path: Path) -> None:
    cfg = load_config(workspace_root=tmp_path, data_dir=tmp_path / "data")
    repo = tmp_path / "repo"
    repo.mkdir()

    # No files exist → empty.
    assert resolve_instruction_paths(cfg, repo) == []

    # Only global.
    g = ensure_global_instructions_initialized(cfg)
    paths = resolve_instruction_paths(cfg, repo)
    assert paths == [str(g.resolve())]

    # Add project — order becomes global, project.
    proj_file = project_instructions_path(repo)
    proj_file.write_text("# project-specific", encoding="utf-8")
    paths = resolve_instruction_paths(cfg, repo)
    assert paths == [str(g.resolve()), str(proj_file.resolve())]


def test_global_instructions_path_under_data_dir(tmp_path: Path) -> None:
    cfg = load_config(workspace_root=tmp_path, data_dir=tmp_path / "data")
    p = global_instructions_path(cfg)
    assert p.parent == cfg.data_dir
    assert p.name == INSTRUCTION_FILENAME


def test_governance_bootstrap_creates_full_skeleton(tmp_path: Path) -> None:
    """`bootstrap_user_dirs` materialises the user-wide layout (PR #7)."""
    from my_deepagent.governance import bootstrap_user_dirs
    from my_deepagent.memory import INDEX_FILENAME as MEMORY_INDEX_FILENAME

    cfg = load_config(workspace_root=tmp_path, data_dir=tmp_path / "data")
    bootstrap_user_dirs(cfg)

    # Global MYDEEPAGENT.md created with template.
    assert global_instructions_path(cfg).is_file()
    # Global memory dir + MEMORY.md created.
    global_mem = Path(cfg.data_dir) / "global" / "memory"
    assert global_mem.is_dir()
    assert (global_mem / MEMORY_INDEX_FILENAME).is_file()
    # User skills dir created.
    assert (Path(cfg.data_dir) / "skills").is_dir()
    # Projects parent dir created.
    assert (Path(cfg.data_dir) / "projects").is_dir()


def test_governance_bootstrap_is_idempotent(tmp_path: Path) -> None:
    from my_deepagent.governance import bootstrap_user_dirs

    cfg = load_config(workspace_root=tmp_path, data_dir=tmp_path / "data")
    bootstrap_user_dirs(cfg)
    gpath = global_instructions_path(cfg)
    gpath.write_text("custom edited content", encoding="utf-8")
    # Second call must not overwrite user edits.
    bootstrap_user_dirs(cfg)
    assert gpath.read_text(encoding="utf-8") == "custom edited content"


# ---------------------------------------------------------------------------
# Integration: instruction paths reach deepagents memory= kwarg
# ---------------------------------------------------------------------------


def test_build_agent_receives_combined_instruction_and_memory_paths(
    tmp_path: Path, monkeypatch: pytest.MonkeyPatch
) -> None:
    """`build_agent(memory_paths_override=[instructions..., memory...])` passes
    the union through to `create_deep_agent(memory=...)`.  Mirrors what
    InteractiveSession does at REPL bootstrap.
    """
    from my_deepagent import session as session_mod
    from my_deepagent.persona import Persona

    captured: dict[str, Any] = {}

    def fake_create_deep_agent(**kwargs: Any) -> Any:
        captured.update(kwargs)
        return object()

    monkeypatch.setattr(session_mod, "create_deep_agent", fake_create_deep_agent)

    cfg = load_config(
        workspace_root=tmp_path,
        data_dir=tmp_path / "data",
        openrouter_api_key="test-key",
    )
    repo = tmp_path / "repo"
    repo.mkdir()

    g = ensure_global_instructions_initialized(cfg)
    proj_file = project_instructions_path(repo)
    proj_file.write_text("# project rule", encoding="utf-8")

    # Simulate a project memory entry.
    mem_entry = tmp_path / "MEM.md"
    mem_entry.write_text("# memory entry", encoding="utf-8")

    persona = Persona(
        name="test-persona",
        version=1,
        backend="openrouter",
        model="openrouter:deepseek/deepseek-chat",
        provider_origin="CN/DeepSeek",
        capabilities=("code_edit",),
        max_risk_level="high",
        system_prompt="System prompt for test persona (must be ≥10 chars)",
        deepagents_backend="state",
    )
    instruction_paths = resolve_instruction_paths(cfg, repo)
    combined = [*instruction_paths, str(mem_entry.resolve())]

    _agent = session_mod.build_agent(
        persona,
        cfg,
        root_dir=repo,
        memory_paths_override=combined,
    )
    assert "memory" in captured
    # Global must come before project, project before mem entry — exact list match.
    expected = [str(g.resolve()), str(proj_file.resolve()), str(mem_entry.resolve())]
    assert captured["memory"] == expected