tests/test_data_processing.py

"""Tests for /api/process/audio endpoint."""

from __future__ import annotations

import io
import json

import pytest
from fastapi.testclient import TestClient

from app.routes.data_processing import _process_rate_store


@pytest.fixture(autouse=True)
def _reset_rate_limiter():
    """The rate limiter's store is module-level and shared across every
    test in this file (10 requests/60s per IP, and TestClient always uses
    the same fake client IP) — without resetting it, tests that pass in
    isolation start failing with 429 once enough tests run before them in
    the same process."""
    _process_rate_store.clear()
    yield
    _process_rate_store.clear()


def _valid_options(**overrides: object) -> str:
    """Return a valid JSON options string with optional overrides."""
    defaults = {
        "pitchShift": False,
        "speedChange": False,
        "bassBoost": False,
        "trimSilence": False,
        "mixAudio": False,
        "addNoise": False,
    }
    defaults.update(overrides)
    return json.dumps(defaults)


def _fake_audio(content: bytes = b"\x00" * 1024) -> io.BytesIO:
    return io.BytesIO(content)


def test_audio_valid_request_accepted(client: TestClient) -> None:
    """Valid audio file + valid options should not return 422."""
    response = client.post(
        "/api/process/audio",
        data={"options": _valid_options()},
        files={"file": ("test.wav", _fake_audio(), "audio/wav")},
    )
    # Should be processed (200) or a processing error (400/500) — never 422
    assert response.status_code != 422


def test_audio_rejects_non_audio(client: TestClient) -> None:
    """Should reject non-audio content type with 400."""
    response = client.post(
        "/api/process/audio",
        data={"options": _valid_options()},
        files={"file": ("test.txt", _fake_audio(), "text/plain")},
    )
    assert response.status_code == 400
    detail = response.json()["detail"]
    assert detail["code"] == "invalid_file_type"


def test_audio_rejects_missing_content_type(client: TestClient) -> None:
    """Should reject file with empty content type as invalid."""
    response = client.post(
        "/api/process/audio",
        data={"options": _valid_options()},
        files={"file": ("test.bin", _fake_audio(), "")},
    )
    assert response.status_code == 400
    detail = response.json()["detail"]
    assert detail["code"] == "invalid_file_type"


def test_audio_rejects_missing_file(client: TestClient) -> None:
    """Should reject request without file."""
    response = client.post(
        "/api/process/audio",
        data={"options": _valid_options()},
    )
    assert response.status_code == 422


def test_audio_rejects_invalid_options_json(client: TestClient) -> None:
    """Should reject malformed JSON in options with 422 + invalid_options code."""
    response = client.post(
        "/api/process/audio",
        data={"options": "not-valid-json"},
        files={"file": ("test.wav", _fake_audio(), "audio/wav")},
    )
    assert response.status_code == 422
    detail = response.json()["detail"]
    assert detail["code"] == "invalid_options"


def test_audio_accepts_camel_case_options(client: TestClient) -> None:
    """Frontend sends camelCase keys — should be accepted."""
    response = client.post(
        "/api/process/audio",
        data={"options": _valid_options(pitchShift=True, addNoise=True)},
        files={"file": ("test.wav", _fake_audio(), "audio/wav")},
    )
    assert response.status_code != 422


def test_audio_defaults_when_options_missing(client: TestClient) -> None:
    """Missing options field should default to all-off, not 422."""
    response = client.post(
        "/api/process/audio",
        files={"file": ("test.wav", _fake_audio(), "audio/wav")},
    )
    # Should proceed to processing (200) or processing error — never 422
    assert response.status_code != 422


def test_audio_rejects_oversized_file_with_413_not_500(client: TestClient) -> None:
    """Regression test: the 413 raised inside the read loop's try block was
    being caught by the bare `except Exception` below it (HTTPException IS
    an Exception) and replaced with a misleading 500. A file over the 30MB
    cap must surface as 413 file_too_large, not 500 internal_error."""
    oversized = _fake_audio(b"\x00" * (31 * 1024 * 1024))
    response = client.post(
        "/api/process/audio",
        data={"options": _valid_options()},
        files={"file": ("big.wav", oversized, "audio/wav")},
    )
    assert response.status_code == 413
    detail = response.json()["detail"]
    assert detail["code"] == "file_too_large"


# ─────────────────────────── /api/process/audio/convert ───────────────────────────

def _valid_convert_options(**overrides: object) -> str:
    defaults = {"targetFormat": "wav", "bitrateKbps": 192}
    defaults.update(overrides)
    return json.dumps(defaults)


def test_convert_valid_request_accepted(client: TestClient) -> None:
    """Valid audio file + valid convert options should not return 422."""
    response = client.post(
        "/api/process/audio/convert",
        data={"options": _valid_convert_options()},
        files={"file": ("test.wav", _fake_audio(), "audio/wav")},
    )
    assert response.status_code != 422


def test_convert_rejects_non_audio(client: TestClient) -> None:
    response = client.post(
        "/api/process/audio/convert",
        data={"options": _valid_convert_options()},
        files={"file": ("test.txt", _fake_audio(), "text/plain")},
    )
    assert response.status_code == 400
    detail = response.json()["detail"]
    assert detail["code"] == "invalid_file_type"


def test_convert_rejects_invalid_target_format(client: TestClient) -> None:
    """targetFormat outside the wav/mp3/flac/ogg enum should 422."""
    response = client.post(
        "/api/process/audio/convert",
        data={"options": _valid_convert_options(targetFormat="exe")},
        files={"file": ("test.wav", _fake_audio(), "audio/wav")},
    )
    assert response.status_code == 422
    detail = response.json()["detail"]
    assert detail["code"] == "invalid_options"


def test_convert_rejects_bitrate_out_of_range(client: TestClient) -> None:
    response = client.post(
        "/api/process/audio/convert",
        data={"options": _valid_convert_options(bitrateKbps=999)},
        files={"file": ("test.wav", _fake_audio(), "audio/wav")},
    )
    assert response.status_code == 422
    detail = response.json()["detail"]
    assert detail["code"] == "invalid_options"


def test_convert_defaults_when_options_missing(client: TestClient) -> None:
    """Missing options should default to wav @ 192kbps, not 422."""
    response = client.post(
        "/api/process/audio/convert",
        files={"file": ("test.wav", _fake_audio(), "audio/wav")},
    )
    assert response.status_code != 422


def test_convert_rejects_oversized_file_with_413_not_500(client: TestClient) -> None:
    """Same regression as the /audio endpoint: 413 must not become 500."""
    oversized = _fake_audio(b"\x00" * (31 * 1024 * 1024))
    response = client.post(
        "/api/process/audio/convert",
        data={"options": _valid_convert_options()},
        files={"file": ("big.wav", oversized, "audio/wav")},
    )
    assert response.status_code == 413
    detail = response.json()["detail"]
    assert detail["code"] == "file_too_large"


# ─────────────────────────── /api/process/audio/organize ───────────────────────────

def test_organize_valid_request_accepted(client: TestClient) -> None:
    response = client.post(
        "/api/process/audio/organize",
        files={"file": ("test.wav", _fake_audio(), "audio/wav")},
    )
    # A silent all-zero WAV is a legitimate 400 (analyzer rejects silence) —
    # never 422 (that's the "malformed request" status, not "bad content").
    assert response.status_code != 422


def test_organize_rejects_non_audio(client: TestClient) -> None:
    response = client.post(
        "/api/process/audio/organize",
        files={"file": ("test.txt", _fake_audio(), "text/plain")},
    )
    assert response.status_code == 400
    detail = response.json()["detail"]
    assert detail["code"] == "invalid_file_type"


def test_organize_rejects_missing_file(client: TestClient) -> None:
    response = client.post("/api/process/audio/organize")
    assert response.status_code == 422


def test_organize_rejects_oversized_file_with_413(client: TestClient) -> None:
    oversized = _fake_audio(b"\x00" * (31 * 1024 * 1024))
    response = client.post(
        "/api/process/audio/organize",
        files={"file": ("big.wav", oversized, "audio/wav")},
    )
    assert response.status_code == 413
    detail = response.json()["detail"]
    assert detail["code"] == "file_too_large"


def test_organize_returns_real_metadata_for_real_audio(client: TestClient) -> None:
    """End-to-end with an actual sine wave — not a silent/zero fixture —
    to prove the analyzer runs and returns genuine acoustic metadata."""
    import io as _io
    import numpy as _np
    import soundfile as _sf

    sr = 22050
    t = _np.linspace(0, 2, sr * 2)
    y = (0.4 * _np.sin(2 * _np.pi * 440 * t)).astype(_np.float32)
    buf = _io.BytesIO()
    _sf.write(buf, y, sr, format="WAV")
    buf.seek(0)

    response = client.post(
        "/api/process/audio/organize",
        files={"file": ("tone.wav", buf, "audio/wav")},
    )
    assert response.status_code == 200
    body = response.json()
    assert body["durationSec"] == pytest.approx(2.0, abs=0.1)
    assert isinstance(body["tempoBpm"], (int, float))
    assert "major" in body["key"] or "minor" in body["key"]
    assert isinstance(body["tags"], list) and len(body["tags"]) > 0