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

"""v0.3 PR #3 — Auto-memory tests.

Covers:
1. Bootstrap creates an empty MEMORY.md index on first call (idempotent).
2. `add_memory_entry` writes a `<slug>.md` file + appends an index pointer line.
3. Auto-slug derivation + collision handling (`-2`, `-3`, … suffix).
4. `remove_memory_entry` deletes file + prunes index line (refuses to delete
   the index itself).
5. `list_memory_paths` puts MEMORY.md first (deepagents reads in order, so the
   index becomes the table-of-contents header in the system prompt).
6. `memory_entries_summary` skips the index, returns (name, size) pairs.
7. ``project_memory_dir`` rejects empty project_key.
8. Different `project_key`s isolate memory between repos.
9. End-to-end via deepagents MemoryMiddleware: `build_agent(..., memory_paths_override=[...])`
   constructs an agent whose middleware stack contains a MemoryMiddleware with
   the right sources.
"""

from __future__ import annotations

from pathlib import Path
from typing import Any

import pytest

from my_deepagent.config import load_config
from my_deepagent.memory import (
    INDEX_FILENAME,
    _slugify,
    add_memory_entry,
    ensure_memory_initialized,
    list_memory_paths,
    memory_entries_summary,
    project_memory_dir,
    remove_memory_entry,
)


@pytest.fixture
def memory_dir(tmp_path: Path) -> Path:
    return tmp_path / "memory"


# ---------------------------------------------------------------------------
# Bootstrap
# ---------------------------------------------------------------------------


def test_ensure_memory_initialized_creates_index(memory_dir: Path) -> None:
    assert not memory_dir.exists()
    index = ensure_memory_initialized(memory_dir)
    assert index.is_file()
    assert index.name == INDEX_FILENAME
    body = index.read_text(encoding="utf-8")
    assert "Auto-memory" in body
    assert "## Entries" in body


def test_ensure_memory_initialized_is_idempotent(memory_dir: Path) -> None:
    ensure_memory_initialized(memory_dir)
    index = memory_dir / INDEX_FILENAME
    original_content = index.read_text(encoding="utf-8")
    # Re-running should not overwrite an existing index.
    ensure_memory_initialized(memory_dir)
    assert index.read_text(encoding="utf-8") == original_content


# ---------------------------------------------------------------------------
# add_memory_entry
# ---------------------------------------------------------------------------


def test_add_memory_entry_writes_file_and_updates_index(memory_dir: Path) -> None:
    result = add_memory_entry(memory_dir, "프로젝트 핵심: 위크닥 CLI MVP")
    assert result.path.is_file()
    body = result.path.read_text(encoding="utf-8")
    assert "프로젝트 핵심" in body
    assert body.startswith("---\nname: ")
    assert "type:" in body
    assert result.scrubbed is False

    index = (memory_dir / INDEX_FILENAME).read_text(encoding="utf-8")
    assert result.path.name in index
    assert "프로젝트 핵심" in index


def test_add_memory_entry_handles_slug_collision(memory_dir: Path) -> None:
    r1 = add_memory_entry(memory_dir, "Same first line")
    r2 = add_memory_entry(memory_dir, "Same first line\nsecond entry body")
    r3 = add_memory_entry(memory_dir, "Same first line\nthird entry body")
    p1, p2, p3 = r1.path, r2.path, r3.path
    assert p1.name != p2.name != p3.name
    stems = sorted([p1.stem, p2.stem, p3.stem])
    assert stems[0] == "same-first-line"
    assert stems[1] == "same-first-line-2"
    assert stems[2] == "same-first-line-3"


def test_add_memory_entry_rejects_empty_content(memory_dir: Path) -> None:
    with pytest.raises(ValueError, match="non-empty"):
        add_memory_entry(memory_dir, "   \n  \t  ")


def test_add_memory_entry_explicit_name_override(memory_dir: Path) -> None:
    r = add_memory_entry(memory_dir, "Random body text", name="My Custom Slug!!")
    assert r.path.stem == "my-custom-slug"


def test_add_memory_entry_scrubs_openrouter_key(memory_dir: Path) -> None:
    r = add_memory_entry(
        memory_dir,
        "save this for me: sk-or-v1-abcdefghijklmnop1234567890",
    )
    body = r.path.read_text(encoding="utf-8")
    assert "sk-or-v1-abcdefghijklmnop" not in body
    assert "<redacted:openrouter-key>" in body
    assert r.scrubbed is True


def test_add_memory_entry_infers_user_type(memory_dir: Path) -> None:
    r = add_memory_entry(memory_dir, "I prefer fish shell over bash")
    assert r.memory_type == "user"


def test_add_memory_entry_infers_feedback_type(memory_dir: Path) -> None:
    r = add_memory_entry(memory_dir, "don't mock the database in integration tests")
    assert r.memory_type == "feedback"


def test_add_memory_entry_explicit_type_overrides_heuristic(memory_dir: Path) -> None:
    r = add_memory_entry(memory_dir, "I prefer fish shell", memory_type="reference")
    assert r.memory_type == "reference"


# ---------------------------------------------------------------------------
# remove_memory_entry
# ---------------------------------------------------------------------------


def test_remove_memory_entry_by_slug(memory_dir: Path) -> None:
    r = add_memory_entry(memory_dir, "to be forgotten")
    assert remove_memory_entry(memory_dir, r.path.stem) is True
    assert not r.path.exists()
    index_body = (memory_dir / INDEX_FILENAME).read_text(encoding="utf-8")
    assert r.path.name not in index_body


def test_remove_memory_entry_by_filename(memory_dir: Path) -> None:
    r = add_memory_entry(memory_dir, "to be forgotten by full filename")
    assert remove_memory_entry(memory_dir, r.path.name) is True
    assert not r.path.exists()


def test_remove_memory_entry_missing_returns_false(memory_dir: Path) -> None:
    ensure_memory_initialized(memory_dir)
    assert remove_memory_entry(memory_dir, "no-such-slug") is False


def test_remove_memory_entry_refuses_to_delete_index(memory_dir: Path) -> None:
    ensure_memory_initialized(memory_dir)
    assert remove_memory_entry(memory_dir, INDEX_FILENAME) is False
    assert (memory_dir / INDEX_FILENAME).is_file()


def test_remove_memory_entry_empty_input(memory_dir: Path) -> None:
    assert remove_memory_entry(memory_dir, "") is False


# ---------------------------------------------------------------------------
# list_memory_paths
# ---------------------------------------------------------------------------


def test_list_memory_paths_puts_index_first(memory_dir: Path) -> None:
    ensure_memory_initialized(memory_dir)
    add_memory_entry(memory_dir, "alpha entry")
    add_memory_entry(memory_dir, "beta entry")
    paths = list_memory_paths(memory_dir)
    assert len(paths) == 3
    assert Path(paths[0]).name == INDEX_FILENAME
    # Remaining ordered lexicographically.
    rest = [Path(p).name for p in paths[1:]]
    assert rest == sorted(rest)


def test_list_memory_paths_missing_dir_returns_empty(tmp_path: Path) -> None:
    assert list_memory_paths(tmp_path / "no-such-dir") == []


# ---------------------------------------------------------------------------
# memory_entries_summary
# ---------------------------------------------------------------------------


def test_memory_entries_summary_skips_index(memory_dir: Path) -> None:
    ensure_memory_initialized(memory_dir)
    add_memory_entry(memory_dir, "first")
    add_memory_entry(memory_dir, "second")
    summary = memory_entries_summary(memory_dir)
    names = [n for (n, _sz) in summary]
    assert INDEX_FILENAME not in names
    assert len(summary) == 2
    for _name, size in summary:
        assert size > 0


def test_memory_entries_summary_missing_dir(tmp_path: Path) -> None:
    assert memory_entries_summary(tmp_path / "no-such") == []


# ---------------------------------------------------------------------------
# project_memory_dir
# ---------------------------------------------------------------------------


def test_project_memory_dir_uses_project_key(tmp_path: Path) -> None:
    cfg = load_config(workspace_root=tmp_path, data_dir=tmp_path / "data")
    d1 = project_memory_dir(cfg, "abcdef0123456789")
    d2 = project_memory_dir(cfg, "fedcba9876543210")
    assert d1 != d2
    assert d1.parent.name == "abcdef0123456789"
    assert d2.parent.name == "fedcba9876543210"


def test_project_memory_dir_rejects_empty_key(tmp_path: Path) -> None:
    cfg = load_config(workspace_root=tmp_path, data_dir=tmp_path / "data")
    with pytest.raises(ValueError, match="non-empty"):
        project_memory_dir(cfg, "")


# ---------------------------------------------------------------------------
# Slugify
# ---------------------------------------------------------------------------


def test_slugify_basic() -> None:
    assert _slugify("Hello World!") == "hello-world"


def test_slugify_unicode_falls_back() -> None:
    # Korean text becomes purely separator characters → fallback "entry".
    assert _slugify("프로젝트 설정") == "entry"


def test_slugify_truncates_to_max_len() -> None:
    long = "a" * 100
    s = _slugify(long, max_len=20)
    assert len(s) <= 20


# ---------------------------------------------------------------------------
# Integration: build_agent threads memory paths through to deepagents
# ---------------------------------------------------------------------------


def test_build_agent_passes_memory_paths_to_deepagents(
    tmp_path: Path, monkeypatch: pytest.MonkeyPatch
) -> None:
    """Ensure `build_agent(..., memory_paths_override=[...])` causes
    `create_deep_agent` to receive the combined memory list.
    """
    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()  # opaque sentinel

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

    # Persona doubles as MemoryMiddleware source when populated; here we leave
    # persona.memory_files empty and rely entirely on the override.
    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",
    )
    cfg = load_config(
        workspace_root=tmp_path,
        data_dir=tmp_path / "data",
        openrouter_api_key="test-key",
    )

    # Build memory paths under tmp_path so the test is hermetic.
    memdir = tmp_path / "mem"
    ensure_memory_initialized(memdir)
    add_memory_entry(memdir, "do not commit secrets")
    paths = list_memory_paths(memdir)

    _agent = session_mod.build_agent(
        persona,
        cfg,
        root_dir=tmp_path,
        memory_paths_override=paths,
    )
    assert "memory" in captured
    assert captured["memory"] == paths


def test_build_agent_omits_memory_kwarg_when_no_paths(
    tmp_path: Path, monkeypatch: pytest.MonkeyPatch
) -> None:
    """Empty memory list → no `memory=` kwarg passed to deepagents (so the
    MemoryMiddleware is not constructed)."""
    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)

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

    _agent = session_mod.build_agent(persona, cfg, root_dir=tmp_path)
    assert "memory" not in captured