specific-ai-cli 0.0.0-stage → 0.1.1

package/dist/platforms/douyin/web/live.js

@@ -0,0 +1,225 @@
+import protobuf from 'protobufjs';
+import { SpecificAIError } from '../../../core/errors.js';
+import * as n from '../../../core/normalize.js';
+import { jsonLoads, urlencode } from '../../../core/py.js';
+import { openSocket, reconnecting } from '../../../core/stream.js';
+import * as api from './api.js';
+import { frontierUrl } from './im.js';
+import * as norm from './normalize.js';
+import * as proto from './proto.js';
+import { APP_VERSION, LIVE, Params, PROFILE, WWW } from './profile.js';
+import { liveSignature } from './sign.js';
+/** im/fetch 的响应（LiveResponse）；不是 protobuf 时多为登录态失效或被风控，报 UPSTREAM。 */
+export function decodeFetch(raw) {
+    try {
+        return proto.decode('Live', 'LiveResponse', proto.inflate(raw));
+    }
+    catch {
+        throw new SpecificAIError('UPSTREAM', `im/fetch 返回的不是 protobuf（${raw.length} 字节），多为登录态失效或被风控`, { detail: { bytes: raw.length } });
+    }
+}
+/** 先 im/fetch 取 cursor / internalExt，再拼 wss 地址（上游 DouyinLive.start_ws）。 */
+export async function liveSocket(d, info, webRid) {
+    const fetched = decodeFetch(await api.webcastFetch(d, info.user_id, info.room_id, `${LIVE}/${webRid}`));
+    const p = new Params();
+    p.add('app_name', 'douyin_web').add('version_code', '180800').add('webcast_sdk_version', '1.0.15').add('update_version_code', '1.0.15');
+    p.add('compress', 'gzip').add('device_platform', 'web').add('cookie_enabled', 'true').add('screen_width', '1707').add('screen_height', '960');
+    p.add('browser_language', 'zh-CN').add('browser_platform', 'Win32').add('browser_name', 'Mozilla').add('browser_version', APP_VERSION);
+    p.add('browser_online', 'true').add('tz_name', 'Etc/GMT-8').add('cursor', fetched.cursor ?? '').add('internal_ext', fetched.internalExt ?? '');
+    p.add('host', LIVE).add('aid', '6383').add('live_id', '1').add('did_rule', '3').add('endpoint', 'live_pc').add('support_wrds', '1');
+    p.add('user_unique_id', info.user_id).add('im_path', '/webcast/im/fetch/').add('identity', 'audience').add('need_persist_msg_count', '15');
+    p.add('insert_task_id', '').add('live_reason', '').add('room_id', info.room_id).add('heartbeatDuration', '0');
+    p.add('signature', liveSignature(d.xb, info.room_id, info.user_id));
+    return {
+        url: `wss://webcast100-ws-web-hl.douyin.com/webcast/im/push/v2/?${urlencode(p.pairs())}`,
+        headers: {
+            Pragma: 'no-cache',
+            'Accept-Language': 'zh-CN,zh;q=0.9,en;q=0.8,en-GB;q=0.7,en-US;q=0.6',
+            'User-Agent': PROFILE.ua,
+            'Cache-Control': 'no-cache',
+            Cookie: d.cookieStr,
+            Origin: LIVE,
+        },
+    };
+}
+const LIVE_TYPES = {
+    WebcastChatMessage: 'ChatMessage',
+    WebcastGiftMessage: 'GiftMessage',
+    WebcastMemberMessage: 'MemberMessage',
+    WebcastLikeMessage: 'LikeMessage',
+    WebcastSocialMessage: 'SocialMessage',
+    WebcastRoomStatsMessage: 'RoomStatsMessage',
+};
+/**
+ * 消息的发送时间：各 Webcast*Message 的字段 1 是公共头 Common，其中字段 4 是 createTime（毫秒）。
+ * 随包的 Live.proto（上游原样）没有声明 Common，这里用 protobufjs 的 Reader 直接读，取不到时为 null。
+ */
+export function messageTime(payload) {
+    try {
+        const r = protobuf.Reader.create(payload);
+        while (r.pos < r.len) {
+            const tag = r.uint32();
+            if (tag >>> 3 !== 1 || (tag & 7) !== 2) {
+                r.skipType(tag & 7);
+                continue;
+            }
+            const common = protobuf.Reader.create(r.bytes());
+            while (common.pos < common.len) {
+                const t = common.uint32();
+                if (t >>> 3 === 4 && (t & 7) === 0)
+                    return n.time(common.uint64().toString());
+                common.skipType(t & 7);
+            }
+            return null;
+        }
+    }
+    catch { }
+    return null;
+}
+/** LiveResponse 的 messagesList → Event[]，不关心的消息跳过。时间取消息自带的 createTime。 */
+function responseEvents(res) {
+    const events = [];
+    for (const item of res.messagesList ?? []) {
+        const name = LIVE_TYPES[item.method];
+        if (!name)
+            continue;
+        const event = norm.liveEvent(item.method, proto.decode('Live', name, item.payload));
+        if (!event)
+            continue;
+        const time = messageTime(item.payload);
+        if (time)
+            event.time = time;
+        events.push(event);
+    }
+    return events;
+}
+/** 一帧推送 → 事件；需要 ack 时返回 ack 帧（上游 on_message）。 */
+export function liveFrame(raw) {
+    const t = proto.type('Live', 'PushFrame');
+    const frame = t.decode(raw);
+    if (frame.payloadType === 'hb' || frame.payloadType === 'ack' || !frame.payload?.length)
+        return { events: [], ack: null };
+    const res = proto.decode('Live', 'LiveResponse', proto.inflate(frame.payload));
+    const ack = res.needAck ? t.encode(t.fromObject({ payloadType: 'ack', payload: Buffer.from(String(res.internalExt ?? ''), 'utf8'), logId: frame.logId })).finish() : null;
+    return { events: responseEvents(res), ack };
+}
+/** 进房时 im/fetch 的响应（LiveResponse，need_persist_msg_count=15）里带回的最近消息。 */
+export function fetchEvents(raw) {
+    return responseEvents(decodeFetch(raw));
+}
+export const heartbeatFrame = () => proto.encode('Live', 'PushFrame', { payloadType: 'hb' });
+/** 直播间页面解析不出房间信息：房间号不对，或者直播已经结束。 */
+export function roomNotFound(webRid) {
+    return new SpecificAIError('UPSTREAM', `未能解析直播间信息：${webRid}`, { hint: '确认直播间号，或者直播已经结束' });
+}
+/**
+ * 弹幕长连。进入重连循环之前先解析一次房间：房间号无效时直接报 UPSTREAM，不当成断线一直重连。
+ * 重连时重新取房间信息（重新开播后 room_id 会变），取不到同样报 UPSTREAM，由 core 连续 5 次后放弃。
+ */
+export async function* listenLive(ctx, d, webRid) {
+    const first = await api.liveInfo(d, webRid);
+    if (!first)
+        throw roomNotFound(webRid);
+    yield* reconnecting(ctx, async function* (attempt) {
+        const info = attempt === 0 ? first : await api.liveInfo(d, webRid);
+        if (!info)
+            throw roomNotFound(webRid);
+        const s = await liveSocket(d, info, webRid);
+        const socket = await openSocket(s.url, { headers: s.headers, proxy: ctx.config.proxy ?? undefined, signal: ctx.signal });
+        const beat = setInterval(() => void socket.send(heartbeatFrame()).catch(() => { }), 5000).unref();
+        try {
+            for await (const raw of socket.messages) {
+                if (typeof raw === 'string')
+                    continue;
+                try {
+                    const { events, ack } = liveFrame(raw);
+                    if (ack)
+                        await socket.send(ack).catch(() => { });
+                    yield* events;
+                }
+                catch (err) {
+                    ctx.log.debug(`解析直播消息失败：${err.message}`);
+                }
+            }
+        }
+        finally {
+            clearInterval(beat);
+            socket.close();
+        }
+    });
+}
+// ================================================================ 私信
+/** frontier-im 的 wss 地址与握手头（上游 DouyinRecvMsg）。 */
+export async function frontierSocket(d) {
+    const deviceId = await api.deviceId(d);
+    return {
+        url: frontierUrl(String(deviceId), d.cookie('sessionid') ?? ''),
+        headers: {
+            Pragma: 'no-cache',
+            'Accept-Language': 'zh-CN,zh;q=0.9,en;q=0.8,en-GB;q=0.7,en-US;q=0.6',
+            'User-Agent': PROFILE.ua,
+            'Cache-Control': 'no-cache',
+            'Sec-WebSocket-Protocol': 'binary, base64, pbbp2',
+            Cookie: d.cookieStr,
+            Origin: WWW,
+        },
+    };
+}
+const MSG_TYPES = { 7: 'text', 27: 'image', 30: 'video', 8: 'card', 77: 'card', 25: 'card', 26: 'card' };
+/** 新消息推送 → Message；已读回执等返回 null（上游 on_message）。 */
+export function imFrame(raw) {
+    const frame = proto.decode('Live', 'PushFrame', raw);
+    if (frame.payloadType !== 'pb' || !frame.payload)
+        return null;
+    const res = proto.decode('Response', 'Response', frame.payload);
+    const m = res.body?.new_message_notify?.message;
+    if (!m)
+        return null;
+    const type = Number(m.message_type);
+    if (type === 50001)
+        return null;
+    let content = {};
+    try {
+        content = jsonLoads(m.content ?? '{}');
+    }
+    catch { }
+    const media = [];
+    const url = (x) => n.url(x?.origin_url_list?.[0] ?? x?.url_list?.[0]);
+    if (type === 27 && url(content.resource_url))
+        media.push(n.media({ type: 'image', url: url(content.resource_url) }));
+    if (type === 5 && url(content.url))
+        media.push(n.media({ type: 'image', url: url(content.url) }));
+    if (type === 17 && url(content.resource_url))
+        media.push(n.media({ type: 'audio', url: url(content.resource_url) }));
+    return n.message({
+        id: n.id(m.server_message_id),
+        conversation_id: String(m.conversation_id ?? ''),
+        from: n.userRef({ id: m.sender }),
+        type: MSG_TYPES[type] ?? 'other',
+        text: type === 7 ? n.str(content.text) : type === 8 ? n.str(content.itemId) : null,
+        media,
+    }, m);
+}
+export function listenMessages(ctx, d) {
+    return reconnecting(ctx, async function* () {
+        const s = await frontierSocket(d);
+        const socket = await openSocket(s.url, { headers: s.headers, proxy: ctx.config.proxy ?? undefined, signal: ctx.signal });
+        try {
+            for await (const raw of socket.messages) {
+                if (typeof raw === 'string')
+                    continue;
+                try {
+                    const msg = imFrame(raw);
+                    if (msg)
+                        yield msg;
+                }
+                catch (err) {
+                    ctx.log.debug(`解析私信推送失败：${err.message}`);
+                }
+            }
+        }
+        finally {
+            socket.close();
+        }
+    });
+}

package/dist/platforms/douyin/web/mssdk.js

@@ -0,0 +1,148 @@
+import { readFileSync } from 'node:fs';
+import { staticFile } from '../../../core/paths.js';
+import { compactJson, jsonLoads } from '../../../core/py.js';
+import * as rand from '../../../core/rand.js';
+import { APP_VERSION, PROFILE } from './profile.js';
+import { encodeStrData } from './sign.js';
+/**
+ * mssdk 上报体（上游 utils/strdata_pure.py、utils/mstoken.py）。设备常量来自上游抓的档案，
+ * 原样复制在 static/douyin/ 下；每次只重算 uuid、采集耗时、时间戳与 nonce。
+ */
+const cache = new Map();
+function profile(name) {
+    if (!cache.has(name))
+        cache.set(name, JSON.parse(readFileSync(staticFile('douyin', name), 'utf8')));
+    return structuredClone(cache.get(name));
+}
+/** Python `json.dumps(float)`：整数值的浮点数带 `.0`。 */
+function pyFloat(x) {
+    return Number.isInteger(x) ? `${x}.0` : x;
+}
+/** 把 `"collectTime":"<n>.0"` 这类占位还原成数字字面量。 */
+function withRawFloats(json, raw) {
+    let out = json;
+    for (const [key, v] of Object.entries(raw))
+        if (typeof v === 'string')
+            out = out.replace(`"${key}":"${v}"`, `"${key}":${v}`);
+    return out;
+}
+/** /web/r/token 的上报体（上游 build_report_body）。 */
+export function strDataReport(aid = 6383, pageId = 6241) {
+    const data = profile('mstoken_profile.json');
+    const n = data.nWID;
+    const [, , , , availW, availH] = PROFILE.geo;
+    Object.assign(n.screen, { height: Number(PROFILE.screenHeight), width: Number(PROFILE.screenWidth), availHeight: availH, availWidth: availW, availTop: 0, availLeft: 0 });
+    n.navigator.hardwareConcurrency = Number(PROFILE.cpuCoreNum);
+    n.navigator.userAgent = PROFILE.ua;
+    n.webgl.renderer = PROFILE.webglRenderer;
+    n.webgl.vendor = PROFILE.webglVendor;
+    n.audio = n.audio ?? {};
+    n.audio.audioContext = n.audio.audioContext ?? {};
+    n.audio.audioContext.state = 'running';
+    const msVersion = n.ms_version ?? '0.0.0.1';
+    delete n.ms_version;
+    const uuid = rand.uuid4();
+    const collectTime = pyFloat(Number(rand.uniform(40, 160).toFixed(10)));
+    n.custom = withRawFloats(compactJson({ version: msVersion, fxgDid: '', uuid, collectTime, aid, pageId }), { collectTime });
+    n.ms_version = msVersion;
+    const plaintext = compactJson({ nWID: n, wID: data.wID });
+    const nonce = rand.randint(0, 255);
+    const ts = rand.now();
+    return compactJson({ magic: 538969122, version: 1, dataType: 8, strData: encodeStrData(plaintext, nonce), tspFromClient: ts, ulr: 0 });
+}
+/** /web/common 的完整 msgType=1 上报（上游 build_common_report_body）。 */
+export function commonReport(aid = 6383, pageId = 6241, sms = false) {
+    const p = profile('mstoken_common_profile.json');
+    const now = rand.now();
+    if (sms)
+        p.ubCode = 12;
+    const n = p.nWID ?? {};
+    if (sms)
+        n.ubCode = 12;
+    n.canvas = n.canvas ?? {};
+    n.canvas.crc32 = 'A7996F6A';
+    n.audio = n.audio ?? {};
+    n.audio.audioContext = n.audio.audioContext ?? {};
+    n.audio.audioContext.state = 'running';
+    let custom;
+    try {
+        custom = typeof n.custom === 'string' ? jsonLoads(n.custom) : { ...(n.custom ?? {}) };
+    }
+    catch {
+        custom = {};
+    }
+    custom.version ??= n.ms_version ?? '0.0.0.1';
+    custom.fxgDid ??= '';
+    custom.uuid = rand.uuid4();
+    custom.collectTime = sms ? 23.799999952316284 : rand.randint(10, 99);
+    const msVersion = n.ms_version ?? '0.0.0.1';
+    delete n.ms_version;
+    n.custom = compactJson(custom);
+    n.ms_version = msVersion;
+    const wid = p.wID ?? {};
+    Object.assign(wid, { msgType: 1, timestamp: String(now), aid, pageId, nap: '11311144242322244122' });
+    const g = PROFILE.geo;
+    Object.assign((p.navigator ??= {}), { appVersion: APP_VERSION, deviceMemory: PROFILE.deviceMemory, hardwareConcurrency: Number(PROFILE.cpuCoreNum) });
+    Object.assign((p.webgl ??= {}), { renderer: PROFILE.webglRenderer, vendor: PROFILE.webglVendor });
+    Object.assign((n.navigator ??= {}), { userAgent: PROFILE.ua, hardwareConcurrency: Number(PROFILE.cpuCoreNum) });
+    Object.assign((n.screen ??= {}), { height: Number(PROFILE.screenHeight), width: Number(PROFILE.screenWidth), availHeight: g[5], availWidth: g[4], availTop: 0, availLeft: 0 });
+    Object.assign((n.webgl ??= {}), { renderer: PROFILE.webglRenderer, vendor: PROFILE.webglVendor });
+    Object.assign((p.screen ??= {}), {
+        innerWidth: g[0],
+        innerHeight: g[1],
+        outerWidth: g[2],
+        outerHeight: g[3],
+        screenX: PROFILE.screenX,
+        screenY: PROFILE.screenY,
+        availWidth: g[4],
+        availHeight: g[5],
+        sizeWidth: g[6],
+        sizeHeight: g[7],
+        clientWidth: g[0],
+        clientHeight: g[1],
+    });
+    p.nWID = n;
+    p.wID = wid;
+    const nonce = rand.randint(0, 255);
+    const strData = encodeStrData(compactJson(p), nonce);
+    return compactJson({ magic: 538969122, version: 1, dataType: 8, strData, tspFromClient: sms ? now + 11 : now + 10, ulr: 0 });
+}
+/** 5 分钟一次的 msgType=2 行为心跳（上游 build_common_behavior_body）。 */
+export function commonBehavior() {
+    const now = rand.now();
+    const g = PROFILE.geo;
+    const plaintext = compactJson({
+        wID: { msgType: 2, privacyMode: 0, timestamp: String(now) },
+        behavior: {
+            beMove: [],
+            beClick: [],
+            beClickEnd: [],
+            beKeyboard: [],
+            windowState: [],
+            gyro: [],
+            focus: [],
+            screen: {
+                innerWidth: g[0],
+                innerHeight: g[1],
+                outerWidth: g[2],
+                outerHeight: g[3],
+                screenX: PROFILE.screenX,
+                screenY: PROFILE.screenY,
+                pageXOffset: 0,
+                pageYOffset: 0,
+                availWidth: g[4],
+                availHeight: g[5],
+                sizeWidth: g[6],
+                sizeHeight: g[7],
+                clientWidth: g[0],
+                clientHeight: g[1],
+                colorDepth: 24,
+                pixelDepth: 24,
+                orientaionType: 'landscape-primary',
+                orientaionAngle: 0,
+            },
+        },
+    });
+    const nonce = rand.randint(0, 255);
+    return compactJson({ magic: 538969122, version: 1, dataType: 8, strData: encodeStrData(plaintext, nonce), tspFromClient: now + 3, ulr: 0 });
+}