| 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) -> Dict[str, Any]: |
| 10 | return { |
| 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 | |
| 18 | |
| 19 | class _FakeConfig: |
| 20 | def __init__(self, data: Dict[str, Any]): |
| 21 | self._data = data |
| 22 | |
| 23 | def get(self, key: str, default: Any = None) -> Any: |
| 24 | return self._data.get(key, default) |
| 25 | |
| 26 | |
| 27 | class _FakeCookieManager: |
| 28 | pass |
| 29 | |
| 30 | |
| 31 | class _NoopRateLimiter: |
| 32 | async def acquire(self): |
| 33 | return |
| 34 | |
| 35 | |
| 36 | class _FakeAPIClient: |
| 37 | def __init__(self): |
| 38 | self.user_info_calls = [] |
| 39 | self.collect_calls = 0 |
| 40 | self.collect_mix_calls = 0 |
| 41 | |
| 42 | async def get_user_info(self, _sec_uid: str): |
| 43 | self.user_info_calls.append(_sec_uid) |
| 44 | return {"uid": "uid-1", "nickname": "tester", "aweme_count": 99} |
| 45 | |
| 46 | async def get_user_post(self, _sec_uid: str, max_cursor: int = 0, count: int = 20): |
| 47 | if max_cursor > 0: |
| 48 | return {"items": [], "has_more": False, "max_cursor": max_cursor, "status_code": 0} |
| 49 | return { |
| 50 | "items": [_make_aweme("111"), _make_aweme("222")], |
| 51 | "has_more": False, |
| 52 | "max_cursor": 0, |
| 53 | "status_code": 0, |
| 54 | } |
| 55 | |
| 56 | async def get_user_like(self, _sec_uid: str, max_cursor: int = 0, count: int = 20): |
| 57 | if max_cursor > 0: |
| 58 | return {"items": [], "has_more": False, "max_cursor": max_cursor, "status_code": 0} |
| 59 | return { |
| 60 | "items": [_make_aweme("222"), _make_aweme("333")], |
| 61 | "has_more": False, |
| 62 | "max_cursor": 0, |
| 63 | "status_code": 0, |
| 64 | } |
| 65 | |
| 66 | async def get_user_mix(self, _sec_uid: str, max_cursor: int = 0, count: int = 20): |
| 67 | return { |
| 68 | "items": [_make_aweme("444")], |
| 69 | "has_more": False, |
| 70 | "max_cursor": 0, |
| 71 | "status_code": 0, |
| 72 | } |
| 73 | |
| 74 | async def get_user_music(self, _sec_uid: str, max_cursor: int = 0, count: int = 20): |
| 75 | return { |
| 76 | "items": [_make_aweme("555")], |
| 77 | "has_more": False, |
| 78 | "max_cursor": 0, |
| 79 | "status_code": 0, |
| 80 | } |
| 81 | |
| 82 | async def get_user_collects(self, _sec_uid: str, max_cursor: int = 0, count: int = 20): |
| 83 | self.collect_calls += 1 |
| 84 | return { |
| 85 | "items": [{"collects_id_str": "collect-1", "collects_name": "默认收藏夹"}], |
| 86 | "has_more": False, |
| 87 | "max_cursor": 0, |
| 88 | "status_code": 0, |
| 89 | } |
| 90 | |
| 91 | async def get_collect_aweme(self, collects_id: str, max_cursor: int = 0, count: int = 20): |
| 92 | assert collects_id == "collect-1" |
| 93 | return { |
| 94 | "items": [_make_aweme("666")], |
| 95 | "has_more": False, |
| 96 | "max_cursor": 0, |
| 97 | "status_code": 0, |
| 98 | } |
| 99 | |
| 100 | |
| 101 | def _build_downloader(tmp_path, mode: List[str]) -> UserDownloader: |
| 102 | config_data = { |
| 103 | "number": {"post": 0, "like": 0, "mix": 0, "music": 0}, |
| 104 | "increase": {"post": False, "like": False, "mix": False, "music": False}, |
| 105 | "mode": mode, |
| 106 | "thread": 2, |
| 107 | "browser_fallback": {"enabled": False}, |
| 108 | } |
| 109 | config = _FakeConfig(config_data) |
| 110 | file_manager = FileManager(str(tmp_path / "Downloaded")) |
| 111 | downloader = UserDownloader( |
| 112 | config=config, |
| 113 | api_client=_FakeAPIClient(), |
| 114 | file_manager=file_manager, |
| 115 | cookie_manager=_FakeCookieManager(), |
| 116 | database=None, |
| 117 | rate_limiter=_NoopRateLimiter(), |
| 118 | retry_handler=None, |
| 119 | queue_manager=QueueManager(max_workers=2), |
| 120 | ) |
| 121 | return downloader |
| 122 | |
| 123 | |
| 124 | def test_user_downloader_processes_modes_and_deduplicates_across_modes(tmp_path, monkeypatch): |
| 125 | downloader = _build_downloader(tmp_path, mode=["post", "like"]) |
| 126 | |
| 127 | async def _always_true(*_args, **_kwargs): |
| 128 | return True |
| 129 | |
| 130 | async def _download_ok(*_args, **_kwargs): |
| 131 | return True |
| 132 | |
| 133 | monkeypatch.setattr(downloader, "_should_download", _always_true) |
| 134 | monkeypatch.setattr(downloader, "_download_aweme_assets", _download_ok) |
| 135 | |
| 136 | result = asyncio.run(downloader.download({"sec_uid": "sec_uid_x"})) |
| 137 | |
| 138 | # 去重后应仅 3 条(111,222,333) |
| 139 | assert result.total == 3 |
| 140 | assert result.success == 3 |
| 141 | assert result.failed == 0 |
| 142 | assert result.skipped == 0 |
| 143 | |
| 144 | |
| 145 | def test_user_downloader_supports_mix_and_music_modes(tmp_path, monkeypatch): |
| 146 | downloader = _build_downloader(tmp_path, mode=["mix", "music"]) |
| 147 | |
| 148 | async def _always_true(*_args, **_kwargs): |
| 149 | return True |
| 150 | |
| 151 | async def _download_ok(*_args, **_kwargs): |
| 152 | return True |
| 153 | |
| 154 | monkeypatch.setattr(downloader, "_should_download", _always_true) |
| 155 | monkeypatch.setattr(downloader, "_download_aweme_assets", _download_ok) |
| 156 | |
| 157 | result = asyncio.run(downloader.download({"sec_uid": "sec_uid_x"})) |
| 158 | |
| 159 | assert result.total == 2 |
| 160 | assert result.success == 2 |
| 161 | |
| 162 | |
| 163 | def test_user_downloader_supports_self_collect_mode(tmp_path, monkeypatch): |
| 164 | downloader = _build_downloader(tmp_path, mode=["collect"]) |
| 165 | |
| 166 | async def _always_true(*_args, **_kwargs): |
| 167 | return True |
| 168 | |
| 169 | async def _download_ok(*_args, **_kwargs): |
| 170 | return True |
| 171 | |
| 172 | monkeypatch.setattr(downloader, "_should_download", _always_true) |
| 173 | monkeypatch.setattr(downloader, "_download_aweme_assets", _download_ok) |
| 174 | |
| 175 | result = asyncio.run(downloader.download({"sec_uid": "self"})) |
| 176 | |
| 177 | assert result.total == 1 |
| 178 | assert result.success == 1 |
| 179 | assert downloader.api_client.user_info_calls == [] |
| 180 | |
| 181 | |
| 182 | def test_user_downloader_rejects_non_self_collect_mode(tmp_path, monkeypatch): |
| 183 | downloader = _build_downloader(tmp_path, mode=["collect"]) |
| 184 | |
| 185 | async def _always_true(*_args, **_kwargs): |
| 186 | return True |
| 187 | |
| 188 | async def _download_ok(*_args, **_kwargs): |
| 189 | return True |
| 190 | |
| 191 | monkeypatch.setattr(downloader, "_should_download", _always_true) |
| 192 | monkeypatch.setattr(downloader, "_download_aweme_assets", _download_ok) |
| 193 | |
| 194 | result = asyncio.run(downloader.download({"sec_uid": "sec_uid_x"})) |
| 195 | |
| 196 | assert result.total == 0 |
| 197 | assert result.success == 0 |
| 198 | assert downloader.api_client.user_info_calls == [] |
| 199 | assert downloader.api_client.collect_calls == 0 |
| 200 | |
| 201 | |
| 202 | def test_user_downloader_post_mode_uses_batch_db_insert(tmp_path, monkeypatch): |
| 203 | """Post-mode should write all aweme records via a single add_aweme_batch |
| 204 | instead of N individual add_aweme commits.""" |
| 205 | from storage.database import Database |
| 206 | |
| 207 | db_path = tmp_path / "test.db" |
| 208 | database = Database(str(db_path)) |
| 209 | asyncio.run(database.initialize()) |
| 210 | |
| 211 | config_data = { |
| 212 | "number": {"post": 0, "like": 0, "mix": 0, "music": 0}, |
| 213 | "increase": {"post": False, "like": False, "mix": False, "music": False}, |
| 214 | "mode": ["post"], |
| 215 | "thread": 2, |
| 216 | "browser_fallback": {"enabled": False}, |
| 217 | } |
| 218 | config = _FakeConfig(config_data) |
| 219 | file_manager = FileManager(str(tmp_path / "Downloaded")) |
| 220 | downloader = UserDownloader( |
| 221 | config=config, |
| 222 | api_client=_FakeAPIClient(), |
| 223 | file_manager=file_manager, |
| 224 | cookie_manager=_FakeCookieManager(), |
| 225 | database=database, |
| 226 | rate_limiter=_NoopRateLimiter(), |
| 227 | retry_handler=None, |
| 228 | queue_manager=QueueManager(max_workers=2), |
| 229 | ) |
| 230 | |
| 231 | add_aweme_calls = {"n": 0} |
| 232 | add_aweme_batch_calls: List[List[Dict[str, Any]]] = [] |
| 233 | |
| 234 | original_add_aweme = database.add_aweme |
| 235 | original_add_aweme_batch = database.add_aweme_batch |
| 236 | |
| 237 | async def counting_add_aweme(record): |
| 238 | add_aweme_calls["n"] += 1 |
| 239 | return await original_add_aweme(record) |
| 240 | |
| 241 | async def counting_add_aweme_batch(records): |
| 242 | add_aweme_batch_calls.append(list(records)) |
| 243 | return await original_add_aweme_batch(records) |
| 244 | |
| 245 | monkeypatch.setattr(database, "add_aweme", counting_add_aweme) |
| 246 | monkeypatch.setattr(database, "add_aweme_batch", counting_add_aweme_batch) |
| 247 | |
| 248 | async def _always_true(*_args, **_kwargs): |
| 249 | return True |
| 250 | |
| 251 | async def _fake_download_aweme_assets(item, _author, *, mode=None, db_batch=None): |
| 252 | if db_batch is not None: |
| 253 | db_batch.append( |
| 254 | { |
| 255 | "aweme_id": item.get("aweme_id"), |
| 256 | "aweme_type": "video", |
| 257 | "title": item.get("desc"), |
| 258 | "author_id": item["author"]["uid"], |
| 259 | "author_name": item["author"]["nickname"], |
| 260 | "create_time": item.get("create_time"), |
| 261 | "file_path": "/tmp", |
| 262 | "metadata": "{}", |
| 263 | } |
| 264 | ) |
| 265 | return True |
| 266 | |
| 267 | monkeypatch.setattr(downloader, "_should_download", _always_true) |
| 268 | monkeypatch.setattr(downloader, "_download_aweme_assets", _fake_download_aweme_assets) |
| 269 | |
| 270 | result = asyncio.run(downloader.download({"sec_uid": "sec_uid_x"})) |
| 271 | |
| 272 | assert result.success == 2 |
| 273 | assert add_aweme_calls["n"] == 0, ( |
| 274 | f"post mode should not call add_aweme per item; got {add_aweme_calls['n']} single inserts" |
| 275 | ) |
| 276 | assert len(add_aweme_batch_calls) == 1 |
| 277 | assert {r["aweme_id"] for r in add_aweme_batch_calls[0]} == {"111", "222"} |
| 278 | |
| 279 | # Verify rows actually landed in the DB. |
| 280 | assert asyncio.run(database.is_downloaded("111")) is True |
| 281 | assert asyncio.run(database.is_downloaded("222")) is True |
| 282 | |
| 283 | asyncio.run(database.close()) |
| 284 | |
| 285 | |
| 286 | def test_user_downloader_rejects_mixed_self_collect_and_regular_modes(tmp_path, monkeypatch): |
| 287 | downloader = _build_downloader(tmp_path, mode=["collect", "post"]) |
| 288 | |
| 289 | async def _always_true(*_args, **_kwargs): |
| 290 | return True |
| 291 | |
| 292 | async def _download_ok(*_args, **_kwargs): |
| 293 | return True |
| 294 | |
| 295 | monkeypatch.setattr(downloader, "_should_download", _always_true) |
| 296 | monkeypatch.setattr(downloader, "_download_aweme_assets", _download_ok) |
| 297 | |
| 298 | result = asyncio.run(downloader.download({"sec_uid": "self"})) |
| 299 | |
| 300 | assert result.total == 0 |
| 301 | assert result.success == 0 |
| 302 | assert downloader.api_client.user_info_calls == [] |
| 303 | assert downloader.api_client.collect_calls == 0 |
| 304 |