ObsiGate/tests/test_file_chat.py

# tests/test_file_chat.py — Chat par fichier (#169)
"""Store + REST routes for the per-file chat history."""

import time

import pytest

from backend import file_chat as _store


@pytest.fixture(autouse=True)
def _isolated_chat_dir(tmp_path, monkeypatch):
    """Keep every test (store AND routes) out of the real ``data/chats``."""
    monkeypatch.setattr(_store, "CHAT_DIR", tmp_path / "chats")


class TestStore:
    def test_get_messages_empty_when_missing(self, tmp_path, monkeypatch):
        monkeypatch.setattr(_store, "CHAT_DIR", tmp_path / "chats")
        assert _store.get_messages("V", "a/b.md") == []

    def test_add_then_get_roundtrip(self, tmp_path, monkeypatch):
        monkeypatch.setattr(_store, "CHAT_DIR", tmp_path / "chats")
        msg = _store.add_message("V", "a/b.md", "bruno", "Bonjour !")
        assert msg["user"] == "bruno"
        assert msg["text"] == "Bonjour !"
        assert msg["id"]
        assert msg["ts"] <= time.time() + 1
        got = _store.get_messages("V", "a/b.md")
        assert len(got) == 1
        assert got[0]["id"] == msg["id"]

    def test_conversation_is_per_file(self, tmp_path, monkeypatch):
        monkeypatch.setattr(_store, "CHAT_DIR", tmp_path / "chats")
        _store.add_message("V", "a.md", "u", "pour a")
        _store.add_message("V", "b.md", "u", "pour b")
        assert [m["text"] for m in _store.get_messages("V", "a.md")] == ["pour a"]
        assert [m["text"] for m in _store.get_messages("V", "b.md")] == ["pour b"]

    def test_path_traversal_never_escapes_chat_dir(self, tmp_path, monkeypatch):
        """The chat file name is hashed: '../../etc' cannot escape CHAT_DIR."""
        chat_dir = tmp_path / "chats"
        monkeypatch.setattr(_store, "CHAT_DIR", chat_dir)
        _store.add_message("V", "../../etc/passwd", "u", "x")
        files = list(chat_dir.iterdir())
        assert len(files) == 1
        assert files[0].parent == chat_dir
        assert "/" not in files[0].name and "\\" not in files[0].name

    def test_retention_cap_drops_oldest(self, tmp_path, monkeypatch):
        monkeypatch.setattr(_store, "CHAT_DIR", tmp_path / "chats")
        n = _store.MAX_MESSAGES + 25
        for i in range(n):
            _store.add_message("V", "a.md", "u", f"m{i}")
        got = _store.get_messages("V", "a.md")
        assert len(got) == _store.MAX_MESSAGES
        assert got[0]["text"] == f"m{25}"  # first 25 dropped
        assert got[-1]["text"] == f"m{n - 1}"

    def test_text_truncated_to_max(self, tmp_path, monkeypatch):
        monkeypatch.setattr(_store, "CHAT_DIR", tmp_path / "chats")
        _store.add_message("V", "a.md", "u", "x" * (_store.MAX_TEXT + 500))
        got = _store.get_messages("V", "a.md")
        assert len(got[0]["text"]) == _store.MAX_TEXT

    def test_corrupt_file_falls_back_to_empty(self, tmp_path, monkeypatch):
        chat_dir = tmp_path / "chats"
        monkeypatch.setattr(_store, "CHAT_DIR", chat_dir)
        chat_dir.mkdir(parents=True)
        target = _store._chat_file("V", "a.md")
        target.write_text("{not json", encoding="utf-8")
        assert _store.get_messages("V", "a.md") == []


class TestRoutes:
    def test_history_requires_existing_vault(self, client, test_vault_dir):
        r = client.get("/api/file/TestVault/chat", params={"path": "note1.md"})
        assert r.status_code == 200
        assert r.json() == {"messages": []}

    def test_post_and_get_roundtrip(self, client, test_vault_dir):
        r = client.post(
            "/api/file/TestVault/chat",
            json={"path": "note1.md", "text": "Vu hier ?"},
        )
        assert r.status_code == 200
        body = r.json()
        assert body["status"] == "ok"
        assert body["message"]["text"] == "Vu hier ?"
        r2 = client.get("/api/file/TestVault/chat", params={"path": "note1.md"})
        assert [m["text"] for m in r2.json()["messages"]] == ["Vu hier ?"]

    def test_post_requires_text(self, client, test_vault_dir):
        r = client.post("/api/file/TestVault/chat", json={"path": "note1.md", "text": "   "})
        assert r.status_code == 400

    def test_post_requires_path(self, client, test_vault_dir):
        r = client.post("/api/file/TestVault/chat", json={"path": "", "text": "hi"})
        assert r.status_code == 400

    def test_unknown_vault_404(self, client, test_vault_dir):
        r = client.get("/api/file/Nope/chat", params={"path": "x.md"})
        assert r.status_code == 404

    def test_path_traversal_rejected(self, client, test_vault_dir):
        r = client.get(
            "/api/file/TestVault/chat", params={"path": "../../etc/passwd"}
        )
        assert r.status_code in (403, 500)
        r2 = client.post(
            "/api/file/TestVault/chat",
            json={"path": "../../etc/passwd", "text": "x"},
        )
        assert r2.status_code in (403, 500)


@pytest.mark.parametrize("path", ["note1.md", "sub/deep/file.md"])
def test_history_per_path_isolated(client, test_vault_dir, path):
    client.post("/api/file/TestVault/chat", json={"path": path, "text": "m"})
    other = "note1.md" if path != "note1.md" else "note2.md"
    r = client.get("/api/file/TestVault/chat", params={"path": other})
    assert r.json()["messages"] == []