feat: migrate Mac mini services into Reyna CLI

tests/test_apple_llm.py

@@ -0,0 +1,74 @@
+
+import pytest
+from unittest.mock import MagicMock, patch
+from reyna_cli.apple_llm_client import AppleLLMClient, get_apple_llm_session, apple_llm_close, apple_llm_status
+
+@pytest.fixture
+def mock_apple_llm():
+    with patch("subprocess.Popen") as mock_popen, \
+         patch("subprocess.run") as mock_run, \
+         patch("pathlib.Path.write_text"), \
+         patch("shutil.rmtree"):
+
+        # Mock the binary build
+        mock_run.return_value = MagicMock(returncode=0)
+
+        # Mock the process
+        mock_proc = MagicMock()
+        mock_proc.poll.return_value = None
+        mock_proc.pid = 1234
+        mock_proc.stdout.readline.side_effect = [
+            '{"id": "0", "ok": true, "text": "Polished text", "ms": 100, "mode": "line"}\\n',
+            '{"id": "1", "ok": true, "text": "Quick reply", "ms": 50, "mode": "quick_reply"}\\n',
+            '{"id": "2", "ok": true, "text": "Chat response", "ms": 200, "mode": "chat"}\\n',
+            '', # EOF
+        ]
+        mock_popen.return_value = mock_proc
+
+        client = AppleLLMClient()
+        yield client
+
+def test_apple_llm_lifecycle(mock_apple_llm):
+    # Test status before start
+    status = apple_llm_status()
+    assert status["active"] is False
+
+    # Start and check status
+    mock_apple_llm.start()
+    status = apple_llm_status()
+    # Note: apple_llm_status uses the global singleton, not the fixture instance.
+    # For the purpose of these tests, we'll mock the global session if needed,
+    # but let's focus on the Client logic first.
+
+def test_apple_llm_call_success(mock_apple_llm):
+    payload = {"mode": "line", "text": "Hello world"}
+    mock_proc = MagicMock()
+    mock_apple_llm.proc = mock_proc
+
+    def reply_on_write(_: str) -> None:
+        event, result_box = mock_apple_llm.pending["0"]
+        result_box["data"] = {"ok": True, "text": "Polished!"}
+        event.set()
+
+    mock_proc.stdin.write.side_effect = reply_on_write
+    with patch.object(mock_apple_llm, "start"):
+        res = mock_apple_llm.call(payload)
+
+    assert res["ok"] is True
+    assert res["text"] == "Polished!"
+
+def test_apple_llm_timeout(mock_apple_llm):
+    with patch("threading.Event.wait", return_value=False):
+        res = mock_apple_llm.call({"mode": "line", "text": "test"}, timeout=0.1)
+        assert res["ok"] is False
+        assert "timeout" in res["error"]
+
+def test_apple_llm_check_mock(mock_apple_llm):
+    with patch("subprocess.run") as mock_run:
+        mock_run.return_value = MagicMock(
+            stdout='{"ok": true, "available": true, "ping": "ok"}',
+            returncode=0
+        )
+        res = mock_apple_llm.check()
+        assert res["ok"] is True
+        assert res["available"] is True