| 1 | from __future__ import annotations |
| 2 | |
| 3 | from typing import Any, Dict, List |
| 4 | |
| 5 | from core.user_modes.base_strategy import BaseUserModeStrategy |
| 6 | from utils.logger import setup_logger |
| 7 | |
| 8 | logger = setup_logger("PostUserModeStrategy") |
| 9 | |
| 10 | |
| 11 | class PostUserModeStrategy(BaseUserModeStrategy): |
| 12 | mode_name = "post" |
| 13 | api_method_name = "get_user_post" |
| 14 | |
| 15 | async def collect_items(self, sec_uid: str, user_info: Dict[str, Any]) -> List[Dict[str, Any]]: |
| 16 | fetcher = getattr(self.downloader.api_client, self.api_method_name, None) |
| 17 | if not callable(fetcher): |
| 18 | logger.error("API client missing get_user_post") |
| 19 | return [] |
| 20 | |
| 21 | aweme_list: List[Dict[str, Any]] = [] |
| 22 | max_cursor = 0 |
| 23 | has_more = True |
| 24 | pagination_restricted = False |
| 25 | |
| 26 | number_limit = int(self.downloader.config.get("number", {}).get(self.mode_name, 0) or 0) |
| 27 | media_filter_enabled = self._media_type_filter_enabled() |
| 28 | |
| 29 | self.downloader._progress_update_step("拉取作品列表", "分页抓取中") |
| 30 | |
| 31 | while has_more: |
| 32 | await self.downloader.rate_limiter.acquire() |
| 33 | request_cursor = max_cursor |
| 34 | page_data = await fetcher(sec_uid, request_cursor, 20) |
| 35 | page = self._normalize_page_data(page_data) |
| 36 | page_items = self.select_items(page) |
| 37 | |
| 38 | if not page_items: |
| 39 | if page.get("status_code") == 0: |
| 40 | pagination_restricted = True |
| 41 | logger.warning( |
| 42 | "User post page empty at cursor=%s (status_code=0); " |
| 43 | "will attempt browser fallback", |
| 44 | request_cursor, |
| 45 | ) |
| 46 | break |
| 47 | |
| 48 | page_items = self._filter_pinned_items(page_items) |
| 49 | aweme_list.extend(page_items) |
| 50 | |
| 51 | self.downloader._progress_update_step("拉取作品列表", f"已抓取 {len(aweme_list)} 条") |
| 52 | |
| 53 | has_more = bool(page.get("has_more", False)) |
| 54 | max_cursor = int(page.get("max_cursor", 0) or 0) |
| 55 | if has_more and max_cursor == request_cursor: |
| 56 | logger.warning( |
| 57 | "max_cursor did not advance (%s), stop paging to avoid loop", |
| 58 | max_cursor, |
| 59 | ) |
| 60 | pagination_restricted = True |
| 61 | break |
| 62 | |
| 63 | if number_limit > 0: |
| 64 | if media_filter_enabled: |
| 65 | if len(self._filter_by_media_type(aweme_list)) >= number_limit: |
| 66 | break |
| 67 | elif len(aweme_list) >= number_limit: |
| 68 | aweme_list = aweme_list[:number_limit] |
| 69 | break |
| 70 | |
| 71 | if pagination_restricted: |
| 72 | self.downloader._progress_update_step("拉取作品列表", "分页受限，尝试浏览器回补") |
| 73 | await self.downloader._recover_user_post_with_browser(sec_uid, user_info, aweme_list) |
| 74 | if not aweme_list: |
| 75 | raise RuntimeError( |
| 76 | "抖音接口未返回作品列表（可能触发了反爬限制），" |
| 77 | "请稍后重试或尝试重新登录抖音刷新 Cookie" |
| 78 | ) |
| 79 | |
| 80 | return aweme_list |
| 81 |