diff --git a/packages/douyin_parser/__init__.py b/packages/douyin_parser/__init__.py index d486e827f..1c762b0ba 100644 --- a/packages/douyin_parser/__init__.py +++ b/packages/douyin_parser/__init__.py @@ -4,7 +4,6 @@ 接口直接返回 aweme_list 包含视频元信息和 play_addr 无水印直链。 此接口不需要任何签名算法、不需要 cookies、不需要特殊 TLS 指纹，稳定性 >95%。 """ - from __future__ import annotations import logging @@ -29,7 +28,34 @@ _APP_UAS = [ _AID = "1128" # Douyin APP aid _FEED_URL = "https://aweme.snssdk.com/aweme/v1/feed/" -_VIDEO_HOST_HINTS = ("douyinvod.com", "bytecdn.com", "365yg.com", "amemv.com", "byteimg.com", "bytedance.com") +# 视频直链 CDN 域名（优先级从高到低）。api-play.amemv.com 是带签名的临时接口，签名过期或 +# 缺失必要 header 会 403/404，因此只作为降级兜底；优先用稳定 CDN。 +_CDN_HOSTS = ("douyinvod.com", "bytecdn.com", "365yg.com", "byteimg.com", "bytedance.com") +_FALLBACK_HOSTS = ("api-play.amemv.com", "api-hl.amemv.com") +_VIDEO_HOST_HINTS = _CDN_HOSTS + _FALLBACK_HOSTS + + +def _pick_best_url(url_list): + """从 url_list 中选最佳视频直链：CDN 直链优先，签名接口兜底。""" + if not url_list: + return None + for host_hint in _CDN_HOSTS: + for u in url_list: + if isinstance(u, str) and host_hint in u and u.endswith(".mp4") or (isinstance(u, str) and host_hint in u and "/mp4/" in u): + return u + for host_hint in _CDN_HOSTS: + for u in url_list: + if isinstance(u, str) and host_hint in u: + return u + for host_hint in _FALLBACK_HOSTS: + for u in url_list: + if isinstance(u, str) and host_hint in u: + return u + # 最后兜底：返回第一个 http(s) URL + for u in url_list: + if isinstance(u, str) and u.startswith("http"): + return u + return None def _extract_aweme_id(url: str) -> Optional[str]: @@ -40,9 +66,7 @@ def _extract_aweme_id(url: str) -> Optional[str]: return None -def fetch_douyin_video_url( - page_url: str, timeout: int = 15, max_retries: int = 3 -) -> Tuple[Optional[str], Optional[str]]: +def fetch_douyin_video_url(page_url: str, timeout: int = 15, max_retries: int = 3) -> Tuple[Optional[str], Optional[str]]: """通过抖音 App Feed API 获取视频无水印直链和文案。 Args: @@ -93,9 +117,7 @@ def fetch_douyin_video_url( continue if len(resp.text) < 100: - logger.warning( - "Feed API 第%d次: 返回内容过短 len=%d body=%s", attempt + 1, len(resp.text), resp.text[:100] - ) + logger.warning("Feed API 第%d次: 返回内容过短 len=%d body=%s", attempt + 1, len(resp.text), resp.text[:100]) time.sleep(0.5 * (attempt + 1)) continue @@ -106,9 +128,7 @@ def fetch_douyin_video_url( status_msg = data.get("status_msg", "") logger.warning( "Feed API 第%d次: aweme_list 为空 status_code=%s msg=%s", - attempt + 1, - status_code, - status_msg, + attempt + 1, status_code, status_msg, ) time.sleep(0.5 * (attempt + 1)) continue @@ -121,28 +141,49 @@ def fetch_douyin_video_url( desc = item.get("desc", "") video = item.get("video") or {} + aweme_type = item.get("aweme_type", 0) - # 提取无水印直链（优先 play_addr_h264 或 play_addr） + # 图文视频 (aweme_type=68): play_addr 通常返回 BGM 的 MP3，不是视频本身。 + # 此时没有可用视频直链，返回 (None, desc) 让调用方仅使用文案。 + images = item.get("images") or [] + if images and not video.get("play_addr_h264", {}).get("url_list"): + logger.info("Feed API 返回图文视频: aweme_id=%s images=%d desc_len=%d（无视频直链，返回文案）", + aweme_id, len(images), len(desc)) + return None, desc + + # 提取无水印直链：优先 download_addr（含 logo 但 CDN 直链稳定），再 play_addr_h264/play_addr play_url = None - for addr_key in ("play_addr_h264", "play_addr", "download_addr", "play_addr_265"): + # 按 key 优先级遍历：download_addr（含水印但CDN直链稳定）→ play_addr_h264 → play_addr → play_addr_lowbr + addr_keys = ("download_addr", "play_addr_h264", "play_addr", "play_addr_lowbr", "play_addr_265") + for addr_key in addr_keys: addr = video.get(addr_key) or {} url_list = addr.get("url_list") or [] - for u in url_list: - if isinstance(u, str) and any(h in u for h in _VIDEO_HOST_HINTS): - play_url = u - break - if play_url: + # 注意：图文视频的 play_addr 里可能是 BGM MP3 而非视频，需过滤 .mp3 + filtered = [u for u in url_list if isinstance(u, str) and not u.endswith(".mp3")] + picked = _pick_best_url(filtered) + if picked: + play_url = picked break + # bit_rate 里的多码率地址作为最后兜底 + if not play_url: + for br_entry in (video.get("bit_rate") or []): + for addr_key in addr_keys: + addr = br_entry.get(addr_key) or {} + url_list = addr.get("url_list") or [] + filtered = [u for u in url_list if isinstance(u, str) and not u.endswith(".mp3")] + picked = _pick_best_url(filtered) + if picked: + play_url = picked + break + if play_url: + break + if play_url: duration = video.get("duration", 0) / 1000 if video.get("duration") else 0 logger.info( - "抖音 Feed API 成功（第%d次）: aweme_id=%s play_len=%d desc_len=%d dur=%.1f", - attempt + 1, - aweme_id, - len(play_url), - len(desc), - duration, + "抖音 Feed API 成功（第%d次）: aweme_id=%s play_len=%d desc_len=%d dur=%.1f type=%d", + attempt + 1, aweme_id, len(play_url), len(desc), duration, aweme_type, ) return play_url, desc @@ -159,12 +200,11 @@ def fetch_douyin_video_url( if __name__ == "__main__": import sys - logging.basicConfig(level=logging.INFO) test_url = sys.argv[1] if len(sys.argv) > 1 else "https://www.douyin.com/video/7661819662322649065" url, desc = fetch_douyin_video_url(test_url) if url: - print("\n✅ SUCCESS!") + print(f"\n✅ SUCCESS!") print(f"desc: {desc}") print(f"play_url: {url[:200]}") else: