CADx-4/tests/test_ceus_video.py

"""VideoFile 单元测试：probe / iter_frames / frame_times_s / 打不开报错。

旧测试读一段真实样例视频的字节；这里在测试里用 cv2 现造一段几十帧的小视频，按路径读。
旧版先把字节落临时文件，那三条「临时文件有没有删」的测试随之不再适用，已删。
"""

from pathlib import Path

import cv2
import numpy as np
import pytest

from cadx.ceus import VideoFile

_FRAMES = 40
_WIDTH, _HEIGHT = 64, 48


@pytest.fixture(scope="module")
def sample_path(tmp_path_factory: pytest.TempPathFactory) -> Path:
    """用 cv2 现造一段 40 帧、64×48、25 fps 的 mp4，每帧亮度不同。"""
    path = tmp_path_factory.mktemp("video") / "sample.mp4"
    writer = cv2.VideoWriter(str(path), cv2.VideoWriter.fourcc(*"mp4v"), 25.0, (_WIDTH, _HEIGHT))
    for index in range(_FRAMES):
        frame = np.full((_HEIGHT, _WIDTH, 3), (index * 6) % 256, dtype=np.uint8)
        frame[index % _HEIGHT, :, 2] = 255
        writer.write(frame)
    writer.release()
    return path


def test_probe_returns_positive_metadata(sample_path: Path) -> None:
    """probe 返回 (fps, frame_count, width, height)：字段顺序/类型/语义范围均须正确。"""
    video = VideoFile(sample_path)
    result = video.probe()
    assert isinstance(result, tuple) and len(result) == 4
    fps, frame_count, width, height = result
    # 字段类型契约（app/ports.py 的 DecodedVideo）：fps 为 float，其余为 int
    assert isinstance(fps, float)
    assert isinstance(frame_count, int)
    assert isinstance(width, int)
    assert isinstance(height, int)
    # 语义合理范围——并钉死字段顺序未颠倒（fps 是小帧率、尺寸是较大像素值）
    assert 0 < fps < 1000  # 帧率不会是几百几千；若与 frame_count(881) 颠倒会触发
    assert frame_count > 0
    assert width > 1 and height > 1


def test_probe_field_order_matches_iter(sample_path: Path) -> None:
    """probe 的 (width, height) 须与 iter_frames 实际帧的 (W, H) 一致（防字段顺序颠倒）。"""
    video = VideoFile(sample_path)
    _fps, _frame_count, width, height = video.probe()
    _idx, frame = next(video.iter_frames())
    assert frame.shape == (height, width, 3)


def test_iter_frames_indices_contiguous_and_count_matches_probe(
    sample_path: Path,
) -> None:
    """iter_frames 产 (idx, frame)：idx 从 0 连续递增，且帧数 == probe 的 frame_count。"""
    video = VideoFile(sample_path)
    _, frame_count, _, _ = video.probe()
    indices = [idx for idx, _frame in video.iter_frames()]

    assert indices == list(range(len(indices)))
    assert len(indices) == frame_count


def test_iter_frames_yields_bgr_uint8_hwc(sample_path: Path) -> None:
    """每帧为 (H, W, 3) uint8 的 ndarray，且 H/W 与 probe 一致。"""
    video = VideoFile(sample_path)
    _, _, width, height = video.probe()
    idx, frame = next(video.iter_frames())

    assert idx == 0
    assert isinstance(frame, np.ndarray)
    assert frame.dtype == np.uint8
    assert frame.shape == (height, width, 3)


def test_frame_times_are_strictly_increasing_and_match_frame_count(
    sample_path: Path,
) -> None:
    video = VideoFile(sample_path)
    _fps, frame_count, _width, _height = video.probe()
    times = video.frame_times_s()

    assert times.shape == (frame_count,)
    assert times[0] >= 0.0
    assert np.all(np.diff(times) > 0.0)


def test_解两遍逐帧像素完全相同(sample_path: Path) -> None:
    """**这条是接缝改成「打开一次」的全部理由**（design/0024）。

    CEUS 那条线要分几趟过同一段视频，每趟只留这一趟的小结果。第二趟必须从第 0 帧重新开始,
    而且看到的像素要和第一趟一模一样——多趟方案整个建立在这个前提上。它今天成立是因为
    每趟各开一个 VideoCapture 顺序读同一个文件，但**这件事此前没有任何一处验证过**。

    **比的是全片，而且是流式比。** 只比前几帧的话，可变帧率或者 seek 行为导致的漂移多半
    出现在后面而不是开头。两趟各自逐帧过、逐帧喂进哈希，峰值内存是两帧——**这条测试自己
    也不许把视频攒起来**，它守的正是这件事。
    """
    import hashlib

    def _digest(video: VideoFile) -> tuple[str, int]:
        h = hashlib.blake2b(digest_size=16)
        count = 0
        for _, frame in video.iter_frames():
            h.update(np.ascontiguousarray(frame).tobytes())
            count += 1
        return h.hexdigest(), count

    video = VideoFile(sample_path)
    first = _digest(video)
    second = _digest(video)

    assert first[1] > 0
    assert first == second, "同一段视频解两遍，逐帧像素必须完全相同"


def test_两趟各自从第零帧开始互不干扰(sample_path: Path) -> None:
    """两趟同时在飞时各读各的游标。

    **断言比的是像素，不是 iter_frames 自己报的下标**——那个下标是本模块里一个 Python
    计数器，就算内层真把外层的读位置搅乱了，它照样从 0 数上去。
    """
    video = VideoFile(sample_path)
    outer = video.iter_frames()
    _, outer_frame_0 = next(outer)
    # 内层整个跑起来，中途丢掉——它不该动到外层的游标
    inner = video.iter_frames()
    _, inner_frame_0 = next(inner)
    _, inner_frame_1 = next(inner)
    del inner
    _, outer_frame_1 = next(outer)

    reference = [frame for _, frame in zip(video.iter_frames(), range(2), strict=False)]

    assert np.array_equal(inner_frame_0, outer_frame_0), "内层没有从第 0 帧开始"
    assert np.array_equal(outer_frame_1, inner_frame_1), "外层被内层带跑了：它接着读到的不是第 1 帧"
    assert len(reference) == 2


def test_bad_bytes_raise_not_silent(tmp_path: Path) -> None:
    """坏文件无法打开 → 报错（不静默返回空 probe / 空迭代器）。"""
    garbage = tmp_path / "garbage.mp4"
    garbage.write_bytes(b"this is not a video file" * 64)
    video = VideoFile(garbage)

    with pytest.raises(RuntimeError):
        video.probe()
    with pytest.raises(RuntimeError):
        list(video.iter_frames())