diff --git a/app/services/grok/services/video.py b/app/services/grok/services/video.py index 6b1feb9..c25d204 100644 --- a/app/services/grok/services/video.py +++ b/app/services/grok/services/video.py @@ -135,6 +135,41 @@ def _log_final_video_payload( logger.info(f"Video upstream payload before send:\n{payload_text}") +def _log_raw_video_stream_event(raw_text: str) -> None: + text = str(raw_text or "").strip() + if not text: + return + try: + payload = orjson.loads(text) + pretty = orjson.dumps(payload, option=orjson.OPT_INDENT_2).decode("utf-8") + except Exception: + pretty = text + logger.info(f"Video upstream raw event:\n{pretty}") + + +def _truncate_video_stream_line(raw_text: str, limit: int = 4000) -> str: + text = str(raw_text or "") + if len(text) <= limit: + return text + return f"{text[:limit]}...(len={len(text)})" + + +def _log_video_stream_line(*, stage: str, raw_text: str) -> None: + text = str(raw_text or "") + if not text: + logger.info(f"Video upstream {stage}: ") + return + logger.info( + "Video upstream line " + f"({stage}): {_truncate_video_stream_line(text)}" + ) + + +def _log_video_stream_end(*, stage: str, reason: str, extra: str = "") -> None: + suffix = f", {extra}" if extra else "" + logger.info(f"Video upstream stream ended ({stage}): reason={reason}{suffix}") + + def _classify_video_error(exc: Exception) -> tuple[str, str, int]: """将底层异常归一化为用户可读错误。""" text = str(exc or "").lower() @@ -175,12 +210,12 @@ async def _try_log_video_share_link( *, local_url: str = "", thumbnail_url: str = "", -) -> None: +) -> str: """视频生成完成后尝试创建分享链接并写入元数据。""" token_text = str(token or "").strip() post_text = str(post_id or "").strip() if not token_text or not post_text: - return + return "" try: logger.info(f"Video create-link attempt: post_id={post_text}") async with AsyncSession() as session: @@ -204,12 +239,47 @@ async def _try_log_video_share_link( "Video create-link completed without shareLink: " f"post_id={post_text}, metadata_path={metadata_path or '-'}" ) + return share_link except Exception as e: details = getattr(e, "details", None) logger.warning( "Video create-link failed: " f"post_id={post_text}, error={e}, details={details}" ) + return "" + + +async def _fetch_share_page_media_urls(share_link: str) -> tuple[str, str]: + share_text = str(share_link or "").strip() + if not share_text: + return "", "" + try: + async with AsyncSession() as session: + response = await session.get(share_text, timeout=get_config("video.timeout")) + html = str(getattr(response, "text", "") or "") + if not html: + return "", "" + video_match = re.search( + r' str: + post_text = str(post_id or "").strip() + if not post_text or not self.token: + return "" + logger.info( + "Video stream high-progress fallback: " + f"try create-link/media-post lookup, post_id={post_text}, progress={progress}" + ) + share_link = await _try_log_video_share_link( + self.token, + post_text, + thumbnail_url=thumbnail_url, + ) + if share_link: + share_video_url, share_thumb_url = await _fetch_share_page_media_urls(share_link) + if share_video_url: + render_started_at = time.perf_counter() + logger.info( + "Video stream high-progress share-link render started: " + f"post_id={post_text}, media_url={share_video_url}, thumbnail_url={share_thumb_url or thumbnail_url or '-'}" + ) + rendered = await self._get_dl().render_video( + share_video_url, + self.token, + share_thumb_url or thumbnail_url, + ) + render_duration_ms = (time.perf_counter() - render_started_at) * 1000 + logger.info( + "Video stream high-progress share-link render completed: " + f"post_id={post_text}, duration_ms={render_duration_ms:.2f}, rendered={rendered}" + ) + return rendered + post = await _fetch_media_post_info(self.token, post_text) + media_url = _normalize_assets_url(str(post.get("mediaUrl") or "").strip()) + thumb_url = _normalize_assets_url( + str(post.get("thumbnailImageUrl") or thumbnail_url or "").strip() + ) + if not media_url: + logger.info( + "Video stream high-progress fallback miss: " + f"post_id={post_text}, progress={progress}" + ) + return "" + render_started_at = time.perf_counter() + logger.info( + "Video stream high-progress fallback render started: " + f"post_id={post_text}, media_url={media_url}, thumbnail_url={thumb_url or '-'}" + ) + rendered = await self._get_dl().render_video(media_url, self.token, thumb_url) + render_duration_ms = (time.perf_counter() - render_started_at) * 1000 + logger.info( + "Video stream high-progress fallback render completed: " + f"post_id={post_text}, duration_ms={render_duration_ms:.2f}, rendered={rendered}" + ) + return rendered + + async def _try_render_from_post_with_retry( + self, + *, + post_id: str, + progress: int = 0, + thumbnail_url: str = "", + retry_delay_seconds: int = 30, + ) -> str: + rendered = await self._try_render_from_post( + post_id=post_id, + progress=progress, + thumbnail_url=thumbnail_url, + ) + if rendered: + return rendered + logger.info( + "Video stream high-progress fallback retry scheduled: " + f"post_id={post_id}, progress={progress}, delay={retry_delay_seconds}s" + ) + await asyncio.sleep(max(1, int(retry_delay_seconds))) + return await self._try_render_from_post( + post_id=post_id, + progress=progress, + thumbnail_url=thumbnail_url, + ) + + async def _wait_before_eof_fallback( + self, + *, + post_id: str, + progress: int, + wait_seconds: int = 60, + ) -> None: + delay = max(1, int(wait_seconds)) + logger.info( + "Video stream EOF fallback wait scheduled: " + f"post_id={post_id}, progress={progress}, delay={delay}s" + ) + await asyncio.sleep(delay) + def _sse(self, content: str = "", role: str = None, finish: str = None) -> str: """Build SSE response.""" delta = {} @@ -1401,16 +1567,32 @@ class VideoStreamProcessor(BaseProcessor): ) -> AsyncGenerator[str, None]: """Process video stream response.""" idle_timeout = self.idle_timeout_override or get_config("video.stream_timeout") + latest_progress = 0 + latest_video_post_id = "" + latest_thumbnail_url = "" + final_rendered = False try: async for line in _with_idle_timeout(response, idle_timeout, self.model): + raw_line = ( + line.decode("utf-8", errors="ignore") + if isinstance(line, (bytes, bytearray)) + else str(line) + ) + _log_video_stream_line(stage="stream/raw", raw_text=raw_line) line = _normalize_line(line) if not line: + _log_video_stream_line( + stage="stream/normalized", raw_text="" + ) continue + _log_video_stream_line(stage="stream/normalized", raw_text=line) try: data = orjson.loads(line) except orjson.JSONDecodeError: + _log_video_stream_line(stage="stream/non-json", raw_text=line) continue + _log_raw_video_stream_event(line) resp = data.get("result", {}).get("response", {}) is_thinking = bool(resp.get("isThinking")) @@ -1438,6 +1620,17 @@ class VideoStreamProcessor(BaseProcessor): if video_resp := resp.get("streamingVideoGenerationResponse"): progress = video_resp.get("progress", 0) + latest_progress = max(latest_progress, int(progress or 0)) + latest_video_post_id = str( + video_resp.get("videoPostId") + or video_resp.get("assetId") + or video_resp.get("videoId") + or latest_video_post_id + ).strip() + latest_thumbnail_url = str( + video_resp.get("thumbnailImageUrl") + or latest_thumbnail_url + ).strip() if is_thinking: if not self.show_think: @@ -1481,6 +1674,7 @@ class VideoStreamProcessor(BaseProcessor): f"post_id={video_post_id or '-'}, duration_ms={render_duration_ms:.2f}" ) yield self._sse(rendered) + final_rendered = True logger.info(f"Video generated: {video_url} (post_id={video_post_id})") if video_post_id and self.token: @@ -1492,15 +1686,73 @@ class VideoStreamProcessor(BaseProcessor): ) continue + if (not final_rendered) and latest_progress >= 90 and latest_video_post_id: + await self._wait_before_eof_fallback( + post_id=latest_video_post_id, + progress=latest_progress, + wait_seconds=60, + ) + rendered = await self._try_render_from_post_with_retry( + post_id=latest_video_post_id, + progress=latest_progress, + thumbnail_url=latest_thumbnail_url, + ) + if rendered: + logger.info( + "Video generated via high-progress fallback: " + f"post_id={latest_video_post_id}, progress={latest_progress}" + ) + if latest_video_post_id and self.token: + await _try_log_video_share_link( + self.token, + latest_video_post_id, + local_url=rendered, + thumbnail_url=latest_thumbnail_url, + ) + yield self._sse(rendered) + final_rendered = True + + _log_video_stream_end(stage="stream", reason="upstream_eof") if self.think_opened: yield self._sse("\n") yield self._sse(finish="stop") yield "data: [DONE]\n\n" except asyncio.CancelledError: + _log_video_stream_end(stage="stream", reason="cancelled") logger.debug( "Video stream cancelled by client", extra={"model": self.model} ) except StreamIdleTimeoutError as e: + if (not final_rendered) and latest_progress >= 90 and latest_video_post_id: + rendered = await self._try_render_from_post_with_retry( + post_id=latest_video_post_id, + progress=latest_progress, + thumbnail_url=latest_thumbnail_url, + ) + if rendered: + if self.think_opened: + yield self._sse("\n") + self.think_opened = False + logger.info( + "Video generated via high-progress timeout fallback: " + f"post_id={latest_video_post_id}, progress={latest_progress}" + ) + if latest_video_post_id and self.token: + await _try_log_video_share_link( + self.token, + latest_video_post_id, + local_url=rendered, + thumbnail_url=latest_thumbnail_url, + ) + yield self._sse(rendered) + yield self._sse(finish="stop") + yield "data: [DONE]\n\n" + return + _log_video_stream_end( + stage="stream", + reason="idle_timeout", + extra=f"timeout={idle_timeout}", + ) raise AppException( message="视频生成失败：网络连接异常，请稍后重试", error_type=ErrorType.SERVER.value, @@ -1508,6 +1760,11 @@ class VideoStreamProcessor(BaseProcessor): status_code=504, ) except RequestsError as e: + _log_video_stream_end( + stage="stream", + reason="request_error", + extra=f"error={e}", + ) if _is_http2_error(e): logger.warning( f"HTTP/2 stream error in video: {e}", extra={"model": self.model} @@ -1528,6 +1785,11 @@ class VideoStreamProcessor(BaseProcessor): status_code=502, ) except Exception as e: + _log_video_stream_end( + stage="stream", + reason="exception", + extra=f"type={type(e).__name__}, error={e}", + ) logger.error( f"Video stream processing error: {e}", extra={"model": self.model, "error_type": type(e).__name__}, @@ -1671,13 +1933,25 @@ class VideoCollectProcessor(BaseProcessor): try: async for line in _with_idle_timeout(response, idle_timeout, self.model): + raw_line = ( + line.decode("utf-8", errors="ignore") + if isinstance(line, (bytes, bytearray)) + else str(line) + ) + _log_video_stream_line(stage="collect/raw", raw_text=raw_line) line = _normalize_line(line) if not line: + _log_video_stream_line( + stage="collect/normalized", raw_text="" + ) continue + _log_video_stream_line(stage="collect/normalized", raw_text=line) try: data = orjson.loads(line) except orjson.JSONDecodeError: + _log_video_stream_line(stage="collect/non-json", raw_text=line) continue + _log_raw_video_stream_event(line) resp = data.get("result", {}).get("response", {}) @@ -1735,15 +2009,31 @@ class VideoCollectProcessor(BaseProcessor): fallback_video_id = fid break + _log_video_stream_end( + stage="collect", + reason="upstream_eof", + extra=f"has_content={bool(content)}, fallback_video_id={fallback_video_id or '-'}", + ) except asyncio.CancelledError: + _log_video_stream_end(stage="collect", reason="cancelled") logger.debug( "Video collect cancelled by client", extra={"model": self.model} ) except StreamIdleTimeoutError as e: + _log_video_stream_end( + stage="collect", + reason="idle_timeout", + extra=f"timeout={idle_timeout}, fallback_video_id={fallback_video_id or '-'}", + ) logger.warning( f"Video collect idle timeout: {e}", extra={"model": self.model} ) except RequestsError as e: + _log_video_stream_end( + stage="collect", + reason="request_error", + extra=f"error={e}, fallback_video_id={fallback_video_id or '-'}", + ) if _is_http2_error(e): logger.warning( f"HTTP/2 stream error in video collect: {e}", @@ -1766,11 +2056,21 @@ class VideoCollectProcessor(BaseProcessor): extra={"model": self.model}, ) raise + _log_video_stream_end( + stage="collect", + reason="upstream_exception", + extra=f"error={e}, fallback_video_id={fallback_video_id or '-'}", + ) logger.error( f"Video collect upstream error: {e}", extra={"model": self.model, "error_type": type(e).__name__}, ) except Exception as e: + _log_video_stream_end( + stage="collect", + reason="exception", + extra=f"type={type(e).__name__}, error={e}, fallback_video_id={fallback_video_id or '-'}", + ) logger.error( f"Video collect processing error: {e}", extra={"model": self.model, "error_type": type(e).__name__}, diff --git a/app/services/grok/utils/cache.py b/app/services/grok/utils/cache.py index a728df1..a9b9d58 100644 --- a/app/services/grok/utils/cache.py +++ b/app/services/grok/utils/cache.py @@ -2,6 +2,8 @@ Local cache utilities. """ +import json +import re from typing import Any, Dict from app.core.storage import DATA_DIR @@ -26,6 +28,47 @@ class CacheService: def _allowed_exts(self, media_type: str): return IMAGE_EXTS if media_type == "image" else VIDEO_EXTS + def _media_meta_dir(self): + return DATA_DIR / "tmp" / "media-meta" + + def _load_video_metadata(self) -> dict[str, dict[str, Any]]: + meta_dir = self._media_meta_dir() + if not meta_dir.exists(): + return {} + + metadata_by_post_id: dict[str, dict[str, Any]] = {} + for file_path in meta_dir.glob("*.json"): + try: + payload = json.loads(file_path.read_text(encoding="utf-8")) + except Exception: + continue + if not isinstance(payload, dict): + continue + if str(payload.get("media_type") or "").strip().lower() != "video": + continue + post_id = str(payload.get("post_id") or "").strip() + if not post_id: + continue + metadata_by_post_id[post_id] = payload + return metadata_by_post_id + + @staticmethod + def _extract_post_id_from_name(name: str) -> str: + text = str(name or "").strip() + if not text: + return "" + match = re.search( + r"generated-([0-9a-fA-F]{8}-[0-9a-fA-F]{4}-[0-9a-fA-F]{4}-[0-9a-fA-F]{4}-[0-9a-fA-F]{12})-", + text, + ) + if match: + return match.group(1) + all_matches = re.findall( + r"[0-9a-fA-F]{8}-[0-9a-fA-F]{4}-[0-9a-fA-F]{4}-[0-9a-fA-F]{4}-[0-9a-fA-F]{12}", + text, + ) + return all_matches[-1] if all_matches else "" + def get_stats(self, media_type: str = "image") -> Dict[str, Any]: cache_dir = self._cache_dir(media_type) if not cache_dir.exists(): @@ -70,8 +113,24 @@ class CacheService: start = max(0, (page - 1) * page_size) paged = items[start : start + page_size] + metadata_by_post_id = self._load_video_metadata() if media_type == "video" else {} + for item in paged: item["view_url"] = f"/v1/files/{media_type}/{item['name']}" + if media_type != "video": + continue + post_id = self._extract_post_id_from_name(item["name"]) + if not post_id: + continue + metadata = metadata_by_post_id.get(post_id) + if not metadata: + continue + item["post_id"] = str(metadata.get("post_id") or "").strip() + item["share_link"] = str(metadata.get("share_link") or "").strip() + item["original_post_id"] = str(metadata.get("original_post_id") or "").strip() + item["media_url"] = str(metadata.get("media_url") or "").strip() + item["thumbnail_url"] = str(metadata.get("thumbnail_url") or "").strip() + item["local_url"] = str(metadata.get("local_url") or "").strip() return {"total": total, "page": page, "page_size": page_size, "items": paged} diff --git a/app/services/grok/utils/download.py b/app/services/grok/utils/download.py index 8a33e6e..9647319 100644 --- a/app/services/grok/utils/download.py +++ b/app/services/grok/utils/download.py @@ -9,6 +9,7 @@ import base64 import hashlib import os import time +import urllib.request from pathlib import Path from typing import List, Optional, Tuple from urllib.parse import urlparse @@ -21,6 +22,7 @@ from app.core.storage import DATA_DIR from app.core.config import get_config from app.core.exceptions import AppException from app.services.reverse.assets_download import AssetsDownloadReverse +from app.services.reverse.utils.headers import build_headers from app.services.grok.utils.locks import _get_download_semaphore, _file_lock @@ -51,6 +53,14 @@ class DownloadService: async def resolve_url( self, path_or_url: str, token: str, media_type: str = "image" ) -> str: + if self._is_public_share_url(path_or_url): + app_url = get_config("app.app_url") + filename = self._public_cache_filename(path_or_url, media_type) + if app_url: + await self.download_file(path_or_url, token, media_type) + return f"{app_url.rstrip('/')}/v1/files/{media_type}/{filename}" + return path_or_url + asset_url = path_or_url path = path_or_url if path_or_url.startswith("http"): @@ -69,6 +79,120 @@ class DownloadService: return f"{app_url.rstrip('/')}/v1/files/{media_type}{path}" return asset_url + @staticmethod + def _is_public_share_url(url: str) -> bool: + text = str(url or "").strip().lower() + return ( + text.startswith("https://imagine-public.x.ai/") + or text.startswith("http://imagine-public.x.ai/") + ) + + @staticmethod + def _is_localhost_url(url: str) -> bool: + parsed = urlparse(str(url or "").strip()) + host = (parsed.hostname or "").lower() + return host in {"localhost", "127.0.0.1", "::1"} + + @staticmethod + def _public_cache_filename(file_url: str, media_type: str = "image") -> str: + parsed = urlparse(str(file_url or "").strip()) + suffix = Path(parsed.path or "").suffix + filename = ( + (parsed.netloc + (parsed.path or "")) + .lstrip("/") + .replace("/", "-") + .replace(":", "-") + ) + if not filename: + filename = f"public-{media_type}" + if suffix and not filename.endswith(suffix): + filename = f"{filename}{suffix}" + return filename + + async def _download_public_url( + self, + file_url: str, + media_type: str = "image", + ) -> Tuple[Optional[Path], str]: + """下载公开分享直链，避免误走 assets.grok.com 反代。""" + started_at = time.perf_counter() + cache_dir = self.image_dir if media_type == "image" else self.video_dir + filename = self._public_cache_filename(file_url, media_type) + cache_path = cache_dir / filename + logger.info( + f"Public download started: media_type={media_type}, file_url={file_url}, cache_path={cache_path}" + ) + + lock_name = ( + f"dl_public_{media_type}_{hashlib.sha1(str(cache_path).encode()).hexdigest()[:16]}" + ) + lock_timeout = max(1, int(get_config("asset.download_timeout"))) + async with _file_lock(lock_name, timeout=lock_timeout): + base_proxy = (get_config("proxy.base_proxy_url") or "").strip() + asset_proxy = (get_config("proxy.asset_proxy_url") or "").strip() + proxy_url = asset_proxy or base_proxy + headers = build_headers( + cookie_token="", + content_type="video/mp4" if media_type == "video" else "image/jpeg", + origin="https://grok.com", + referer="https://grok.com/", + ) + if media_type == "video": + headers["Range"] = "bytes=0-" + + def _urllib_download(): + opener = ( + urllib.request.build_opener( + urllib.request.ProxyHandler( + {"http": proxy_url, "https": proxy_url} + ) + ) + if proxy_url + else urllib.request.build_opener() + ) + req = urllib.request.Request(file_url, headers=headers, method="GET") + with opener.open(req, timeout=get_config("asset.download_timeout")) as resp: + status = int(getattr(resp, "status", 200) or 200) + body = resp.read() + raw_headers = {str(k).lower(): str(v) for k, v in dict(resp.headers.items()).items()} + return status, raw_headers, body + + status_code, response_headers, response_body = await asyncio.to_thread( + _urllib_download + ) + if status_code not in (200, 206): + raise AppException( + f"Public download failed, status={status_code}", + code="public_download_failed", + ) + logger.info( + f"Public download response received: media_type={media_type}, file_url={file_url}, " + f"status={status_code}, content_type={(response_headers.get('content-type', 'application/octet-stream').split(';')[0])}" + ) + + tmp_path = cache_path.with_suffix(cache_path.suffix + ".tmp") + try: + async with aiofiles.open(tmp_path, "wb") as f: + await f.write(response_body) + os.replace(tmp_path, cache_path) + finally: + if tmp_path.exists() and not cache_path.exists(): + try: + tmp_path.unlink() + except Exception: + pass + + mime = response_headers.get( + "content-type", "application/octet-stream" + ).split(";")[0] + duration_ms = (time.perf_counter() - started_at) * 1000 + logger.info( + f"Public downloaded: {file_url}, media_type={media_type}, cache_path={cache_path}, " + f"mime={mime}, duration_ms={duration_ms:.2f}" + ) + asyncio.create_task(self._check_limit()) + return cache_path, mime + async def render_image( self, url: str, token: str, image_id: str = "image" ) -> str: @@ -101,7 +225,11 @@ class DownloadService: final_video_url = await self.resolve_url(video_url, token, "video") final_thumb_url = "" if thumbnail_url: - final_thumb_url = await self.resolve_url(thumbnail_url, token, "image") + try: + final_thumb_url = await self.resolve_url(thumbnail_url, token, "image") + except Exception as e: + logger.warning(f"Render video thumbnail resolve failed, ignore thumbnail: {e}") + final_thumb_url = "" duration_ms = (time.perf_counter() - started_at) * 1000 logger.info( f"Render video resolved: final_video_url={final_video_url}, " @@ -196,6 +324,9 @@ class DownloadService: Returns: Tuple[Optional[Path], str]: The path of the downloaded file and the MIME type. """ + if self._is_public_share_url(file_path): + return await self._download_public_url(file_path, media_type) + started_at = time.perf_counter() async with _get_download_semaphore(): file_path = self._normalize_path(file_path) diff --git a/app/services/reverse/assets_download.py b/app/services/reverse/assets_download.py index ce214b0..b312e58 100644 --- a/app/services/reverse/assets_download.py +++ b/app/services/reverse/assets_download.py @@ -44,12 +44,26 @@ class AssetsDownloadReverse: yield data[i:i + chunk_size] @staticmethod - async def _urllib_get(url: str, headers: dict[str, str], timeout: int) -> "AssetsDownloadReverse._SimpleResponse": + async def _urllib_get( + url: str, + headers: dict[str, str], + timeout: int, + proxy_url: str = "", + ) -> "AssetsDownloadReverse._SimpleResponse": """使用标准库 urllib 兜底下载，绕过 curl TLS 实现。""" req = urllib.request.Request(url=url, headers=headers, method="GET") def _do_get(): - with urllib.request.urlopen(req, timeout=timeout) as resp: + opener = None + if proxy_url: + proxy_handler = urllib.request.ProxyHandler( + {"http": proxy_url, "https": proxy_url} + ) + opener = urllib.request.build_opener(proxy_handler) + else: + opener = urllib.request.build_opener() + + with opener.open(req, timeout=timeout) as resp: status = int(getattr(resp, "status", 200) or 200) body = resp.read() raw_headers = {} @@ -137,13 +151,13 @@ class AssetsDownloadReverse: try: return await _single_get(use_impersonate=True, use_proxy=True) except Exception as first_err: - # TLS/握手类问题常出现在代理或浏览器指纹组合，降级直连重试一次。 + # TLS/握手类问题常出现在浏览器指纹组合，降级为不伪装浏览器后重试。 logger.warning( "AssetsDownloadReverse primary request failed, fallback direct: " f"error={first_err}" ) try: - return await _single_get(use_impersonate=False, use_proxy=False) + return await _single_get(use_impersonate=False, use_proxy=True) except Exception as second_err: # curl 栈仍失败时，使用标准库 urllib 再兜底一次，绕开 curl TLS 实现。 logger.warning( @@ -154,6 +168,7 @@ class AssetsDownloadReverse: url=url, headers=headers, timeout=timeout, + proxy_url=proxy_url, ) if fallback_resp.status_code != 200: logger.error( diff --git a/app/services/reverse/media_post.py b/app/services/reverse/media_post.py index 07392ad..987e188 100644 --- a/app/services/reverse/media_post.py +++ b/app/services/reverse/media_post.py @@ -4,6 +4,7 @@ Reverse interface: media post create/get. import asyncio import json +import re import urllib.request from pathlib import Path from dataclasses import dataclass @@ -87,9 +88,15 @@ class MediaPostReverse: except Exception as e: logger.warning(f"MediaPost metadata create-link failed: post_id={post_text}, error={e}") + canonical_post_id = post_text + if share_link: + match = re.search(r"/imagine/post/([0-9a-fA-F-]{32,36})", share_link) + if match: + canonical_post_id = match.group(1) + post = post_payload.get("post", {}) if isinstance(post_payload, dict) else {} metadata: dict[str, Any] = { - "post_id": post_text, + "post_id": canonical_post_id, "share_link": share_link, "media_type": str(media_type or post.get("mediaType") or "").strip(), "media_url": str(post.get("mediaUrl") or "").strip(), @@ -103,7 +110,7 @@ class MediaPostReverse: } if extra: metadata.update(extra) - path = await MediaPostReverse.write_metadata(post_text, metadata) + path = await MediaPostReverse.write_metadata(canonical_post_id, metadata) if path: metadata["metadata_path"] = str(path) return metadata diff --git a/app/static/public/js/video.js b/app/static/public/js/video.js index 9eb18fa..e2caed9 100644 --- a/app/static/public/js/video.js +++ b/app/static/public/js/video.js @@ -209,19 +209,59 @@ if (editExtendPostId) editExtendPostId.textContent = shortHash(currentExtendPostId); } + const UUID_RE = /[0-9a-fA-F]{8}-[0-9a-fA-F]{4}-[0-9a-fA-F]{4}-[0-9a-fA-F]{4}-[0-9a-fA-F]{12}/g; + // 从缓存视频文件名中提取 parentPostId // 文件名格式示例: users-xxx-generated-{postId}-generated_video_hd.mp4 function extractPostIdFromFileName(name) { const s = String(name || '').trim(); if (!s) return ''; // 尝试 generated-{uuid}- 模式 - const m = s.match(/generated-([0-9a-fA-F-]{32,36})-/); + const m = s.match(/generated-([0-9a-fA-F]{8}-[0-9a-fA-F]{4}-[0-9a-fA-F]{4}-[0-9a-fA-F]{4}-[0-9a-fA-F]{12})-/); if (m) return m[1]; // 回退：匹配最后一个 UUID 格式 - const allUuids = s.match(/[0-9a-fA-F-]{32,36}/g); + const allUuids = s.match(UUID_RE); return allUuids && allUuids.length ? allUuids[allUuids.length - 1] : ''; } + function extractPostIdFromShareLink(link) { + const text = String(link || '').trim(); + if (!text) return ''; + const match = text.match(/\/imagine\/post\/([0-9a-fA-F]{8}-[0-9a-fA-F]{4}-[0-9a-fA-F]{4}-[0-9a-fA-F]{4}-[0-9a-fA-F]{12})(?:[/?#]|$)/i); + return match ? match[1] : ''; + } + + function resolveVideoPostId(meta = {}) { + const directPostId = String(meta.postId || meta.post_id || '').trim(); + if (directPostId) return directPostId; + const shareLinkPostId = extractPostIdFromShareLink(meta.shareLink || meta.share_link || ''); + if (shareLinkPostId) return shareLinkPostId; + const originalPostId = String(meta.originalPostId || meta.original_post_id || '').trim(); + if (originalPostId) return originalPostId; + const namePostId = extractPostIdFromFileName(meta.name || ''); + if (namePostId) return namePostId; + return extractPostIdFromFileName(meta.url || ''); + } + + function applyResolvedVideoIdentity(meta = {}, sourceLabel = '') { + const resolvedPostId = resolveVideoPostId(meta); + if (!resolvedPostId) { + if (sourceLabel) { + console.warn(`[视频标识] ${sourceLabel} 未解析到 post_id`, meta); + } + return ''; + } + currentExtendPostId = resolvedPostId; + currentFileAttachmentId = resolvedPostId; + if (!originalFileAttachmentId) { + originalFileAttachmentId = resolvedPostId; + } + if (sourceLabel) { + console.log(`[视频标识] ${sourceLabel} 解析 post_id:`, resolvedPostId); + } + return resolvedPostId; + } + function debugLog(...args) { // console.log('[video-extend-debug]', ...args); } @@ -1570,24 +1610,13 @@ }); } - function bindEditVideoSource(url) { + function bindEditVideoSource(url, meta = {}) { const safeUrl = String(url || '').trim(); selectedVideoUrl = safeUrl; if (editHint) { editHint.classList.toggle('hidden', Boolean(safeUrl)); } - // 从 URL 中提取 postId （支持历史面板点击编辑） - const postId = extractPostIdFromFileName(safeUrl); - if (postId) { - currentExtendPostId = postId; - currentFileAttachmentId = postId; - // 首次设置 originalFileAttachmentId（后续延长不覆盖） - if (!originalFileAttachmentId) { - originalFileAttachmentId = postId; - debugLog('bindEditVideoSource: set originalFileAttachmentId =', postId); - } - debugLog('bindEditVideoSource: extracted postId =', postId); - } + applyResolvedVideoIdentity({ ...meta, url: safeUrl }, 'bindEditVideoSource'); if (!editVideo) return; enforceInlinePlayback(editVideo); editVideo.src = safeUrl; @@ -1617,7 +1646,12 @@ } if (editHint) editHint.classList.add('hidden'); if (editBody) editBody.classList.remove('hidden'); - bindEditVideoSource(url); + bindEditVideoSource(url, item ? { + postId: item.dataset.postId || '', + shareLink: item.dataset.shareLink || '', + originalPostId: item.dataset.originalPostId || '', + name: item.dataset.name || '', + } : {}); } function closeEditPanel() { @@ -1747,9 +1781,12 @@ const html = items.map((item, idx) => { const name = String(item.name || ''); const url = String(item.view_url || ''); + const postId = String(item.post_id || ''); + const shareLink = String(item.share_link || ''); + const originalPostId = String(item.original_post_id || ''); const size = formatBytes(item.size_bytes); const mtime = formatMtime(item.mtime_ms); - return `
+ return `
@@ -1788,25 +1825,20 @@ } } - function useCachedVideo(url, name) { + function useCachedVideo(url, name, meta = {}) { const safeUrl = String(url || '').trim(); if (!safeUrl) return; selectedVideoItemId = `cache-${Date.now()}`; selectedVideoUrl = safeUrl; - // 从文件名中自动提取 parentPostId 用于视频延长 - const extractedPostId = extractPostIdFromFileName(String(name || '')); - if (extractedPostId) { - currentExtendPostId = extractedPostId; - currentFileAttachmentId = extractedPostId; - // 从缓存重新选择视频时重置 originalFileAttachmentId（开启新的延长链） - originalFileAttachmentId = extractedPostId; - debugLog('useCachedVideo: extracted postId =', extractedPostId); - } + originalFileAttachmentId = ''; + applyResolvedVideoIdentity({ ...meta, name, url: safeUrl }, 'useCachedVideo'); if (imageUrlInput) imageUrlInput.value = safeUrl; if (imageFileName && name) imageFileName.textContent = name; if (enterEditBtn) enterEditBtn.disabled = false; closeCacheVideoModal(); - openEditPanel(); + bindEditVideoSource(safeUrl, { ...meta, name }); + if (editHint) editHint.classList.add('hidden'); + if (editBody) editBody.classList.remove('hidden'); setEditMeta(); } @@ -2404,11 +2436,12 @@ editVideo.src = videoUrl; editVideo.load(); } - // 从新视频 URL 提取 postId 用于链式延长 - const newPostId = extractPostIdFromFileName(videoUrl); + // 优先从 create-link 元数据对应的 post_id，最后才回退到 URL + const newPostId = applyResolvedVideoIdentity({ url: videoUrl }, 'extend-sse'); if (newPostId) { - currentExtendPostId = newPostId; // 更新当前 postId - currentFileAttachmentId = newPostId; + if (item) { + item.dataset.postId = newPostId; + } console.log('[SSE 调试] 从新视频成功提取到新的 extend_post_id:', newPostId); } else { console.warn('[SSE 调试] 未能从新视频地址提取出新的 extend_post_id!', videoUrl); @@ -2708,7 +2741,11 @@ if (!target.classList.contains('cache-video-use')) return; const row = target.closest('.cache-video-item'); if (!row) return; - useCachedVideo(row.getAttribute('data-url') || '', row.getAttribute('data-name') || ''); + useCachedVideo(row.getAttribute('data-url') || '', row.getAttribute('data-name') || '', { + postId: row.getAttribute('data-post-id') || '', + shareLink: row.getAttribute('data-share-link') || '', + originalPostId: row.getAttribute('data-original-post-id') || '', + }); }); } @@ -2726,7 +2763,13 @@ return; } // 提取 postId 用于延长 - const postId = extractPostIdFromFileName(bUrl); + const postId = resolveVideoPostId({ + postId: item.dataset.postId || '', + shareLink: item.dataset.shareLink || '', + originalPostId: item.dataset.originalPostId || '', + name: item.dataset.name || '', + url: bUrl, + }); if (postId) { currentExtendPostId = postId; currentFileAttachmentId = postId; @@ -2750,7 +2793,12 @@ return; } if (!target.classList.contains('video-download')) { - bindEditVideoSource(selectedVideoUrl); + bindEditVideoSource(selectedVideoUrl, { + postId: item.dataset.postId || '', + shareLink: item.dataset.shareLink || '', + originalPostId: item.dataset.originalPostId || '', + name: item.dataset.name || '', + }); return; } event.preventDefault(); diff --git a/config.defaults.toml b/config.defaults.toml index 81c71c3..1426e44 100644 --- a/config.defaults.toml +++ b/config.defaults.toml @@ -124,8 +124,6 @@ concurrent = 100 timeout = 60 # 流式空闲超时时间（秒） stream_timeout = 60 -# 视频延长流式空闲超时时间（秒） -extension_stream_timeout = 300 # ==================== 语音配置 ==================== [voice]