"""Tests for enhanced dedup: text + structure dimensions (Issue #P2-后端3)."""

from __future__ import annotations

import sys
from unittest.mock import MagicMock

# ---------------------------------------------------------------------------
# Mock heavy deps before importing dedup module
# ---------------------------------------------------------------------------
_ORIGINAL_MODULES = dict(sys.modules)
_MOCKED_MODULE_NAMES: list[str] = []


def _mock_if_absent(name: str, mock_obj=None):
    """仅在模块不在 sys.modules 中时注入 mock，并记录以便清理。"""
    if name not in sys.modules:
        sys.modules[name] = mock_obj if mock_obj is not None else MagicMock()
        _MOCKED_MODULE_NAMES.append(name)


# Mock heavy deps
_mock_if_absent("ffmpeg")
_mock_if_absent("ffmpeg.utils")
_mock_if_absent("worker_app.celery_app")
_mock_if_absent("worker_app.db")
_mock_if_absent("packages.adapters.sqlalchemy_impl.generated_video_repository")
_mock_if_absent("packages.adapters.sqlalchemy_impl.models")
_mock_if_absent("packages.shared.storage")

# Mock cv2 and numpy if not available
try:
    import cv2 as _cv2

    if not isinstance(_cv2, MagicMock):
        _HAS_CV2 = True
    else:
        _HAS_CV2 = False
except ImportError:
    _HAS_CV2 = False
    _mock_if_absent("cv2")
    _mock_if_absent("numpy")

import pytest

from apps.worker.video_processing.dedup import (
    STRUCTURE_WEIGHT,
    TEXT_WEIGHT,
    VISUAL_WEIGHT,
    compute_structure_similarity,
    compute_text_similarity,
)


class TestTextSimilarity:
    """Tests for compute_text_similarity."""

    def test_identical_texts(self):
        """相同文本返回 1.0。"""
        assert compute_text_similarity("你好世界", "你好世界") == 1.0

    def test_empty_texts(self):
        """都为空返回 1.0。"""
        assert compute_text_similarity("", "") == 1.0

    def test_one_empty(self):
        """一个为空返回 0.0。"""
        assert compute_text_similarity("你好", "") == 0.0
        assert compute_text_similarity("", "你好") == 0.0

    def test_completely_different(self):
        """完全不同文本返回低相似度。"""
        sim = compute_text_similarity("你好世界", "abcdefgh")
        assert sim < 0.3

    def test_partial_overlap(self):
        """部分重叠文本返回中等相似度。"""
        sim = compute_text_similarity("今天天气真好", "今天天气不错")
        assert 0.3 < sim < 0.9

    def test_case_insensitive(self):
        """英文大小写不敏感。"""
        sim = compute_text_similarity("Hello World", "hello world")
        assert sim == 1.0

    def test_whitespace_ignored(self):
        """空白字符被忽略。"""
        sim = compute_text_similarity("你好 世界", "你好世界")
        assert sim == 1.0

    def test_punctuation_ignored(self):
        """标点符号被忽略。"""
        sim = compute_text_similarity("你好，世界！", "你好世界")
        assert sim == 1.0

    def test_long_texts(self):
        """长文本也能计算。"""
        t1 = "这是一段很长的配音文本，用于测试文案查重功能"
        t2 = "这是一段较长的配音文字，用于测试文案去重功能"
        sim = compute_text_similarity(t1, t2)
        assert 0.0 <= sim <= 1.0


class TestStructureSimilarity:
    """Tests for compute_structure_similarity."""

    def test_identical_structures(self):
        """完全相同结构返回 1.0。"""
        clips = [
            {"clip_type": "video", "duration": 5.0},
            {"clip_type": "title", "duration": 2.0},
            {"clip_type": "video", "duration": 8.0},
        ]
        assert compute_structure_similarity(clips, clips) == 1.0

    def test_empty_clips(self):
        """都为空返回 1.0。"""
        assert compute_structure_similarity([], []) == 1.0

    def test_one_empty(self):
        """一个为空返回 0.0。"""
        clips = [{"clip_type": "video", "duration": 5.0}]
        assert compute_structure_similarity(clips, []) == 0.0
        assert compute_structure_similarity([], clips) == 0.0

    def test_different_count(self):
        """片段数不同，相似度降低。"""
        clips1 = [
            {"clip_type": "video", "duration": 5.0},
            {"clip_type": "title", "duration": 2.0},
        ]
        clips2 = [
            {"clip_type": "video", "duration": 5.0},
            {"clip_type": "title", "duration": 2.0},
            {"clip_type": "video", "duration": 3.0},
            {"clip_type": "title", "duration": 1.0},
        ]
        sim = compute_structure_similarity(clips1, clips2)
        assert 0.0 < sim < 0.8

    def test_different_types(self):
        """片段类型不同，类型相似度低。"""
        clips1 = [
            {"clip_type": "video", "duration": 5.0},
            {"clip_type": "video", "duration": 3.0},
        ]
        clips2 = [
            {"clip_type": "title", "duration": 5.0},
            {"clip_type": "title", "duration": 3.0},
        ]
        sim = compute_structure_similarity(clips1, clips2)
        assert sim <= 0.6  # 类型全部不同，但数量和时长相同贡献 0.6

    def test_different_duration_distribution(self):
        """时长分布不同，时长相似度低。"""
        clips1 = [
            {"clip_type": "video", "duration": 10.0},  # 占比 80%
            {"clip_type": "title", "duration": 2.5},  # 占比 20%
        ]
        clips2 = [
            {"clip_type": "video", "duration": 2.0},  # 占比 20%
            {"clip_type": "title", "duration": 8.0},  # 占比 80%
        ]
        sim = compute_structure_similarity(clips1, clips2)
        assert 0.7 < sim < 0.9  # 类型相同但时长分布不同，sim=0.82

    def test_similar_structure(self):
        """相似结构返回较高相似度。"""
        clips1 = [
            {"clip_type": "video", "duration": 5.0},
            {"clip_type": "title", "duration": 2.0},
            {"clip_type": "video", "duration": 8.0},
        ]
        clips2 = [
            {"clip_type": "video", "duration": 5.5},
            {"clip_type": "title", "duration": 2.2},
            {"clip_type": "video", "duration": 7.5},
        ]
        sim = compute_structure_similarity(clips1, clips2)
        assert sim > 0.8

    def test_single_clip(self):
        """单片段也能计算。"""
        clips1 = [{"clip_type": "video", "duration": 10.0}]
        clips2 = [{"clip_type": "video", "duration": 12.0}]
        sim = compute_structure_similarity(clips1, clips2)
        assert sim > 0.5  # 类型相同，数量相同，只是时长不同


class TestDimensionWeights:
    """Tests for dimension weight constants."""

    def test_weights_sum_to_one(self):
        """多维度权重之和为 1.0。"""
        assert abs(VISUAL_WEIGHT + TEXT_WEIGHT + STRUCTURE_WEIGHT - 1.0) < 1e-9

    def test_visual_weight_is_half(self):
        """视觉权重为 0.5。"""
        assert VISUAL_WEIGHT == 0.5

    def test_text_weight_is_quarter(self):
        """文案权重为 0.25。"""
        assert TEXT_WEIGHT == 0.25

    def test_structure_weight_is_quarter(self):
        """结构权重为 0.25。"""
        assert STRUCTURE_WEIGHT == 0.25