Peach: Local-first video metadata scraper and media manager.
feat(downloads): 待找与找资源的 JavDB 磁链改走匿名接口
longmeidao53849317a3ce
diff --git a/CHANGELOG.md b/CHANGELOG.mdindex 4320837a4..9ff1892c4 100644--- a/CHANGELOG.md+++ b/CHANGELOG.md@@ -39,9 +39,11 @@ Removed / Fixed / Security。- **账本**:「应用与维护」新增「账本备份」:可以立即备份账本,也可以从列表里的一份备份恢复。列表写明每份备份的时间、来源(手动或写入前自动)和大小;恢复只在写入端执行,会先检查备份并把当前账本另存一份,被中断后下次启动时自动收拾到完整的账本。- **复核**:有 NFO 或升级后重新核对资料的影片,在线来源与现有资料的差异不再一律交人复核。只差写法、打码或店铺装饰的,和只有社区站有异议的,保留现有资料;来源多给的标签自动补上,一个不删;官方来源一致的标题与发行日期直接更新。厂牌、系列、出演者的官方异议,以及 NFO 与官方来源对不上的,仍留在复核页。升级后对已有的这批差异自动整理一次,自动写入的可用 `revert_auto_landing.py` 按规则名整批撤回。- **采集**:随附用户脚本更新至 0.7.2。浮卡「其他任务」展开、折叠有高度过渡,进出番号的进度页沿用设置面板的进场动效。别的页面添加的任务结束后一直留在列表里,直到点行尾的关闭键、进度页的「标为已读」或列表底部的「全部标为已读」;没有进行中的任务时,收成图标的角标显示未读的已结束数量。列表底部的「重试失败项」按失败原因逐项重试:整理失败的重试整理,磁力失效的换一个磁力,其余从原网页重新加入。Peach 自动重新提交的云下载显示「等待自动重新提交」。更新脚本并刷新来源页面后生效。+- **采集**:想要清单和云下载找资源时,JavDB 一环改用新作页同一套匿名接口取磁链区和评论里的链接,不再读 JavDB 网页,也就不再受网页的登录墙和 Cloudflare 影响。同番号对应多部作品时提示里列出各自的来源编号。### 修复+- **采集**:新作页资源区里 JavDB 磁链区的条目显示实际大小(如 4.12GB),不再一律写「大小未知」。- **采集**:HEYZO 编号小于 1000 的影片(如 `HEYZO-0707`)不再被收成三位、查不到资料;升级时把已入库的三位写法改回四位并重新处理,手改过的番号不动。- **播放**:生成过时间轴缩略图的视频,详情页进度条预览只显示指向时刻的那一格,不再把一整行画面挤在一个框里。- **播放**:进度条较窄时,靠近两端的画面预览不再被裁掉一截。diff --git a/docs/JAV_DISCOVERY.md b/docs/JAV_DISCOVERY.mdindex 0bd1c4eb1..f3ce67ec9 100644--- a/docs/JAV_DISCOVERY.md+++ b/docs/JAV_DISCOVERY.md@@ -85,8 +85,8 @@资源协议用文字「磁链」「ED2K」「网页链接」。标题和元数据占据资源区宽度,窄资源区的操作键在下方换行,宽资源区并列显示。字节容量复用馆藏 `fmtSize` 口径显示,有明确单位的容量按来源展示;纯数字且单位未取得时显示「大小未标注」,悬停可查看来源原值。复制和下载使用原始链接。-云下载的直接添加优先取得可用高清候选;「想要」的待找清单在同一请求预算内继续汇集 manko.fun、JavDB 和 sukebei,按画质与版本排序并去重。JavDB 资源区与评论中的磁链、ED2K 和网页链接按原协议保留,取得范围提示随来源返回。待找缓存与快速添加候选分开,已取得的多个资源保留展开入口。+云下载的直接添加优先取得可用高清候选;「想要」的待找清单在同一请求预算内继续汇集 manko.fun、JavDB 和 sukebei,按画质与版本排序并去重。JavDB 一环与新作页共用匿名资源接口,读磁链区与前 4 页评论;资源区与评论中的磁链、ED2K 和网页链接按原协议保留,取得范围提示随来源返回。待找缓存与快速添加候选分开,已取得的多个资源保留展开入口。-新作页的匿名资源接口合并磁链区与前 8 页评论的链接,资源按协议身份去重并保留评论锚点。评论取数失败时保留磁链区结果,并在资源区提示未取得的范围;重新查询同时刷新磁链和评论。资源缓存使用包含评论的独立身份。+新作页的匿名资源接口合并磁链区与前 8 页评论的链接,资源按协议身份去重并保留评论锚点。磁链区体积由来源的 MiB 数换算成 MB 或 GB 显示。评论取数失败时保留磁链区结果,并在资源区提示未取得的范围;重新查询同时刷新磁链和评论。资源缓存使用包含评论的独立身份。待找资源在浏览器和服务端本机缓存。发行未满三个月保存 7 天,更早作品保存 1 年,最多保留 128 部;服务重启后有效期从原查询时间继续计算。已取得的结果优先读取,不等待其他作品的查询锁;过期或点击「重新查询」才访问来源。主动查询失败时保留上次成功资源;来源失败或受限且没有结果的状态只短暂缓存,不写入持久结果。缓存位于私有来源目录,不写入账本。diff --git a/docs/REUSE.md b/docs/REUSE.mdindex aee038d4d..9935f07e6 100644--- a/docs/REUSE.md+++ b/docs/REUSE.md@@ -731,7 +731,7 @@ SFW 开关键可在偏好设置的「内容显示」里改。主键按 `Keyboard- CloudDrive 引导用现有 `settings_file`、`platform.root_online`、`scan_location` 与 React 配置页,来源为 `local`、`115`、`pikpak`。挂载由用户安装的 CloudDrive 负责([官方帮助](https://www.clouddrive2.com/help.html):Windows 用盘符、macOS 用目录挂载点),Peach 不捆绑其二进制,挂载引导不依赖管理 API,路径处理用 `pathlib`、`os.scandir`、`tomllib`;表单、来源归属与扫描选择归 Peach。云下载和账号代理配置经官方 proto 子集(`downloads_clouddrive.proto`,version 1.1.2)走 gRPC,令牌存 CredentialStore,不用非官方 SDK(ADR-0089)。- 媒体目录校验沿用 `media.normalized_path` 的平台策略:Windows 用标准库 `os.path.abspath` 保留盘符,macOS 用 `Path.resolve` 解析文件系统链接。实现只依赖项目 Python 标准库(PSF 许可),不新增依赖;校验器独立于账本路径翻译与媒体仓储。映射盘转 UNC 的边界由注入文件系统解析结果的回归覆盖,首次设置的本地来源也验证配置保留盘符。直接输入 UNC 仍按盘符路径契约拒绝;真实 NAS 挂载需在目标机器验收。- 非 Windows 表单的内部路径由 `media_configuration.prepare_sources` 补齐,复用既有来源校验、配置写入和 `platform` 路径转换;不改数据库路径模型或执行迁移。标准库 [PureWindowsPath / PurePosixPath](https://docs.python.org/3.14/library/pathlib.html#pure-paths) 用于比较两种路径,[hashlib.sha256](https://docs.python.org/3.14/library/hashlib.html) 按来源与实际目录生成稳定候选前缀,已有配置和显式输入优先,冲突时换盘符。媒体库名称使用实际目录名,前端复用现有 `Disclosure` 和 `Input`,无新增依赖。-- JAV 入库资源查询复用 `sources.javdb` 的身份核对、`SourceTransport` 的 Cookie 与冷却、`HostLimitedTransport` 的 3 秒间隔、`beautifulsoup4==4.15.0` 和 `downloads.parse_magnet`。JavPack 的 `JavDB.magnet.user.js` 0.0.2(GPL-3.0,仓库 revision `b546e1881147e15796c697e8e086db9d3ac0c5be`)只作页面字段参考;它依赖浏览器用户脚本 API,不能直接作为 Python 服务端依赖。Peach 负责评论链接、协议分流、去重、60 秒缓存和最多三页评论的预算,不新增依赖。2026-10-02 单作品只读实测取得 4 条磁链和含 ed2k 的评论页;评论入口从 `.review-tab[data-url]` 读取,受限评论保留来源限制提示。磁链可交现有云下载服务,ed2k 与其他网页链接提供复制。+- JAV 入库资源查询与新作页共用 `javdb_public.PublicAPI.resources`:匿名 API 的作品身份核对、3 秒间隔、请求预算、来源冷却和按发行日期保存的快照,加上 `downloads.parse_magnet` 与 `sources.javdb.parse_comment_text`。磁链区的 `size` 是 MiB 整数,换算成带单位的文字后才进入体积过滤与排序。Peach 负责评论链接、协议分流和去重,不新增依赖。待找清单读磁链区与前 4 页评论,预算跟随本轮查询期限;作品身份歧义按来源原文报告。磁链可交现有云下载服务,ed2k 与其他网页链接提供复制。- 磁力优先链复用同一套 `SourceTransport`、`HostLimitedTransport`、`downloads.parse_magnet` 与 Torznab 后备,不引入新的抓取运行时。manko.fun 的页面请求实际落到 `healertanker.com/swx`,响应 `data` 是 basE91 包装的 JSON;实现只移植该公开响应形态与编号精确匹配,不复用会话脚本。Sukebei 直接读搜索表格;JavBus、16mag、OneJAV、Free JAV Torrent 由现有自配 Prowlarr/Jackett 索引器续接。Akiba Online 与 Reddit 只作线索,不自动取磁力。候选按分辨率、做种、推广附件数量与正片体积排序;总文件数不再是一票否决,manko.fun 候选绕过水印门槛,其他来源只能标为待到盘检查。- 云下载后处理复用 `task_run` 后继串行、`process_library(retry_ids=...)`、`junk.isolate`、`organize.build_plan(asset_ids=...)`、SQLite backup 与整理批次日志。没有另造第二套刮削、垃圾词库或移动器。- `junk.scored_candidates` 统一复核与自动隔离的证据;`junk.isolate` 要求显式资产或来源目录,保护人工决定、个人记录、字幕与正片身份,写前重查证据,仅写可撤销的回收站状态。扫描、推送、媒体探测、资料处理、云下载与维护脚本共用该入口;页面只补展示成本与缩略图字段。复用当前树的词库、目录关系、`record_holders`、`LedgerDatabase` 与历史评分回归,使用 Python 标准库和现有 SQLite JSON 支持,无新增依赖、协议或媒体读取。流程顺序见[垃圾文件](OPERATIONS.md#垃圾文件)。diff --git a/src/peach/javdb_public.py b/src/peach/javdb_public.pyindex c5c8adee1..972fa9930 100644--- a/src/peach/javdb_public.py+++ b/src/peach/javdb_public.py@@ -118,6 +118,13 @@ def preview_video_url(movie: dict) -> str:return value if allowed else ""+def magnet_size(value: object) -> str:+ """磁链区的体积是 MiB 整数,写成带单位的文字;其他形态原样保留。"""+ if type(value) in (int, float) and value > 0:+ return f"{value / 1024:.2f}GB" if value >= 1024 else f"{value:g}MB"+ return str(value or "")++class PublicSourceUnavailable(ValueError):"""公开来源当前未取得。"""@@ -493,6 +500,53 @@ class PublicAPI:self._rows(f"/v1/movies/{movie['id']}/magnets", "magnets"),released=movie.get('release_date'), refresh=refresh)+ def resources(self, code: str, *, refresh=False, review_pages=MAX_REVIEW_PAGES) -> dict:+ """磁链区与评论正文的资源链接;一侧未取得只记警告,作品身份未取得直接抛出。"""+ from .downloads import parse_magnet+ from .sources.javdb import merge_resources, parse_comment_text++ movie = self.movie(code, refresh=True) if refresh else self.movie(code)+ source_url = "https://javdb.com/v/" + movie["id"]+ warnings = []+ items = {}+ magnet_failed = comments_failed = False+ try:+ rows = self.magnets(code, refresh=refresh, movie=movie)+ except PublicSourceUnavailable as error:+ rows = []+ magnet_failed = True+ warnings.append("磁链区资源未取得:" + str(error))+ for row in rows:+ try:+ magnet = parse_magnet(str(row.get("hash") or ""))+ except ValueError:+ continue+ if magnet.info_hash in items:+ continue+ attributes = [label for key, label in (("cnsub", "中字"), ("hd", "高清")) if row.get(key)]+ items[magnet.info_hash] = {"id": "magnet:" + magnet.info_hash, "protocol": "magnet",+ "info_hash": magnet.info_hash, "uri": magnet.uri, "name": row.get("name") or code,+ "size": magnet_size(row.get("size")), "files": str(row.get("files_count") or ""),+ "date": str(row.get("created_at") or ""), "attributes": attributes,+ "source": "JavDB", "source_url": source_url, "origins": ["JavDB"]}+ comments = []+ try:+ for page in range(1, review_pages + 1):+ reviews = self.reviews(code, page, refresh=refresh, movie=movie)+ for review in reviews:+ origin = source_url + "#review-item-" + str(review.get("id") or "")+ comments.extend(parse_comment_text(str(review.get("content") or ""), origin,+ str(review.get("created_at") or "")))+ if len(reviews) < PAGE_SIZE:+ break+ else:+ warnings.append(f"已查询前 {review_pages} 页评论资源。")+ except PublicSourceUnavailable as error:+ comments_failed = True+ warnings.append("评论资源未取得:" + str(error))+ return {"items": merge_resources([*items.values(), *comments]), "warnings": warnings,+ "source_url": source_url, "magnet_failed": magnet_failed, "comments_failed": comments_failed}+def official_preview(self, code, fetch, *, released=None, refresh=False, provider='prestige', source_id=None):"""预告按来源独立保存;DMM、JavDB 和未提供的预告每天可再检查。"""def read():diff --git a/src/peach/magnet_chain.py b/src/peach/magnet_chain.pyindex 3f2b9dee9..2acdd9331 100644--- a/src/peach/magnet_chain.py+++ b/src/peach/magnet_chain.py@@ -20,9 +20,11 @@ from bs4 import BeautifulSoupfrom .catalog_rules import normalise_code_key, release_code_from_text, same_release_codefrom .downloads import parse_magnetfrom .jav_cover_fetch import HostLimitedTransport+from .javdb_public import (_CACHE_ROOT, PUBLIC_API, PublicIdentityAmbiguous, PublicRequestBudget,+ PublicSourceUnavailable, public_cache_scope)from .resource_search import qualitiesfrom .scraping_access import SourceTransport-from .sources import JavDBSource, Session, SourceFailure+from .sources import FailureReason, Session, SourceFailureHIGH_QUALITY = 1080MAX_RESULTS = 8@@ -218,12 +220,20 @@ class Fetcher:return rank_candidates(found)def javdb(self, code: str) -> list[dict]:+ """匿名 API 现取磁链区与前四页评论;快照兜底的失败也记成警告,预算跟随本轮期限。"""source = CHAIN[1]- result = JavDBSource().resources(code, session=self.session)+ budget = PublicRequestBudget(max_requests=12, max_seconds=max(1, self.deadline - time.monotonic()))+ try:+ with public_cache_scope(_CACHE_ROOT.get()) as reads, PUBLIC_API.request_budget(budget):+ result = PUBLIC_API.resources(code, refresh=True, review_pages=4)+ except PublicSourceUnavailable as error:+ reason = FailureReason.PARSE_ERROR if isinstance(error, PublicIdentityAmbiguous) else FailureReason.NETWORK+ raise SourceFailure(reason, str(error)) from error+ stale = [f"已沿用上次结果:{read['error']}" for read in reads if read['error']]+ result["warnings"] = list(dict.fromkeys([*result["warnings"], *stale]))self.resource_result = result- return rank_candidates([candidate for item in result.get("items", ())- if (candidate := _normalise({**item, "source_url": result.get("source_url")},- source, code))])+ return rank_candidates([candidate for item in result["items"] if (candidate := _normalise(+ {**item, "file_count": int(item["files"]) if item["files"].isdigit() else 0}, source, code))])def sukebei(self, code: str) -> list[dict]:from .sources.base import SiteConfigdiff --git a/src/peach/sources/javdb.py b/src/peach/sources/javdb.pyindex 7b5823b68..8107eceae 100644--- a/src/peach/sources/javdb.py+++ b/src/peach/sources/javdb.py@@ -66,54 +66,6 @@ def maker_writing(value: str) -> str:class JavDBSource(SiteSource):DEFAULT = JAVDB- def resources(self, code: str, *, session: Session) -> dict:- """核对作品身份后合并资源区与评论链接;最多读取三页评论。"""- with self.holding(session):- page = self.fetch(code, session=session)- self.parse(page, code)- warnings = []- try:- items = parse_magnets(page)- except SourceFailure as error:- items = []- warnings.append(str(error))- soup = BeautifulSoup(page.text, "html.parser")- items.extend(parse_comment_links(page))- entry = soup.select_one('.review-tab[data-url]')- next_url = str(entry.get("data-url", "")) if entry else ""- seen = set()- for _ in range(3):- if not next_url:- break- url = urllib.parse.urljoin(page.url, next_url)- parts, original = urllib.parse.urlsplit(url), urllib.parse.urlsplit(page.url)- if (parts.netloc != original.netloc or parts.scheme != "https"- or not parts.path.startswith(original.path + "/reviews/") or url in seen):- warnings.append("部分评论未取得,请打开来源页查看。")- break- seen.add(url)- try:- with self.holding(session):- comments = self._page(url, session=session)- body = BeautifulSoup(comments.text, "html.parser")- if not body.select_one('.review-items'):- raise SourceFailure(FailureReason.PARSE_ERROR, "评论区域未取得")- items.extend(parse_comment_links(comments))- except Exception:- warnings.append("评论链接未取得,请打开来源页查看。")- break- if body.select_one('.more a[href="/plans"]'):- warnings.append("来源只返回了部分评论,资源链接可能不完整。")- following = body.select_one('a[rel="next"], .pagination-next:not([disabled])')- next_url = str(following.get('href', '')) if following else ""- else:- if next_url:- warnings.append("已读取前三页评论,其余请打开来源页查看。")- if not entry and not soup.select_one('.review-items'):- warnings.append("评论入口未取得。")- return {"items": merge_resources(items), "warnings": list(dict.fromkeys(warnings)),- "source_url": page.url}-def search_url(self, code: str) -> str:return f"{self.config.base_url}/search?q={urllib.parse.quote(code)}&f=all"@@ -168,67 +120,12 @@ class JavDBSource(SiteSource):'source_id': urllib.parse.urlsplit(page.url).path.rsplit('/', 1)[-1], 'javdb_rule': RULE})-def parse_magnets(page: Page) -> list[dict]:- """JavDB 磁链区:名称、大小、文件数、标签与收录日期;按 infohash 去重。"""- from ..downloads import parse_magnet-- soup = BeautifulSoup(page.text, "html.parser")- container = soup.select_one("#magnets-content")- if container is None:- raise SourceFailure(FailureReason.PARSE_ERROR, "JavDB 磁链区域未取得")- found: dict[str, dict] = {}- invalid = False- for row in container.select(".item"):- link = row.select_one('a[href^="magnet:"]')- if link is None:- continue- try:- magnet = parse_magnet(str(link.get("href", "")))- except ValueError:- invalid = True- continue- def text(selector, row=row):- node = row.select_one(selector)- return node.get_text(" ", strip=True) if node else ""- meta = re.split(r"[,,]", text(".meta"), maxsplit=1)- found.setdefault(magnet.info_hash, {- "id": "magnet:" + magnet.info_hash, "protocol": "magnet",- "info_hash": magnet.info_hash, "uri": magnet.uri,- "name": text(".name") or magnet.name or magnet.info_hash,- "size": meta[0].strip(), "files": meta[1].strip() if len(meta) > 1 else "",- "date": text(".time") or text(".date"),- "attributes": list(dict.fromkeys(tag.get_text(" ", strip=True)- for tag in row.select(".tag") if tag.get_text(strip=True))),- "source": "JavDB", "source_url": page.url, "origins": [page.url],- })- if invalid and not found:- raise SourceFailure(FailureReason.PARSE_ERROR, "JavDB 磁链格式无法识别")- return list(found.values())--_COMMENT_LINK = re.compile(r"(?:magnet:\?[^\s<>\"']+|ed2k://\|file\|[^\r\n<>]+?\|/|https?://[^\s<>\"']+)", re.I)_ED2K = re.compile(r"ed2k://\|file\|([^|]+)\|(\d+)\|([a-f0-9]{32})\|(?:[^\r\n]*?\|)?/", re.I)-def parse_comment_links(page: Page) -> list[dict]:- """只取评论正文的链接,跳过作者主页、点赞与站点推广入口。"""- found = []- soup = BeautifulSoup(page.text, "html.parser")- for row in soup.select('.review-item'):- content = row.select_one('.content')- if content is None:- continue- links = [str(a.get('href', '')) for a in content.select('a[href]')]- links.extend(_COMMENT_LINK.findall(content.get_text(' ', strip=True)))- date = row.select_one('.time')- # 片段引用保留评论 id;正文中的任意 URL 都不会被服务器追踪请求。- origin = page.url + ("#" + str(row['id']) if row.get('id') else "")- found.extend(comment_resources(links, origin, date.get_text(' ', strip=True) if date else ""))- return merge_resources(found)--def parse_comment_text(text: str, origin: str, date: str = "") -> list[dict]:- """匿名评论正文与网页评论使用同一资源解析。"""+ """匿名评论正文里的磁链、ed2k 与网页链接;正文中的 URL 都不会被服务器追踪请求。"""return comment_resources(_COMMENT_LINK.findall(text), origin, date)diff --git a/src/peach/web_jav_detail.py b/src/peach/web_jav_detail.pyindex 3605b530e..20fd22a19 100644--- a/src/peach/web_jav_detail.py+++ b/src/peach/web_jav_detail.py@@ -346,56 +346,16 @@ def _preview_fields(contract, code, source, *, refresh=False):def q_jav_resources(contract, args) -> dict:"""公开资源列表;只返回候选,下载提交由既有入口接受用户选择。"""import time- from .downloads import parse_magnet- from .sources.javdb import merge_resources, parse_comment_textcode = detail_code(args)try:- movie = PUBLIC_API.movie(code, refresh=True) if args.get('refresh') == '1' else PUBLIC_API.movie(code)+ found = PUBLIC_API.resources(code, refresh=args.get('refresh') == '1')except PublicSourceUnavailable as error:return {"state": "error", "items": [], "error": str(error), "checked_at": None}- warnings = []- magnet_failed = False- try:- rows = PUBLIC_API.magnets(code, refresh=args.get('refresh') == '1', movie=movie)- except PublicSourceUnavailable as error:- rows = []- magnet_failed = True- warnings.append("磁链区资源未取得:" + str(error))- items = {}- source_url = "https://javdb.com/v/" + movie["id"]- for row in rows:- try:- magnet = parse_magnet(str(row.get("hash") or ""))- except ValueError:- continue- if magnet.info_hash in items:- continue- attributes = [label for key, label in (("cnsub", "中字"), ("hd", "高清")) if row.get(key)]- items[magnet.info_hash] = {"id": "magnet:" + magnet.info_hash, "protocol": "magnet",- "info_hash": magnet.info_hash, "uri": magnet.uri, "name": row.get("name") or code,- "size": str(row.get("size") or ""), "files": str(row.get("files_count") or ""),- "date": str(row.get("created_at") or ""), "attributes": attributes,- "source": "JavDB", "source_url": source_url, "origins": ["JavDB"]}- comments = []- comments_failed = False- try:- for page in range(1, MAX_REVIEW_PAGES + 1):- reviews = PUBLIC_API.reviews(code, page, refresh=args.get('refresh') == '1', movie=movie)- for review in reviews:- origin = source_url + "#review-item-" + str(review.get("id") or "")- comments.extend(parse_comment_text(str(review.get("content") or ""), origin,- str(review.get("created_at") or "")))- if len(reviews) < PAGE_SIZE:- break- else:- warnings.append("已查询前 8 页评论资源。")- except PublicSourceUnavailable as error:- comments_failed = True- warnings.append("评论资源未取得:" + str(error))- merged = merge_resources([*items.values(), *comments])- failed = magnet_failed and comments_failed and not merged- return {"state": "error" if failed else "ready", "items": merged,- "error": ";".join(warnings) if failed else "", "source_url": source_url,+ warnings = found["warnings"]+ magnet_failed, comments_failed = found["magnet_failed"], found["comments_failed"]+ failed = magnet_failed and comments_failed and not found["items"]+ return {"state": "error" if failed else "ready", "items": found["items"],+ "error": ";".join(warnings) if failed else "", "source_url": found["source_url"],"checked_at": time.strftime("%Y-%m-%dT%H:%M:%SZ", time.gmtime()), "warnings": warnings,"cache_stale": magnet_failed or comments_failed, "complete": not (magnet_failed or comments_failed)}diff --git a/src/peach/web_resource_search.py b/src/peach/web_resource_search.pyindex 80212139d..8be3104d4 100644--- a/src/peach/web_resource_search.py+++ b/src/peach/web_resource_search.py@@ -1,4 +1,5 @@"""资源搜索的只读 Web 契约;下载仍由云下载写入口接收。"""+from .javdb_public import public_cache_scopefrom .magnet_chain import HIGH_QUALITY, rank_candidates, search as direct_searchfrom .resource_search import Filters, Indexers, Search@@ -23,9 +24,10 @@ def q_resource_search(contract, args) -> dict:skip = [str(value) for value in args.get("skip") or () if isinstance(args.get("skip"), (list, tuple))]excluded = set(blocked) | set(skip)code = str(args.get("code") or "")- direct = direct_search(- contract.follow_secrets_root, code, min_size=filters.min_size,- max_size=filters.max_size, goal=filters.goal, stop_on_preferred=not skip)+ with public_cache_scope(contract.cover_root.parent / 'jav-public'):+ direct = direct_search(+ contract.follow_secrets_root, code, min_size=filters.min_size,+ max_size=filters.max_size, goal=filters.goal, stop_on_preferred=not skip)direct["items"] = [row for row in direct["items"] if row["info_hash"] not in excluded]# manko.fun 给出高清候选时按用户指定的优先级短路,不再消耗后备索引器请求。if not skip and direct["items"] and direct["items"][0]["source"] == "manko.fun" \diff --git a/src/peach/web_wants.py b/src/peach/web_wants.pyindex 1be137c13..688f6aadd 100644--- a/src/peach/web_wants.py+++ b/src/peach/web_wants.py@@ -214,6 +214,7 @@ def _ids(body) -> list[int]:def q_want_magnets(contract, args) -> dict:"""只读资源查询;未发售、已入库与没有番号的条目不请求来源。"""+ from .javdb_public import public_cache_scopefrom .wants_magnets import search_forraw = str(args.get("id") or "")@@ -225,8 +226,10 @@ def q_want_magnets(contract, args) -> dict:raise KeyError("想要条目不存在")if not row["code"] or wants.phase(row, date.today()) in {"unreleased", "acquired"}:return {"ok": True, "state": "unavailable", "items": [], "error": "", "checked_at": None}- return search_for(contract.follow_secrets_root).query(- row["code"], refresh=str(args.get('refresh') or '') == '1', released=row['release_date'])+ # JavDB 匿名资料与详情页共用快照目录。+ with public_cache_scope(contract.cover_root.parent / 'jav-public'):+ return search_for(contract.follow_secrets_root).query(+ row["code"], refresh=str(args.get('refresh') or '') == '1', released=row['release_date'])def _add(connection, body) -> dict:diff --git a/tests/test_javdb_public.py b/tests/test_javdb_public.pyindex 46a4884b3..ae8cb042c 100644--- a/tests/test_javdb_public.py+++ b/tests/test_javdb_public.py@@ -637,6 +637,15 @@ class DetailProjectionTests(unittest.TestCase):self.assertEqual(result['items'][0]['attributes'], ['中字', '高清'])self.assertEqual(result['source_url'], 'https://javdb.com/v/Exact')+ def test_resource_sizes_are_mebibytes_written_with_units(self):+ rows = [{'hash': 'a' * 40, 'size': 4219}, {'hash': 'b' * 40, 'size': 700},+ {'hash': 'c' * 40, 'size': 0}, {'hash': 'd' * 40, 'size': '3 GB'}]+ with patch('peach.web_jav_detail.PUBLIC_API.movie', return_value={'id': 'Exact'}), \+ patch('peach.web_jav_detail.PUBLIC_API.magnets', return_value=rows), \+ patch('peach.web_jav_detail.PUBLIC_API.reviews', return_value=[]):+ result = q_jav_resources(self.contract, {'code': 'EXM-001'})+ self.assertEqual([item['size'] for item in result['items']], ['4.12GB', '700MB', '', '3 GB'])+def test_resources_include_comment_ed2k_magnets_and_urls_with_origins(self):digest = 'a' * 40ed2k = 'ed2k://|file|www.98T.la@MIKR-096-U_1_chf3_prob4.mp4|19735225787|0936618B7ED1E5A0235FC6FBE27E104A|h=U5OZKZ4TELDTNANYLRNGQX5LCTDKAM52|/'diff --git a/tests/test_magnet_chain.py b/tests/test_magnet_chain.pyindex 1d2423311..9478d31d9 100644--- a/tests/test_magnet_chain.py+++ b/tests/test_magnet_chain.py@@ -88,6 +88,33 @@ class MagnetChainTests(unittest.TestCase):now[0] += 120self.assertEqual(lookup.query('ABC-123')['state'], 'error')self.assertEqual(len(calls), 2)+ def test_javdb_step_reads_the_anonymous_api_and_keeps_comment_links(self):+ from peach.magnet_chain import Fetcher+ rows = [{'hash': HASH_A, 'name': 'ABC-123-C.torrent', 'size': 4219, 'cnsub': True, 'hd': True,+ 'files_count': 3, 'created_at': '2026-10-08'}]+ ed2k = 'ed2k://|file|ABC-123.mp4|1234|' + 'b' * 32 + '|/'+ fetcher = Fetcher(Path('.'), transport=object())+ with patch('peach.magnet_chain.PUBLIC_API.movie', return_value={'id': 'Exact'}) as movie, \+ patch('peach.magnet_chain.PUBLIC_API.magnets', return_value=rows) as magnets, \+ patch('peach.magnet_chain.PUBLIC_API.reviews', return_value=[{'id': 7, 'content': ed2k}]):+ found = fetcher.javdb('ABC-123')+ movie.assert_called_once_with('ABC-123', refresh=True)+ magnets.assert_called_once_with('ABC-123', refresh=True, movie={'id': 'Exact'})+ self.assertEqual(len(found), 1)+ self.assertEqual((found[0]['size'], found[0]['size_text']), (int(4.12 * 1024 ** 3), '4.12GB'))+ self.assertEqual((found[0]['chinese'], found[0]['file_count']), (True, 3))+ self.assertEqual(found[0]['source_url'], 'https://javdb.com/v/Exact')+ self.assertEqual([row['protocol'] for row in fetcher.resource_result['items']], ['magnet', 'ed2k'])++ def test_javdb_identity_ambiguity_is_reported_with_source_ids(self):+ from peach.javdb_public import PublicIdentityAmbiguous+ from peach.magnet_chain import Fetcher, search+ ambiguous = PublicIdentityAmbiguous('同番号对应多个作品,请核对来源身份:Aa1、Bb2')+ with patch.object(Fetcher, 'manko', return_value=[]), patch.object(Fetcher, 'sukebei', return_value=[]), \+ patch('peach.magnet_chain.PUBLIC_API.movie', side_effect=ambiguous):+ result = search(Path('.'), 'ABC-123', transport=object(), stop_on_preferred=False)+ self.assertEqual(result['warnings'], ['JavDB:同番号对应多个作品,请核对来源身份:Aa1、Bb2'])+def test_base91_decodes_the_api_envelope_payload(self):# basE91 官方示例:编码后的这一段解回 ASCII "test"。self.assertEqual(decode_base91("fPNKd"), b"test")diff --git a/tests/test_metadata_sources.py b/tests/test_metadata_sources.pyindex 5babe269c..1c0c5ec3b 100644--- a/tests/test_metadata_sources.py+++ b/tests/test_metadata_sources.py@@ -87,58 +87,26 @@ class JavDBResourceTests(unittest.TestCase):JavDBSource().parse(page, "ABW-221")def test_resource_area_and_comment_links_share_identity(self):- from peach.sources.javdb import parse_magnets, parse_comment_links, merge_resources+ from peach.sources.javdb import parse_comment_text, merge_resourcesdigest = 'a' * 40file_hash = 'b' * 32- html = f'''<div id="magnets-content"><div class="item">- <a href="magnet:?xt=urn:btih:{digest}&dn=demo"><span class="name">演示</span></a>- <span class="meta">4 GB, 2 個文件</span><span class="tag">字幕</span><span class="time">2026-10-01</span>- </div></div><dl class="review-items"><dt class="review-item" id="review-item-1">- <a href="https://ignored.test/profile">作者</a><span class="time">2026-10-02</span>- <div class="content"><a href="magnet:?xt=urn:btih:{digest.upper()}">另一条写法</a>- ed2k://|file|demo%20file.mkv|123456|{file_hash}|/- <a href="ed2k://|file|another-name|123456|{file_hash.upper()}|/">重复</a>- <a href="https://EXAMPLE.test/file?q=1">补充</a> https://example.test/file?q=1- <a href="javascript:alert(1)">忽略</a></div></dt></dl>'''- page = Page('https://javdb.com/v/demo', html.encode())- items = merge_resources([*parse_magnets(page), *parse_comment_links(page)])+ magnet = {'id': 'magnet:' + digest, 'protocol': 'magnet', 'info_hash': digest,+ 'uri': f'magnet:?xt=urn:btih:{digest}', 'name': '演示', 'size': '4.00GB', 'files': '2',+ 'date': '2026-10-01', 'attributes': ['中字'], 'source': 'JavDB',+ 'source_url': 'https://javdb.com/v/demo', 'origins': ['JavDB']}+ text = (f'magnet:?xt=urn:btih:{digest.upper()} ed2k://|file|another-name|123456|{file_hash}|/\n'+ f'ed2k://|file|demo%20file.mkv|123456|{file_hash.upper()}|/\n'+ 'https://EXAMPLE.test/file?q=1 https://example.test/file?q=1 javascript:alert(1)')+ comments = parse_comment_text(text, 'https://javdb.com/v/demo#review-item-1', '2026-10-02')+ items = merge_resources([magnet, *comments])self.assertEqual([item['protocol'] for item in items], ['magnet', 'ed2k', 'url'])- self.assertEqual(items[0]['size'], '4 GB')- self.assertEqual(items[0]['attributes'], ['字幕'])+ self.assertEqual(items[0]['size'], '4.00GB')+ self.assertEqual(items[0]['attributes'], ['中字'])self.assertEqual(len(items[0]['origins']), 2)self.assertEqual(items[1]['name'], 'another-name')self.assertEqual(items[1]['size'], '123,456 B')self.assertEqual(items[2]['uri'], 'https://example.test/file?q=1')- def test_missing_magnet_region_is_not_an_empty_result(self):- from peach.sources.javdb import parse_magnets- with self.assertRaises(SourceFailure):- parse_magnets(Page('https://javdb.com/v/demo', b'<html>login</html>'))- self.assertEqual(parse_magnets(Page('https://javdb.com/v/demo', b'<div id="magnets-content"></div>')), [])-- def test_comment_failure_keeps_magnets_and_reports_partial_results(self):- from peach.sources.javdb import JavDBSource- source = JavDBSource()- page = Page('https://javdb.com/v/demo', b'<div id="magnets-content"></div>'- b'<a class="review-tab" data-url="/v/demo/reviews/lastest"></a>')- with patch.object(source, 'fetch', return_value=page), patch.object(source, 'parse'), \- patch.object(source, '_page', side_effect=SourceFailure(FailureReason.AUTH_REQUIRED, 'login')):- result = source.resources('DEMO-001', session=Session(lambda *args: None))- self.assertEqual(result['items'], [])- self.assertEqual(result['source_url'], 'https://javdb.com/v/demo')- self.assertTrue(result['warnings'])-- def test_comment_urls_cannot_fetch_other_hosts(self):- from peach.sources.javdb import JavDBSource- source = JavDBSource()- page = Page('https://javdb.com/v/demo', b'<div id="magnets-content"></div>'- b'<a class="review-tab" data-url="https://example.test/private"></a>')- with patch.object(source, 'fetch', return_value=page), patch.object(source, 'parse'), \- patch.object(source, '_page') as fetch_page:- result = source.resources('DEMO-001', session=Session(lambda *args: None))- fetch_page.assert_not_called()- self.assertTrue(result['warnings'])-class DemoSource(SiteSource):"""最小的一站:作品页是 `/<番号>`,页面正文就是标题。"""