| 1 | import asyncio |
| 2 | from typing import Any, Dict, List |
| 3 | |
| 4 | from control.queue_manager import QueueManager |
| 5 | from core.user_downloader import UserDownloader |
| 6 | from storage.file_manager import FileManager |
| 7 | |
| 8 | |
| 9 | def _make_aweme(aweme_id: str, **overrides: Any) -> Dict[str, Any]: |
| 10 | aweme = { |
| 11 | "aweme_id": aweme_id, |
| 12 | "desc": f"desc-{aweme_id}", |
| 13 | "create_time": 1700000000, |
| 14 | "author": {"nickname": "tester", "uid": "uid-1"}, |
| 15 | "video": {"play_addr": {"url_list": ["https://example.com/video.mp4"]}}, |
| 16 | } |
| 17 | aweme.update(overrides) |
| 18 | return aweme |
| 19 | |
| 20 | |
| 21 | class _FakeConfig: |
| 22 | def __init__(self, data: Dict[str, Any]): |
| 23 | self._data = data |
| 24 | |
| 25 | def get(self, key: str, default: Any = None) -> Any: |
| 26 | return self._data.get(key, default) |
| 27 | |
| 28 | |
| 29 | class _FakeCookieManager: |
| 30 | pass |
| 31 | |
| 32 | |
| 33 | class _NoopRateLimiter: |
| 34 | async def acquire(self): |
| 35 | return |
| 36 | |
| 37 | |
| 38 | class _FakeProgressReporter: |
| 39 | def __init__(self): |
| 40 | self.step_updates: List[tuple[str, str]] = [] |
| 41 | self.item_totals: List[tuple[int, str]] = [] |
| 42 | self.item_events: List[tuple[str, str]] = [] |
| 43 | |
| 44 | def update_step(self, step: str, detail: str = "") -> None: |
| 45 | self.step_updates.append((step, detail)) |
| 46 | |
| 47 | def set_item_total(self, total: int, detail: str = "") -> None: |
| 48 | self.item_totals.append((total, detail)) |
| 49 | |
| 50 | def advance_item(self, status: str, detail: str = "") -> None: |
| 51 | self.item_events.append((status, detail)) |
| 52 | |
| 53 | |
| 54 | class _FakeAPIClient: |
| 55 | def __init__(self): |
| 56 | self.user_post_calls: List[int] = [] |
| 57 | self.browser_calls = 0 |
| 58 | self.detail_calls: List[str] = [] |
| 59 | self.detail_call_kwargs: List[Dict[str, Any]] = [] |
| 60 | self.browser_call_kwargs: List[Dict[str, Any]] = [] |
| 61 | self.browser_post_items: Dict[str, Dict[str, Any]] = {} |
| 62 | self.browser_post_stats: Dict[str, int] = {} |
| 63 | |
| 64 | async def get_user_post(self, _sec_uid: str, max_cursor: int = 0, _count: int = 20): |
| 65 | self.user_post_calls.append(max_cursor) |
| 66 | if max_cursor == 0: |
| 67 | return { |
| 68 | "status_code": 0, |
| 69 | "aweme_list": [_make_aweme("111")], |
| 70 | "has_more": 1, |
| 71 | "max_cursor": 123, |
| 72 | "not_login_module": {"guide_login_tip_exist": True}, |
| 73 | } |
| 74 | return {"status_code": 0} |
| 75 | |
| 76 | async def collect_user_post_ids_via_browser(self, *_args, **_kwargs): |
| 77 | self.browser_calls += 1 |
| 78 | self.browser_call_kwargs.append(dict(_kwargs)) |
| 79 | return ["111", "222", "333"] |
| 80 | |
| 81 | async def get_video_detail(self, aweme_id: str, **kwargs): |
| 82 | self.detail_calls.append(aweme_id) |
| 83 | self.detail_call_kwargs.append(kwargs) |
| 84 | return _make_aweme(aweme_id) |
| 85 | |
| 86 | def pop_browser_post_aweme_items(self): |
| 87 | data = self.browser_post_items |
| 88 | self.browser_post_items = {} |
| 89 | return data |
| 90 | |
| 91 | def pop_browser_post_stats(self): |
| 92 | data = self.browser_post_stats |
| 93 | self.browser_post_stats = {} |
| 94 | return data |
| 95 | |
| 96 | |
| 97 | def _build_downloader( |
| 98 | tmp_path, |
| 99 | api_client, |
| 100 | browser_enabled: bool, |
| 101 | progress_reporter=None, |
| 102 | number_post: int = 0, |
| 103 | ) -> UserDownloader: |
| 104 | config_data = { |
| 105 | "number": {"post": number_post}, |
| 106 | "increase": {"post": False}, |
| 107 | "mode": ["post"], |
| 108 | "thread": 2, |
| 109 | "browser_fallback": { |
| 110 | "enabled": browser_enabled, |
| 111 | "headless": True, |
| 112 | "max_scrolls": 10, |
| 113 | "idle_rounds": 2, |
| 114 | "wait_timeout_seconds": 5, |
| 115 | }, |
| 116 | } |
| 117 | config = _FakeConfig(config_data) |
| 118 | file_manager = FileManager(str(tmp_path / "Downloaded")) |
| 119 | downloader = UserDownloader( |
| 120 | config=config, |
| 121 | api_client=api_client, |
| 122 | file_manager=file_manager, |
| 123 | cookie_manager=_FakeCookieManager(), |
| 124 | database=None, |
| 125 | rate_limiter=_NoopRateLimiter(), |
| 126 | retry_handler=None, |
| 127 | queue_manager=QueueManager(max_workers=2), |
| 128 | ) |
| 129 | downloader.progress_reporter = progress_reporter |
| 130 | return downloader |
| 131 | |
| 132 | |
| 133 | def test_user_post_browser_fallback_recovers_missing_pages(tmp_path, monkeypatch): |
| 134 | api_client = _FakeAPIClient() |
| 135 | downloader = _build_downloader(tmp_path, api_client, browser_enabled=True) |
| 136 | |
| 137 | async def _always_true(*_args, **_kwargs): |
| 138 | return True |
| 139 | |
| 140 | monkeypatch.setattr(downloader, "_should_download", _always_true) |
| 141 | monkeypatch.setattr(downloader, "_download_aweme_assets", _always_true) |
| 142 | |
| 143 | result = asyncio.run( |
| 144 | downloader._download_user_post( |
| 145 | "sec_uid_x", |
| 146 | {"uid": "uid-1", "nickname": "tester", "aweme_count": 3}, |
| 147 | ) |
| 148 | ) |
| 149 | |
| 150 | assert result.total == 3 |
| 151 | assert result.success == 3 |
| 152 | assert api_client.browser_calls == 1 |
| 153 | assert api_client.browser_call_kwargs[0].get("expected_count") == 0 |
| 154 | assert api_client.detail_calls == ["222", "333"] |
| 155 | assert all(call.get("suppress_error") is True for call in api_client.detail_call_kwargs) |
| 156 | |
| 157 | |
| 158 | def test_user_post_browser_fallback_can_be_disabled(tmp_path, monkeypatch): |
| 159 | api_client = _FakeAPIClient() |
| 160 | downloader = _build_downloader(tmp_path, api_client, browser_enabled=False) |
| 161 | |
| 162 | async def _always_true(*_args, **_kwargs): |
| 163 | return True |
| 164 | |
| 165 | monkeypatch.setattr(downloader, "_should_download", _always_true) |
| 166 | monkeypatch.setattr(downloader, "_download_aweme_assets", _always_true) |
| 167 | |
| 168 | result = asyncio.run( |
| 169 | downloader._download_user_post( |
| 170 | "sec_uid_x", |
| 171 | {"uid": "uid-1", "nickname": "tester", "aweme_count": 3}, |
| 172 | ) |
| 173 | ) |
| 174 | |
| 175 | assert result.total == 1 |
| 176 | assert result.success == 1 |
| 177 | assert api_client.browser_calls == 0 |
| 178 | assert api_client.detail_calls == [] |
| 179 | assert api_client.detail_call_kwargs == [] |
| 180 | |
| 181 | |
| 182 | def test_user_post_browser_fallback_prefers_browser_aweme_items(tmp_path, monkeypatch): |
| 183 | api_client = _FakeAPIClient() |
| 184 | api_client.browser_post_items = { |
| 185 | "222": _make_aweme("222"), |
| 186 | "333": _make_aweme("333"), |
| 187 | } |
| 188 | downloader = _build_downloader(tmp_path, api_client, browser_enabled=True) |
| 189 | |
| 190 | async def _always_true(*_args, **_kwargs): |
| 191 | return True |
| 192 | |
| 193 | monkeypatch.setattr(downloader, "_should_download", _always_true) |
| 194 | monkeypatch.setattr(downloader, "_download_aweme_assets", _always_true) |
| 195 | |
| 196 | result = asyncio.run( |
| 197 | downloader._download_user_post( |
| 198 | "sec_uid_x", |
| 199 | {"uid": "uid-1", "nickname": "tester", "aweme_count": 3}, |
| 200 | ) |
| 201 | ) |
| 202 | |
| 203 | assert result.total == 3 |
| 204 | assert result.success == 3 |
| 205 | assert api_client.detail_calls == [] |
| 206 | |
| 207 | |
| 208 | def test_user_post_browser_fallback_expected_count_uses_number_limit(tmp_path, monkeypatch): |
| 209 | api_client = _FakeAPIClient() |
| 210 | downloader = _build_downloader( |
| 211 | tmp_path, |
| 212 | api_client, |
| 213 | browser_enabled=True, |
| 214 | number_post=2, |
| 215 | ) |
| 216 | |
| 217 | async def _always_true(*_args, **_kwargs): |
| 218 | return True |
| 219 | |
| 220 | monkeypatch.setattr(downloader, "_should_download", _always_true) |
| 221 | monkeypatch.setattr(downloader, "_download_aweme_assets", _always_true) |
| 222 | |
| 223 | result = asyncio.run( |
| 224 | downloader._download_user_post( |
| 225 | "sec_uid_x", |
| 226 | {"uid": "uid-1", "nickname": "tester", "aweme_count": 999}, |
| 227 | ) |
| 228 | ) |
| 229 | |
| 230 | assert result.total == 2 |
| 231 | assert api_client.browser_calls == 1 |
| 232 | assert api_client.browser_call_kwargs[0].get("expected_count") == 2 |
| 233 | |
| 234 | |
| 235 | def test_user_post_skips_pinned_before_number_limit(tmp_path, monkeypatch): |
| 236 | class _PinnedAPIClient(_FakeAPIClient): |
| 237 | async def get_user_post(self, _sec_uid: str, max_cursor: int = 0, _count: int = 20): |
| 238 | self.user_post_calls.append(max_cursor) |
| 239 | if max_cursor == 0: |
| 240 | return { |
| 241 | "status_code": 0, |
| 242 | "aweme_list": [ |
| 243 | _make_aweme("111", is_top=1), |
| 244 | _make_aweme("222", is_top=1), |
| 245 | _make_aweme("333", is_top=0), |
| 246 | ], |
| 247 | "has_more": 1, |
| 248 | "max_cursor": 456, |
| 249 | } |
| 250 | return { |
| 251 | "status_code": 0, |
| 252 | "aweme_list": [_make_aweme("444", is_top=0)], |
| 253 | "has_more": 0, |
| 254 | "max_cursor": max_cursor, |
| 255 | } |
| 256 | |
| 257 | api_client = _PinnedAPIClient() |
| 258 | downloader = _build_downloader( |
| 259 | tmp_path, |
| 260 | api_client, |
| 261 | browser_enabled=False, |
| 262 | number_post=2, |
| 263 | ) |
| 264 | |
| 265 | downloaded_ids: List[str] = [] |
| 266 | |
| 267 | async def _always_true(*_args, **_kwargs): |
| 268 | return True |
| 269 | |
| 270 | async def _download_aweme_assets(item, *_args, **_kwargs): |
| 271 | downloaded_ids.append(str(item.get("aweme_id"))) |
| 272 | return True |
| 273 | |
| 274 | monkeypatch.setattr(downloader, "_should_download", _always_true) |
| 275 | monkeypatch.setattr(downloader, "_download_aweme_assets", _download_aweme_assets) |
| 276 | |
| 277 | result = asyncio.run( |
| 278 | downloader._download_user_post( |
| 279 | "sec_uid_x", |
| 280 | {"uid": "uid-1", "nickname": "tester", "aweme_count": 4}, |
| 281 | ) |
| 282 | ) |
| 283 | |
| 284 | assert result.total == 2 |
| 285 | assert result.success == 2 |
| 286 | assert downloaded_ids == ["333", "444"] |
| 287 | |
| 288 | |
| 289 | def test_user_post_reports_step_and_item_progress(tmp_path, monkeypatch): |
| 290 | api_client = _FakeAPIClient() |
| 291 | reporter = _FakeProgressReporter() |
| 292 | downloader = _build_downloader( |
| 293 | tmp_path, |
| 294 | api_client, |
| 295 | browser_enabled=True, |
| 296 | progress_reporter=reporter, |
| 297 | ) |
| 298 | |
| 299 | async def _fake_should_download(aweme_id): |
| 300 | return aweme_id != "222" |
| 301 | |
| 302 | async def _fake_download_aweme_assets(item, *_args, **_kwargs): |
| 303 | return item.get("aweme_id") != "333" |
| 304 | |
| 305 | monkeypatch.setattr(downloader, "_should_download", _fake_should_download) |
| 306 | monkeypatch.setattr(downloader, "_download_aweme_assets", _fake_download_aweme_assets) |
| 307 | |
| 308 | result = asyncio.run( |
| 309 | downloader._download_user_post( |
| 310 | "sec_uid_x", |
| 311 | {"uid": "uid-1", "nickname": "tester", "aweme_count": 3}, |
| 312 | ) |
| 313 | ) |
| 314 | |
| 315 | assert result.total == 3 |
| 316 | assert result.success == 1 |
| 317 | assert result.skipped == 1 |
| 318 | assert result.failed == 1 |
| 319 | assert reporter.item_totals == [(3, "作品待下载")] |
| 320 | assert ("下载作品", "待处理 3 条") in reporter.step_updates |
| 321 | statuses = [status for status, _detail in reporter.item_events] |
| 322 | assert statuses.count("success") == 1 |
| 323 | assert statuses.count("skipped") == 1 |
| 324 | assert statuses.count("failed") == 1 |
| 325 |