| """Tests for session extraction utilities.""" |
|
|
| from __future__ import annotations |
|
|
| import json |
| from typing import Any |
| from unittest.mock import MagicMock, patch |
|
|
| from extract import EXPECTED_KEYS, MODEL_NAME, SYSTEM_PROMPT, extract_session |
|
|
|
|
| def _sample_payload() -> dict[str, Any]: |
| """Return a complete sample extraction payload.""" |
| return { |
| "duration_minutes": 12, |
| "themes": ["scaffolding", "audio"], |
| "approaches_tried": [{"approach": "build a Gradio app", "outcome": "worked"}], |
| "dead_ends": [{"position": 0.4, "what_happened": "pytest was missing"}], |
| "breakthroughs": [{"position": 0.7, "what_worked": "mocked the client"}], |
| "gotchas": ["network access needed approval"], |
| "sentiment_arc": {"start": "curious", "end": "steady"}, |
| "skill_md": "# Skill\n\nUse the transcript.", |
| "slug_voice": [ |
| "You waited through the install.", |
| "The button found its job.", |
| "The mock kept things tidy.", |
| "The transcript became a map.", |
| "The Space now has a trail.", |
| ], |
| } |
|
|
|
|
| def test_extract_session_returns_expected_keys() -> None: |
| """The extractor parses chat completion JSON into the expected dict shape.""" |
| payload = _sample_payload() |
| response = { |
| "choices": [ |
| { |
| "message": { |
| "content": f"```json\n{json.dumps(payload)}\n```", |
| } |
| } |
| ] |
| } |
|
|
| with patch("extract.InferenceClient") as client_class: |
| client = MagicMock() |
| client.chat.completions.create.return_value = response |
| client_class.return_value = client |
|
|
| extraction = extract_session("session transcript") |
|
|
| assert EXPECTED_KEYS.issubset(extraction.keys()) |
| client.chat.completions.create.assert_called_once_with( |
| model=MODEL_NAME, |
| messages=[ |
| {"role": "system", "content": SYSTEM_PROMPT}, |
| {"role": "user", "content": "session transcript"}, |
| ], |
| response_format={"type": "json_object"}, |
| ) |
|
|
