Files

377 lines
12 KiB
Python
Raw Permalink Blame History

This file contains ambiguous Unicode characters
This file contains Unicode characters that might be confused with other characters. If you think that this is intentional, you can safely ignore this warning. Use the Escape button to reveal them.
"""duplication 单测.
domain 层查重记录纯逻辑模块,0 外部依赖。
覆盖:DuplicateSegment 工厂/校验、DuplicationRecord 创建/状态流转/重试。
"""
from __future__ import annotations
from packages.domain.duplication import DuplicateSegment, DuplicationRecord
class TestDuplicateSegmentCreate:
"""DuplicateSegment.create 工厂方法测试."""
def test_create_valid(self):
"""正常创建."""
seg = DuplicateSegment.create(
source_start=1.0,
source_end=5.0,
matched_video_id="vid123",
matched_video_name="测试视频",
matched_start=10.0,
matched_end=14.0,
similarity=85.5,
)
assert seg.source_start == 1.0
assert seg.source_end == 5.0
assert seg.matched_video_id == "vid123"
assert seg.matched_video_name == "测试视频"
assert seg.matched_start == 10.0
assert seg.matched_end == 14.0
assert seg.similarity == 85.5
assert isinstance(seg.id, str)
assert len(seg.id) > 0
def test_create_generates_unique_id(self):
"""每次创建生成不同的 id."""
seg1 = DuplicateSegment.create(0, 1, "v", "n", 0, 1, 50.0)
seg2 = DuplicateSegment.create(0, 1, "v", "n", 0, 1, 50.0)
assert seg1.id != seg2.id
def test_create_negative_source_start(self):
"""source_start 为负抛出 ValueError."""
try:
DuplicateSegment.create(-1, 5, "v", "n", 0, 1, 50.0)
assert False, "应该抛出 ValueError"
except ValueError as e:
assert "source" in str(e).lower()
def test_create_source_end_equals_start(self):
"""source_end 等于 source_start 无效."""
try:
DuplicateSegment.create(5, 5, "v", "n", 0, 1, 50.0)
assert False
except ValueError as e:
assert "source" in str(e).lower()
def test_create_source_end_less_than_start(self):
"""source_end 小于 source_start 无效."""
try:
DuplicateSegment.create(5, 3, "v", "n", 0, 1, 50.0)
assert False
except ValueError as e:
assert "source" in str(e).lower()
def test_create_negative_matched_start(self):
"""matched_start 为负无效."""
try:
DuplicateSegment.create(0, 5, "v", "n", -1, 1, 50.0)
assert False
except ValueError as e:
assert "matched" in str(e).lower()
def test_create_matched_end_invalid(self):
"""matched_end <= matched_start 无效."""
try:
DuplicateSegment.create(0, 5, "v", "n", 5, 5, 50.0)
assert False
except ValueError as e:
assert "matched" in str(e).lower()
def test_create_similarity_zero(self):
"""similarity = 0 是合法的."""
seg = DuplicateSegment.create(0, 1, "v", "n", 0, 1, 0.0)
assert seg.similarity == 0.0
def test_create_similarity_100(self):
"""similarity = 100 是合法的."""
seg = DuplicateSegment.create(0, 1, "v", "n", 0, 1, 100.0)
assert seg.similarity == 100.0
def test_create_similarity_negative(self):
"""similarity < 0 无效."""
try:
DuplicateSegment.create(0, 1, "v", "n", 0, 1, -1.0)
assert False
except ValueError as e:
assert "similarity" in str(e).lower()
def test_create_similarity_over_100(self):
"""similarity > 100 无效."""
try:
DuplicateSegment.create(0, 1, "v", "n", 0, 1, 101.0)
assert False
except ValueError as e:
assert "similarity" in str(e).lower()
class TestDuplicationRecordCreate:
"""DuplicationRecord.create 工厂方法测试."""
def test_create_minimal(self):
"""最简创建."""
rec = DuplicationRecord.create(
user_id="user1",
filename="test.mp4",
file_size=1024,
storage_key="oss://bucket/test.mp4",
)
assert rec.user_id == "user1"
assert rec.filename == "test.mp4"
assert rec.file_size == 1024
assert rec.storage_key == "oss://bucket/test.mp4"
assert rec.duration_seconds == 0.0
assert rec.status == "pending"
assert rec.duplicate_rate is None
assert rec.duplicate_count == 0
assert rec.segments == []
assert rec.error_message == ""
assert isinstance(rec.id, str)
assert len(rec.id) > 0
def test_create_with_duration(self):
"""带时长创建."""
rec = DuplicationRecord.create(
user_id="user1",
filename="test.mp4",
file_size=1024,
storage_key="oss://key",
duration_seconds=120.5,
)
assert rec.duration_seconds == 120.5
def test_create_strips_whitespace(self):
"""user_id 和 filename 会 strip."""
rec = DuplicationRecord.create(
user_id=" user1 ",
filename=" test.mp4 ",
file_size=1024,
storage_key="oss://key",
)
assert rec.user_id == "user1"
assert rec.filename == "test.mp4"
def test_create_empty_user_id(self):
"""空 user_id 无效."""
try:
DuplicationRecord.create("", "test.mp4", 1024, "oss://key")
assert False
except ValueError as e:
assert "user_id" in str(e)
def test_create_whitespace_user_id(self):
"""纯空白 user_id 无效."""
try:
DuplicationRecord.create(" ", "test.mp4", 1024, "oss://key")
assert False
except ValueError as e:
assert "user_id" in str(e)
def test_create_empty_filename(self):
"""空 filename 无效."""
try:
DuplicationRecord.create("user1", "", 1024, "oss://key")
assert False
except ValueError as e:
assert "filename" in str(e)
def test_create_whitespace_filename(self):
"""纯空白 filename 无效."""
try:
DuplicationRecord.create("user1", " ", 1024, "oss://key")
assert False
except ValueError as e:
assert "filename" in str(e)
def test_create_zero_file_size(self):
"""file_size = 0 无效."""
try:
DuplicationRecord.create("user1", "test.mp4", 0, "oss://key")
assert False
except ValueError as e:
assert "file_size" in str(e)
def test_create_negative_file_size(self):
"""file_size 为负无效."""
try:
DuplicationRecord.create("user1", "test.mp4", -1, "oss://key")
assert False
except ValueError as e:
assert "file_size" in str(e)
def test_create_unique_id(self):
"""不同记录 id 不同."""
r1 = DuplicationRecord.create("u", "f", 1, "k")
r2 = DuplicationRecord.create("u", "f", 1, "k")
assert r1.id != r2.id
def test_create_has_timestamps(self):
"""有创建和更新时间."""
rec = DuplicationRecord.create("u", "f", 1, "k")
assert rec.created_at is not None
assert rec.updated_at is not None
# 两者应该很接近(都是 now
delta = (rec.updated_at - rec.created_at).total_seconds()
assert abs(delta) < 1.0
class TestDuplicationRecordStatusFlow:
"""状态流转测试."""
def _make_record(self):
return DuplicationRecord.create("user1", "test.mp4", 1024, "oss://key")
def test_initial_status_pending(self):
"""初始状态 pending."""
rec = self._make_record()
assert rec.status == "pending"
def test_mark_processing(self):
"""标记为处理中."""
rec = self._make_record()
old_updated = rec.updated_at
rec.mark_processing()
assert rec.status == "processing"
assert rec.updated_at >= old_updated
def test_mark_completed(self):
"""标记为完成."""
rec = self._make_record()
seg = DuplicateSegment.create(0, 1, "v", "n", 0, 1, 80.0)
rec.mark_completed(duplicate_rate=45.5, duplicate_count=3, segments=[seg])
assert rec.status == "completed"
assert rec.duplicate_rate == 45.5
assert rec.duplicate_count == 3
assert len(rec.segments) == 1
assert rec.segments[0].similarity == 80.0
def test_mark_completed_zero_rate(self):
"""重复率为 0 合法."""
rec = self._make_record()
rec.mark_completed(0.0, 0, [])
assert rec.status == "completed"
assert rec.duplicate_rate == 0.0
assert rec.duplicate_count == 0
assert rec.segments == []
def test_mark_completed_full_rate(self):
"""重复率 100 合法."""
rec = self._make_record()
rec.mark_completed(100.0, 1, [])
assert rec.duplicate_rate == 100.0
def test_mark_completed_negative_rate(self):
"""重复率为负无效."""
rec = self._make_record()
try:
rec.mark_completed(-1, 0, [])
assert False
except ValueError as e:
assert "duplicate_rate" in str(e)
def test_mark_completed_over_100(self):
"""重复率超过 100 无效."""
rec = self._make_record()
try:
rec.mark_completed(101, 0, [])
assert False
except ValueError as e:
assert "duplicate_rate" in str(e)
def test_mark_failed(self):
"""标记为失败."""
rec = self._make_record()
rec.mark_failed("网络超时")
assert rec.status == "failed"
assert rec.error_message == "网络超时"
def test_mark_failed_empty_message(self):
"""失败信息可以为空字符串."""
rec = self._make_record()
rec.mark_failed("")
assert rec.status == "failed"
assert rec.error_message == ""
def test_can_retry_failed(self):
"""failed 状态可以重试."""
rec = self._make_record()
rec.mark_failed("error")
assert rec.can_retry() is True
def test_cannot_retry_pending(self):
"""pending 状态不可重试."""
rec = self._make_record()
assert rec.can_retry() is False
def test_cannot_retry_processing(self):
"""processing 状态不可重试."""
rec = self._make_record()
rec.mark_processing()
assert rec.can_retry() is False
def test_cannot_retry_completed(self):
"""completed 状态不可重试."""
rec = self._make_record()
rec.mark_completed(50, 1, [])
assert rec.can_retry() is False
def test_reset_for_retry(self):
"""重置回 pending."""
rec = self._make_record()
rec.mark_failed("error")
seg = DuplicateSegment.create(0, 1, "v", "n", 0, 1, 50.0)
rec.segments = [seg]
rec.video_fingerprint = {"hash": "abc"}
rec.duplicate_rate = 50.0
rec.duplicate_count = 5
rec.reset_for_retry()
assert rec.status == "pending"
assert rec.duplicate_rate is None
assert rec.duplicate_count == 0
assert rec.error_message == ""
assert rec.segments == []
assert rec.video_fingerprint is None
def test_reset_updates_timestamp(self):
"""重置更新 updated_at."""
rec = self._make_record()
rec.mark_failed("error")
old_updated = rec.updated_at
rec.reset_for_retry()
assert rec.updated_at >= old_updated
class TestDuplicationRecordSegments:
"""segments 列表相关测试."""
def _make_record(self):
return DuplicationRecord.create("user1", "test.mp4", 1024, "oss://key")
def test_segments_default_empty(self):
"""初始 segments 为空列表."""
rec = self._make_record()
assert rec.segments == []
def test_segments_independent_list(self):
"""不同记录的 segments 是独立列表."""
r1 = self._make_record()
r2 = self._make_record()
r1.segments.append("fake")
assert len(r2.segments) == 0
def test_completed_with_multiple_segments(self):
"""完成时带多个片段."""
rec = self._make_record()
segs = [
DuplicateSegment.create(0, 1, "v1", "n1", 0, 1, 90.0),
DuplicateSegment.create(2, 3, "v2", "n2", 5, 6, 70.0),
DuplicateSegment.create(4, 5, "v3", "n3", 10, 11, 85.0),
]
rec.mark_completed(60.0, 3, segs)
assert len(rec.segments) == 3
assert rec.segments[0].similarity == 90.0
assert rec.segments[1].matched_video_id == "v2"
assert rec.segments[2].matched_video_name == "n3"