下载盘、页面验证、站点页、交给智能体、后台任务、提供方、凭证、其他、抓取间隔、 需要处理的事项、候选作品 / 匹配到多部作品、作品(不叫条目)、挂载点、下载中; 识别查询不再说「远端」,回收站与收录版本提示去掉公文腔;数据保留标签与设置页一致;同步断言。 Co-Authored-By: Claude Opus 5.5 <[email protected]> Claude-Session: https://claude.ai/code/session_01Cxwgnn2pbzQ7rXv3rq7fei
182 lines
6.5 KiB
TypeScript
182 lines
6.5 KiB
TypeScript
import { hostAllowed, type PageMetrics } from "@umt/adapters";
|
|
import type {
|
|
BrowserControlLease,
|
|
BrowserController,
|
|
BrowserCoordinateSpace,
|
|
BrowserEgressProxy,
|
|
BrowserEvidenceRef,
|
|
BrowserSession,
|
|
BrowserTab,
|
|
NavigationBudget,
|
|
} from "@umt/contracts";
|
|
import type { BrowserControlLeaseRow, BrowserSessionRow, BrowserTabRow } from "@umt/db";
|
|
|
|
// C1-E 浏览器服务层:BrowserProvider(camofox)之上的会话/标签/控制权持久化(数据库为准)、
|
|
// 站点域名白名单 + 站点限流、凭证注入、接管状态机、下载产物暂存核验。
|
|
// 适配器只是内存镜像:服务重启后按数据库恢复,旧元素引用一律失效。
|
|
|
|
export const iso = (d: Date | string | null | undefined) => (d ? new Date(d).toISOString() : null);
|
|
export type Json = Record<string, unknown>;
|
|
export const obj = (v: unknown): Json => (v && typeof v === "object" && !Array.isArray(v) ? (v as Json) : {});
|
|
|
|
export const REASON_TEXT: Record<string, string> = {
|
|
cf_challenge: "站点要求人机验证(Cloudflare),需要你过一次",
|
|
captcha: "登录页需要验证码",
|
|
challenge: "页面需要人工验证",
|
|
two_factor: "需要二次验证或本人设备确认",
|
|
challenge_unsolved: "页面验证没能自动通过",
|
|
login_required: "登录已失效:接管后在画面里登录一次,交还后全站生效",
|
|
lease_expired: "人工接管超时未交还",
|
|
agent_request: "智能体请求人工处理",
|
|
human_released_waiting: "人工交还时选择继续等待",
|
|
};
|
|
|
|
/** 重连回到的地址:会话记录的地址是 http(s)、在站点域名内、没有被脱敏过才用;否则 null(调用方退回站点首页) */
|
|
export function restoreUrl(url: string | null, budget: NavigationBudget): string | null {
|
|
const u = url ? safeUrl(url) : null;
|
|
if (!u || !/^https?:$/.test(u.protocol) || !hostAllowed(u.hostname, budget.domainAllowlist)) return null;
|
|
if (url!.includes("[REDACTED")) return null;
|
|
return url;
|
|
}
|
|
|
|
export function safeUrl(s: string): URL | null {
|
|
try {
|
|
return new URL(s);
|
|
} catch {
|
|
return null;
|
|
}
|
|
}
|
|
|
|
export function holderName(holder: string): string {
|
|
return holder.includes(":") ? holder.slice(holder.indexOf(":") + 1) : holder;
|
|
}
|
|
|
|
export function toSession(r: BrowserSessionRow): BrowserSession {
|
|
return {
|
|
id: r.id,
|
|
providerId: r.provider_id,
|
|
siteId: r.site_id,
|
|
accountId: r.account_id,
|
|
profileKey: r.profile_key,
|
|
controller: r.controller as BrowserController,
|
|
currentUrl: r.current_url,
|
|
lastActivityAt: iso(r.last_activity_at)!,
|
|
upstreamRef: obj(r.upstream_ref),
|
|
recoveryState: obj(r.recovery_state),
|
|
closedAt: iso(r.closed_at),
|
|
closeReason: r.close_reason,
|
|
createdAt: iso(r.created_at)!,
|
|
updatedAt: iso(r.updated_at)!,
|
|
};
|
|
}
|
|
|
|
export function toTab(r: BrowserTabRow): BrowserTab {
|
|
return {
|
|
id: r.id,
|
|
sessionId: r.session_id,
|
|
upstreamTabId: r.upstream_tab_id,
|
|
url: r.url,
|
|
title: r.title,
|
|
status: r.status as BrowserTab["status"],
|
|
lastSnapshotAt: iso(r.last_snapshot_at),
|
|
closedAt: iso(r.closed_at),
|
|
createdAt: iso(r.created_at)!,
|
|
updatedAt: iso(r.updated_at)!,
|
|
};
|
|
}
|
|
|
|
export function toLease(r: BrowserControlLeaseRow): BrowserControlLease {
|
|
return {
|
|
id: r.id,
|
|
sessionId: r.session_id,
|
|
holderKind: r.holder_kind as BrowserControlLease["holderKind"],
|
|
holder: r.holder,
|
|
reason: r.reason,
|
|
status: r.status as BrowserControlLease["status"],
|
|
acquiredAt: iso(r.acquired_at)!,
|
|
expiresAt: iso(r.expires_at)!,
|
|
releasedAt: iso(r.released_at),
|
|
releaseReason: r.release_reason,
|
|
createdAt: iso(r.created_at)!,
|
|
updatedAt: iso(r.updated_at)!,
|
|
};
|
|
}
|
|
|
|
/** 视觉子流程可以接手的等待原因(验证码 / 页面挑战 / 上次没处理成功) */
|
|
export const SOLVABLE = ["captcha", "challenge", "challenge_unsolved", "cf_challenge"];
|
|
|
|
/** capture / visionAct 的结果(能力输出的公共部分) */
|
|
export interface VisionShot {
|
|
evidence: BrowserEvidenceRef | null;
|
|
url: string | null;
|
|
controller: BrowserController;
|
|
sensitive: boolean;
|
|
imageForModel: boolean;
|
|
note: string | null;
|
|
textSnapshot: string | null;
|
|
coordinates: BrowserCoordinateSpace | null;
|
|
waitingReason: string | null;
|
|
modelImage: { mimeType: "image/png" | "image/jpeg"; data: string; width: number; height: number } | null;
|
|
}
|
|
|
|
/** 模型坐标系:缩放后的截图 ↔ 页面视口;取不到页面度量时按原始截图尺寸、DPR 1 推定 */
|
|
export function coordinateSpace(
|
|
src: { width: number; height: number },
|
|
img: { width: number; height: number },
|
|
m: PageMetrics | null,
|
|
): BrowserCoordinateSpace {
|
|
return {
|
|
imageWidth: img.width,
|
|
imageHeight: img.height,
|
|
sourceWidth: src.width,
|
|
sourceHeight: src.height,
|
|
viewportWidth: m?.viewportWidth ?? src.width,
|
|
viewportHeight: m?.viewportHeight ?? src.height,
|
|
devicePixelRatio: m?.devicePixelRatio ?? 1,
|
|
assumed: !m,
|
|
};
|
|
}
|
|
|
|
/** 截图像素(交给模型的缩放图)→ 页面 CSS 像素:按像素中心换算,四舍五入 */
|
|
export function screenToPage(sp: BrowserCoordinateSpace, x: number, y: number): { pageX: number; pageY: number } {
|
|
return {
|
|
pageX: Math.round(((x + 0.5) * sp.viewportWidth) / sp.imageWidth - 0.5),
|
|
pageY: Math.round(((y + 0.5) * sp.viewportHeight) / sp.imageHeight - 0.5),
|
|
};
|
|
}
|
|
|
|
/** 截图像素尺寸:PNG IHDR / JPEG SOF */
|
|
export function imageSize(data: Uint8Array): { width: number | null; height: number | null } {
|
|
if (data.length > 24 && data[0] === 0x89 && data[1] === 0x50) {
|
|
const v = new DataView(data.buffer, data.byteOffset, data.byteLength);
|
|
return { width: v.getUint32(16), height: v.getUint32(20) };
|
|
}
|
|
if (data[0] === 0xff && data[1] === 0xd8) {
|
|
let i = 2;
|
|
while (i + 9 < data.length) {
|
|
if (data[i] !== 0xff) break;
|
|
const marker = data[i + 1]!;
|
|
const len = (data[i + 2]! << 8) | data[i + 3]!;
|
|
if (marker >= 0xc0 && marker <= 0xc3)
|
|
return { width: (data[i + 7]! << 8) | data[i + 8]!, height: (data[i + 5]! << 8) | data[i + 6]! };
|
|
i += 2 + len;
|
|
}
|
|
}
|
|
return { width: null, height: null };
|
|
}
|
|
|
|
/** 代理地址 → 浏览器上下文代理(账号口令拆出来;Playwright 不认 URL 里的口令) */
|
|
export function egressProxy(url: string): BrowserEgressProxy | null {
|
|
try {
|
|
const u = new URL(url);
|
|
if (!["http:", "https:", "socks4:", "socks5:"].includes(u.protocol) || !u.hostname) return null;
|
|
return {
|
|
server: `${u.protocol}//${u.host}`,
|
|
...(u.username ? { username: decodeURIComponent(u.username) } : {}),
|
|
...(u.password ? { password: decodeURIComponent(u.password) } : {}),
|
|
};
|
|
} catch {
|
|
return null;
|
|
}
|
|
}
|