tests/test_api.py

from io import BytesIO
import re

import numpy as np
import pytest
import soundfile as sf
from fastapi.testclient import TestClient

from app.main import app


client = TestClient(app)


@pytest.fixture(autouse=True)
def disable_llm_for_api_tests(monkeypatch):
    monkeypatch.delenv("DOGVOICE_LLM_API_KEY", raising=False)


def test_health():
    response = client.get("/health")
    assert response.status_code == 200
    assert response.json()["ok"] is True
    assert response.json()["llm"]["configured"] is False


def test_analyze_generated_wav_and_feedback():
    wav = _make_wav()
    response = client.post(
        "/analyze-dog-audio",
        files={"file": ("bark.wav", wav, "audio/wav")},
        data={"context_note": "门外有人经过，狗狗叫了几声。"},
    )
    assert response.status_code == 200
    data = response.json()
    assert data["ok"] is True
    assert data["result_id"]
    assert data["scene"] == "audio_only"
    assert data["arousal"] in {"Low", "Medium", "High"}
    assert data["arousal_label"] in {"低", "中", "高"}
    assert data["valence"] in {"Negative", "Neutral", "Positive"}
    assert data["valence_label"] in {"偏负向", "中性", "偏正向"}
    assert data["confidence_label"] in {"低", "中", "高"}
    assert data["model_backend_label"] in {"狗叫情绪识别模型", "备用声学特征分析"}
    assert data["primary_intent"]["label"]
    assert data["primary_intent"]["likelihood"] in {"较高", "中等", "较低"}
    assert len(data["intent_candidates"]) == 3
    assert data["sound_profile"]["signals"]
    assert data["care"]["urgency_label"]
    assert data["care"]["next_steps"]
    assert data["rule_summary"]
    assert data["user_context_note"] == "门外有人经过，狗狗叫了几声。"
    assert data["professional_note"]
    assert data["science_basis"]
    assert data["behavior_knowledge"]
    _assert_clean_visible_text(data["professional_note"])
    _assert_clean_visible_text(" ".join(f"{item['label']} {item['text']}" for item in data["science_basis"]))
    assert data["llm_interpretation"]["status"] == "disabled"
    assert "High" not in data["summary"]
    assert "Medium" not in data["summary"]
    assert "Low" not in data["summary"]
    assert "Positive" not in data["summary"]
    assert "Neutral" not in data["summary"]
    assert "Negative" not in data["summary"]
    assert data["timeline"]
    assert data["timeline"][0]["arousal_label"]
    assert data["timeline"][0]["valence_label"]

    feedback = client.post(
        "/feedback",
        data={
            "result_id": data["result_id"],
            "rating": "partial",
            "actual_reason": "door_or_outside_noise",
            "note": "测试反馈",
        },
    )
    assert feedback.status_code == 200
    assert feedback.json()["ok"] is True

    followup = client.post(
        "/follow-up",
        data={
            "result_id": data["result_id"],
            "message": "它最近独处时也会叫，这个需要担心吗？",
        },
    )
    assert followup.status_code == 200
    followup_data = followup.json()
    assert followup_data["ok"] is True
    assert followup_data["reply"]
    assert followup_data["professional_note"]
    assert followup_data["science_basis"]
    assert followup_data["behavior_knowledge"]
    _assert_clean_visible_text(followup_data["professional_note"])
    _assert_clean_visible_text(" ".join(f"{item['label']} {item['text']}" for item in followup_data["science_basis"]))


def test_rejects_short_wav():
    wav = _make_wav(duration=0.5)
    response = client.post(
        "/analyze-dog-audio",
        files={"file": ("short.wav", wav, "audio/wav")},
    )
    assert response.status_code == 400
    assert response.json()["detail"]["ok"] is False


def test_legacy_scene_field_is_ignored():
    wav = _make_wav()
    response = client.post(
        "/analyze-dog-audio",
        files={"file": ("bark.wav", wav, "audio/wav")},
        data={"scene": "food"},
    )
    assert response.status_code == 200
    assert response.json()["scene"] == "audio_only"


def _make_wav(duration: float = 3.0, sample_rate: int = 22_050) -> BytesIO:
    t = np.linspace(0, duration, int(sample_rate * duration), endpoint=False)
    carrier = np.sin(2 * np.pi * 520 * t)
    envelope = (np.sin(2 * np.pi * 3 * t) > 0).astype(float) * 0.45 + 0.05
    signal = (carrier * envelope).astype(np.float32)
    buffer = BytesIO()
    sf.write(buffer, signal, sample_rate, format="WAV")
    buffer.seek(0)
    return buffer


def _assert_clean_visible_text(text: str) -> None:
    for term in ("Dogbug", "Dogbark", "Dock", "提示词", "资料库检索结果", "专业依据摘要", "规则底稿", "声音需要结合场景解读", "当时场景", "选择场景"):
        assert term not in text
    assert not re.search(r"[A-Za-z]", text)