xiaoxia-saas/tests/unit/test_duplication_domain.py

"""
Duplication 查重记录领域模型单元测试
"""

import pytest

from packages.domain.duplication import DuplicateSegment, DuplicationRecord


class TestDuplicateSegmentCreate:
    """DuplicateSegment.create 测试"""

    def test_create_success(self):
        seg = DuplicateSegment.create(
            source_start=10.0,
            source_end=20.0,
            matched_video_id="vid_123",
            matched_video_name="测试视频",
            matched_start=5.0,
            matched_end=15.0,
            similarity=85.5,
        )
        assert seg.id is not None
        assert len(seg.id) == 32
        assert seg.source_start == 10.0
        assert seg.source_end == 20.0
        assert seg.matched_video_id == "vid_123"
        assert seg.matched_video_name == "测试视频"
        assert seg.matched_start == 5.0
        assert seg.matched_end == 15.0
        assert seg.similarity == 85.5

    def test_invalid_source_negative_start(self):
        with pytest.raises(ValueError, match="invalid source segment range"):
            DuplicateSegment.create(
                source_start=-1.0,
                source_end=10.0,
                matched_video_id="v1",
                matched_video_name="n1",
                matched_start=0,
                matched_end=10,
                similarity=50,
            )

    def test_invalid_source_end_before_start(self):
        with pytest.raises(ValueError, match="invalid source segment range"):
            DuplicateSegment.create(
                source_start=20.0,
                source_end=10.0,
                matched_video_id="v1",
                matched_video_name="n1",
                matched_start=0,
                matched_end=10,
                similarity=50,
            )

    def test_invalid_source_end_equals_start(self):
        with pytest.raises(ValueError, match="invalid source segment range"):
            DuplicateSegment.create(
                source_start=10.0,
                source_end=10.0,
                matched_video_id="v1",
                matched_video_name="n1",
                matched_start=0,
                matched_end=10,
                similarity=50,
            )

    def test_invalid_matched_negative_start(self):
        with pytest.raises(ValueError, match="invalid matched segment range"):
            DuplicateSegment.create(
                source_start=0,
                source_end=10,
                matched_video_id="v1",
                matched_video_name="n1",
                matched_start=-5,
                matched_end=10,
                similarity=50,
            )

    def test_invalid_matched_end_before_start(self):
        with pytest.raises(ValueError, match="invalid matched segment range"):
            DuplicateSegment.create(
                source_start=0,
                source_end=10,
                matched_video_id="v1",
                matched_video_name="n1",
                matched_start=15,
                matched_end=10,
                similarity=50,
            )

    def test_invalid_similarity_negative(self):
        with pytest.raises(ValueError, match="similarity must be between 0 and 100"):
            DuplicateSegment.create(
                source_start=0,
                source_end=10,
                matched_video_id="v1",
                matched_video_name="n1",
                matched_start=0,
                matched_end=10,
                similarity=-1,
            )

    def test_invalid_similarity_over_100(self):
        with pytest.raises(ValueError, match="similarity must be between 0 and 100"):
            DuplicateSegment.create(
                source_start=0,
                source_end=10,
                matched_video_id="v1",
                matched_video_name="n1",
                matched_start=0,
                matched_end=10,
                similarity=101,
            )

    def test_similarity_boundary_zero(self):
        seg = DuplicateSegment.create(
            source_start=0,
            source_end=10,
            matched_video_id="v1",
            matched_video_name="n1",
            matched_start=0,
            matched_end=10,
            similarity=0,
        )
        assert seg.similarity == 0

    def test_similarity_boundary_100(self):
        seg = DuplicateSegment.create(
            source_start=0,
            source_end=10,
            matched_video_id="v1",
            matched_video_name="n1",
            matched_start=0,
            matched_end=10,
            similarity=100,
        )
        assert seg.similarity == 100


class TestDuplicationRecordCreate:
    """DuplicationRecord.create 测试"""

    def test_create_minimal(self):
        record = DuplicationRecord.create(
            user_id="user123",
            filename="test.mp4",
            file_size=1024000,
            storage_key="oss://bucket/test.mp4",
        )
        assert record.id is not None
        assert len(record.id) == 32
        assert record.user_id == "user123"
        assert record.filename == "test.mp4"
        assert record.file_size == 1024000
        assert record.storage_key == "oss://bucket/test.mp4"
        assert record.status == "pending"
        assert record.duplicate_rate is None
        assert record.duplicate_count == 0
        assert record.segments == []
        assert record.duration_seconds == 0.0
        assert record.created_at is not None
        assert record.updated_at is not None

    def test_create_with_duration(self):
        record = DuplicationRecord.create(
            user_id="u1",
            filename="video.mp4",
            file_size=5000,
            storage_key="key",
            duration_seconds=120.5,
        )
        assert record.duration_seconds == 120.5

    def test_create_strips_whitespace(self):
        record = DuplicationRecord.create(
            user_id="  user456  ",
            filename="  my video.mp4  ",
            file_size=100,
            storage_key="key",
        )
        assert record.user_id == "user456"
        assert record.filename == "my video.mp4"

    def test_empty_user_id_raises(self):
        with pytest.raises(ValueError, match="user_id cannot be empty"):
            DuplicationRecord.create(
                user_id="   ",
                filename="test.mp4",
                file_size=100,
                storage_key="key",
            )

    def test_empty_filename_raises(self):
        with pytest.raises(ValueError, match="filename cannot be empty"):
            DuplicationRecord.create(
                user_id="u1",
                filename="  ",
                file_size=100,
                storage_key="key",
            )

    def test_zero_file_size_raises(self):
        with pytest.raises(ValueError, match="file_size must be positive"):
            DuplicationRecord.create(
                user_id="u1",
                filename="test.mp4",
                file_size=0,
                storage_key="key",
            )

    def test_negative_file_size_raises(self):
        with pytest.raises(ValueError, match="file_size must be positive"):
            DuplicationRecord.create(
                user_id="u1",
                filename="test.mp4",
                file_size=-100,
                storage_key="key",
            )


class TestDuplicationRecordLifecycle:
    """生命周期状态转换测试"""

    def test_mark_processing(self):
        record = DuplicationRecord.create(user_id="u1", filename="t.mp4", file_size=100, storage_key="k")
        old_updated = record.updated_at
        record.mark_processing()
        assert record.status == "processing"
        assert record.updated_at >= old_updated

    def test_mark_completed(self):
        record = DuplicationRecord.create(user_id="u1", filename="t.mp4", file_size=100, storage_key="k")
        record.mark_processing()
        segments = [
            DuplicateSegment.create(
                source_start=0,
                source_end=10,
                matched_video_id="v1",
                matched_video_name="n1",
                matched_start=0,
                matched_end=10,
                similarity=90,
            )
        ]
        record.mark_completed(
            duplicate_rate=25.5,
            duplicate_count=1,
            segments=segments,
        )
        assert record.status == "completed"
        assert record.duplicate_rate == 25.5
        assert record.duplicate_count == 1
        assert len(record.segments) == 1
        assert record.error_message == ""

    def test_mark_completed_zero_rate(self):
        record = DuplicationRecord.create(user_id="u1", filename="t.mp4", file_size=100, storage_key="k")
        record.mark_completed(duplicate_rate=0.0, duplicate_count=0, segments=[])
        assert record.status == "completed"
        assert record.duplicate_rate == 0.0
        assert record.duplicate_count == 0
        assert record.segments == []

    def test_mark_completed_100_rate(self):
        record = DuplicationRecord.create(user_id="u1", filename="t.mp4", file_size=100, storage_key="k")
        record.mark_completed(duplicate_rate=100.0, duplicate_count=5, segments=[])
        assert record.duplicate_rate == 100.0

    def test_mark_completed_invalid_rate_negative(self):
        record = DuplicationRecord.create(user_id="u1", filename="t.mp4", file_size=100, storage_key="k")
        with pytest.raises(ValueError, match="duplicate_rate must be between 0 and 100"):
            record.mark_completed(duplicate_rate=-1, duplicate_count=0, segments=[])

    def test_mark_completed_invalid_rate_over_100(self):
        record = DuplicationRecord.create(user_id="u1", filename="t.mp4", file_size=100, storage_key="k")
        with pytest.raises(ValueError, match="duplicate_rate must be between 0 and 100"):
            record.mark_completed(duplicate_rate=101, duplicate_count=0, segments=[])

    def test_mark_failed(self):
        record = DuplicationRecord.create(user_id="u1", filename="t.mp4", file_size=100, storage_key="k")
        record.mark_processing()
        record.mark_failed("网络超时")
        assert record.status == "failed"
        assert record.error_message == "网络超时"
        assert record.duplicate_rate is None

    def test_mark_failed_from_pending(self):
        record = DuplicationRecord.create(user_id="u1", filename="t.mp4", file_size=100, storage_key="k")
        record.mark_failed("文件损坏")
        assert record.status == "failed"
        assert record.error_message == "文件损坏"


class TestDuplicationRecordRetry:
    """重试逻辑测试"""

    def test_can_retry_failed(self):
        record = DuplicationRecord.create(user_id="u1", filename="t.mp4", file_size=100, storage_key="k")
        record.mark_failed("error")
        assert record.can_retry() is True

    def test_cannot_retry_pending(self):
        record = DuplicationRecord.create(user_id="u1", filename="t.mp4", file_size=100, storage_key="k")
        assert record.can_retry() is False

    def test_cannot_retry_processing(self):
        record = DuplicationRecord.create(user_id="u1", filename="t.mp4", file_size=100, storage_key="k")
        record.mark_processing()
        assert record.can_retry() is False

    def test_cannot_retry_completed(self):
        record = DuplicationRecord.create(user_id="u1", filename="t.mp4", file_size=100, storage_key="k")
        record.mark_completed(duplicate_rate=10, duplicate_count=1, segments=[])
        assert record.can_retry() is False

    def test_reset_for_retry(self):
        record = DuplicationRecord.create(user_id="u1", filename="t.mp4", file_size=100, storage_key="k")
        record.mark_processing()
        segments = [
            DuplicateSegment.create(
                source_start=0,
                source_end=5,
                matched_video_id="v1",
                matched_video_name="n1",
                matched_start=0,
                matched_end=5,
                similarity=80,
            )
        ]
        record.mark_completed(duplicate_rate=30, duplicate_count=1, segments=segments)
        record.status = "failed"
        record.error_message = "something wrong"
        record.video_fingerprint = {"hash": "abc"}

        record.reset_for_retry()
        assert record.status == "pending"
        assert record.duplicate_rate is None
        assert record.duplicate_count == 0
        assert record.error_message == ""
        assert record.segments == []
        assert record.video_fingerprint is None
        assert record.updated_at is not None

    def test_reset_for_retry_from_pending(self):
        """即使从 pending 也能重置（调用方负责判断 can_retry）"""
        record = DuplicationRecord.create(user_id="u1", filename="t.mp4", file_size=100, storage_key="k")
        record.reset_for_retry()
        assert record.status == "pending"
        assert record.duplicate_count == 0