@chainlesschain/personal-data-hub 0.1.0 → 0.2.1

package/lib/adapters/ai-chat-history/vendors/coze.js

@@ -0,0 +1,179 @@
+/**
+ * 扣子 (Coze, ByteDance) vendor adapter — Phase 10.2 wiring.
+ *
+ * Reference: docs/design/Adapter_AIChat_History.md §6.7
+ *   - login   https://www.coze.cn/
+ *   - convs   GET /api/conversation/list
+ *   - msgs    GET /api/conversation/<id>/message
+ *   - cookies 字节通用 s_v_web_id (+ session)
+ *   - 特别 — agent 平台：tool_calls 多，workflow 可能嵌套。v1 flatten 为消息序列；
+ *             保留 extra.toolCalls 与 extra.botId.
+ */
+
+"use strict";
+
+const BASE = "https://www.coze.cn";
+const CONV_LIST_PATH = "/api/conversation/list";
+const MSG_LIST_PATH = (id) => `/api/conversation/${encodeURIComponent(id)}/message`;
+const USER_INFO_PATH = "/api/user/info";
+
+const DEFAULT_PAGE_SIZE = 30;
+
+function _ensureClient(ctx) {
+  if (!ctx || !ctx.httpClient) throw new Error("coze: ctx.httpClient required");
+  return ctx.httpClient;
+}
+
+async function validateCookie(ctx) {
+  const client = _ensureClient(ctx);
+  try {
+    const data = await client.getJson(BASE + USER_INFO_PATH, {
+      session: ctx.session,
+      matchDomain: "www.coze.cn",
+    });
+    if (data && data.code === 0 && data.data) {
+      return { ok: true, userId: data.data.user_id || data.data.uid };
+    }
+    return { ok: false, reason: "UNEXPECTED_RESPONSE_SHAPE" };
+  } catch (err) {
+    return { ok: false, reason: err.code || err.message };
+  }
+}
+
+async function *listConversations(ctx, opts = {}) {
+  const client = _ensureClient(ctx);
+  const pageSize = Number.isFinite(opts.pageSize) ? opts.pageSize : DEFAULT_PAGE_SIZE;
+  const sinceTs = opts.since && opts.since.lastUpdatedAt ? Number(opts.since.lastUpdatedAt) : 0;
+
+  let cursor = "0";
+  let safety = 0;
+  while (safety < 200) {
+    safety++;
+    const url = new URL(BASE + CONV_LIST_PATH);
+    url.searchParams.set("limit", String(pageSize));
+    url.searchParams.set("cursor", cursor);
+    const data = await client.getJson(url.toString(), {
+      session: ctx.session,
+      matchDomain: "www.coze.cn",
+    });
+    const items = _extractList(data);
+    if (items.length === 0) return;
+
+    let stopped = false;
+    for (const c of items) {
+      const updatedAt = _toMs(c.last_updated_time || c.updated_at || c.created_at);
+      if (sinceTs && updatedAt <= sinceTs) {
+        stopped = true;
+        break;
+      }
+      yield {
+        vendor: "coze",
+        originalId: String(c.conversation_id || c.id),
+        title: c.title || c.name || undefined,
+        modelName: undefined,
+        createdAt: _toMs(c.created_at || c.create_time),
+        updatedAt,
+        messageCount: c.message_count || undefined,
+        archived: Boolean(c.archived),
+        extra: { botId: c.bot_id, scene: c.scene },
+      };
+    }
+    if (stopped) return;
+    cursor = (data && data.data && data.data.next_cursor) || "";
+    if (!cursor || items.length < pageSize) return;
+  }
+}
+
+async function *listMessages(ctx, conversationId, _opts = {}) {
+  const client = _ensureClient(ctx);
+  const url = new URL(BASE + MSG_LIST_PATH(conversationId));
+  url.searchParams.set("limit", "200");
+  const data = await client.getJson(url.toString(), {
+    session: ctx.session,
+    matchDomain: "www.coze.cn",
+  });
+  const msgs = _extractList(data);
+
+  msgs.sort((a, b) => _toMs(a.created_at || a.create_time) - _toMs(b.created_at || b.create_time));
+
+  for (const m of msgs) {
+    yield {
+      vendor: "coze",
+      originalId: String(m.message_id || m.id),
+      conversationId: String(conversationId),
+      role: _normalizeRole(m.role || m.sender),
+      content: _buildContent(m),
+      createdAt: _toMs(m.created_at || m.create_time),
+      parentMessageId: m.parent_id ? String(m.parent_id) : undefined,
+      modelName: undefined,
+      extra: {
+        toolCalls: Array.isArray(m.tool_calls) ? m.tool_calls : undefined,
+        botId: m.bot_id,
+        workflowRunId: m.workflow_run_id,
+      },
+    };
+  }
+}
+
+function _extractList(data) {
+  if (!data) return [];
+  if (data.data && Array.isArray(data.data.list)) return data.data.list;
+  if (data.data && Array.isArray(data.data.conversations)) return data.data.conversations;
+  if (data.data && Array.isArray(data.data.messages)) return data.data.messages;
+  if (data.data && Array.isArray(data.data.message_list)) return data.data.message_list;
+  if (Array.isArray(data.data)) return data.data;
+  return [];
+}
+
+function _normalizeRole(r) {
+  if (r === "user" || r === "USER") return "user";
+  if (r === "assistant" || r === "ASSISTANT" || r === "bot") return "assistant";
+  if (r === "system" || r === "SYSTEM") return "system";
+  if (r === "tool" || r === "TOOL") return "tool";
+  return r ? String(r).toLowerCase() : "assistant";
+}
+
+function _buildContent(m) {
+  const content = { text: m.content || m.text || "" };
+  if (Array.isArray(m.attachments) && m.attachments.length > 0) {
+    content.attachments = m.attachments
+      .map((a) => ({
+        type: (a.type === "image" || /image/i.test(a.mime_type || "")) ? "image" : "file",
+        filename: a.name,
+        url: a.url,
+        mimeType: a.mime_type,
+        size: a.size,
+      }))
+      .filter((a) => a.url || a.filename);
+  }
+  return content;
+}
+
+function _toMs(t) {
+  if (typeof t === "number") return t > 1e12 ? t : t * 1000;
+  if (typeof t === "string") {
+    const n = Number(t);
+    if (Number.isFinite(n)) return n > 1e12 ? n : n * 1000;
+    const d = Date.parse(t);
+    return Number.isFinite(d) ? d : 0;
+  }
+  return 0;
+}
+
+const SPEC = {
+  name: "coze",
+  displayName: "扣子",
+  androidPackage: "com.coze.space",
+  loginUrl: "https://www.coze.cn/",
+  cookieDomains: ["www.coze.cn", ".coze.cn"],
+  rateLimits: { perMinute: 20, minIntervalMs: 2000 },
+
+  validateCookie,
+  listConversations,
+  listMessages,
+};
+
+module.exports = {
+  SPEC,
+  _internal: { _toMs, _normalizeRole, _buildContent, _extractList, BASE },
+};

package/lib/adapters/ai-chat-history/vendors/deepseek.js

@@ -0,0 +1,199 @@
+/**
+ * DeepSeek vendor adapter — Phase 10.2 wiring.
+ *
+ * Reference: docs/design/Adapter_AIChat_History.md §6.1
+ *   - login   https://chat.deepseek.com/
+ *   - convs   GET /api/v0/chat_session/fetch_page?count=N&before=<cursor>
+ *   - msgs    GET /api/v0/chat/history_messages?chat_session_id=<id>
+ *   - cookies userToken / session cookie
+ *
+ * The vendor exposes three async surfaces:
+ *
+ *   validateCookie(ctx)        — HEAD on userinfo endpoint, classifies cookie state.
+ *   listConversations(ctx, o)  — paginate via `before` cursor, yield RawConversation.
+ *   listMessages(ctx, id, o)   — return chronological RawMessage[] for one conversation.
+ *
+ * Each surface receives `ctx.httpClient` (a configured HttpClient bound to
+ * this vendor's rateLimits + cookie session). Tests pass a stub HttpClient
+ * with deterministic responses; production wires the real one via
+ * AIChatHistoryAdapter.
+ *
+ * Response shape notes (reverse-engineered, may shift — capture in fixtures
+ * before changing wire parser):
+ *
+ *   /chat_session/fetch_page →
+ *     { data: { biz_data: { chat_sessions: [{ id, title, model, inserted_at,
+ *       updated_at, current_message_id, ... }] } } }
+ *
+ *   /chat/history_messages →
+ *     { data: { biz_data: { chat_messages: [{ id, parent_id, role,
+ *       message_content, content, inserted_at, model, thinking_enabled,
+ *       files: [...] }] } } }
+ */
+
+"use strict";
+
+const BASE = "https://chat.deepseek.com";
+const CONV_PAGE_PATH = "/api/v0/chat_session/fetch_page";
+const MSG_PATH = "/api/v0/chat/history_messages";
+const USER_INFO_PATH = "/api/v0/user/get_user_info";
+
+const DEFAULT_PAGE_SIZE = 30;
+
+function _ensureClient(ctx) {
+  if (!ctx || !ctx.httpClient) {
+    throw new Error("deepseek: ctx.httpClient required (AIChatHistoryAdapter must wire one)");
+  }
+  return ctx.httpClient;
+}
+
+async function validateCookie(ctx) {
+  const client = _ensureClient(ctx);
+  try {
+    const data = await client.getJson(BASE + USER_INFO_PATH, { session: ctx.session });
+    if (data && data.code === 0 && data.data) {
+      return { ok: true, userId: data.data.biz_data && data.data.biz_data.user_id };
+    }
+    return { ok: false, reason: "UNEXPECTED_RESPONSE_SHAPE" };
+  } catch (err) {
+    return { ok: false, reason: err.code || err.message };
+  }
+}
+
+/**
+ * Yield one RawConversation per remote chat session, newest first. Pagination
+ * uses the `before` cursor returned by deepseek (the inserted_at of the last
+ * returned session). Stop when `chat_sessions` is empty OR we hit
+ * opts.since (timestamp watermark from prior sync).
+ */
+async function *listConversations(ctx, opts = {}) {
+  const client = _ensureClient(ctx);
+  const limit = Number.isFinite(opts.pageSize) ? opts.pageSize : DEFAULT_PAGE_SIZE;
+  const sinceTs = opts.since && opts.since.lastUpdatedAt ? Number(opts.since.lastUpdatedAt) : 0;
+  let cursor = null;
+
+  while (true) {
+    const url = new URL(BASE + CONV_PAGE_PATH);
+    url.searchParams.set("count", String(limit));
+    if (cursor != null) url.searchParams.set("before", String(cursor));
+
+    const data = await client.getJson(url.toString(), { session: ctx.session });
+    const sessions = data && data.data && data.data.biz_data && data.data.biz_data.chat_sessions;
+    if (!Array.isArray(sessions) || sessions.length === 0) return;
+
+    let stopped = false;
+    for (const s of sessions) {
+      const updatedAt = _toMs(s.updated_at || s.inserted_at);
+      if (sinceTs && updatedAt <= sinceTs) {
+        stopped = true;
+        break;
+      }
+      yield {
+        vendor: "deepseek",
+        originalId: String(s.id),
+        title: s.title || undefined,
+        modelName: s.model || undefined,
+        createdAt: _toMs(s.inserted_at),
+        updatedAt,
+        messageCount: s.message_count || undefined,
+        archived: false,
+        extra: {
+          currentMessageId: s.current_message_id,
+        },
+      };
+    }
+    if (stopped) return;
+    // Advance cursor — `before` expects the oldest timestamp from this page.
+    const last = sessions[sessions.length - 1];
+    cursor = last.inserted_at || last.updated_at;
+    if (!cursor) return;
+  }
+}
+
+/**
+ * Yield each message in a conversation. DeepSeek already returns messages in
+ * chronological order — we re-yield them as-is.
+ */
+async function *listMessages(ctx, conversationId, _opts = {}) {
+  const client = _ensureClient(ctx);
+  const url = new URL(BASE + MSG_PATH);
+  url.searchParams.set("chat_session_id", String(conversationId));
+
+  const data = await client.getJson(url.toString(), { session: ctx.session });
+  const msgs = data && data.data && data.data.biz_data && data.data.biz_data.chat_messages;
+  if (!Array.isArray(msgs)) return;
+
+  for (const m of msgs) {
+    yield {
+      vendor: "deepseek",
+      originalId: String(m.id),
+      conversationId: String(conversationId),
+      role: _normalizeRole(m.role),
+      content: _buildContent(m),
+      createdAt: _toMs(m.inserted_at),
+      parentMessageId: m.parent_id ? String(m.parent_id) : undefined,
+      modelName: m.model || undefined,
+      extra: {
+        thinking: m.thinking_content || undefined,
+        thinkingEnabled: Boolean(m.thinking_enabled),
+      },
+    };
+  }
+}
+
+function _normalizeRole(r) {
+  if (r === "USER" || r === "user") return "user";
+  if (r === "ASSISTANT" || r === "assistant") return "assistant";
+  if (r === "SYSTEM" || r === "system") return "system";
+  if (r === "TOOL" || r === "tool") return "tool";
+  return r || "assistant";
+}
+
+function _buildContent(m) {
+  const content = { text: m.content || m.message_content || "" };
+  if (Array.isArray(m.files) && m.files.length > 0) {
+    content.attachments = m.files
+      .map((f) => ({
+        type: f.kind === "image" ? "image" : "file",
+        filename: f.name || f.file_name,
+        url: f.url || f.download_url,
+        size: f.size,
+        mimeType: f.mime_type,
+      }))
+      .filter((a) => a.url || a.filename);
+  }
+  return content;
+}
+
+function _toMs(t) {
+  if (typeof t === "number") {
+    // deepseek uses seconds since epoch on this endpoint.
+    return t > 1e12 ? t : t * 1000;
+  }
+  if (typeof t === "string") {
+    const n = Number(t);
+    if (Number.isFinite(n)) return n > 1e12 ? n : n * 1000;
+    const d = Date.parse(t);
+    return Number.isFinite(d) ? d : 0;
+  }
+  return 0;
+}
+
+const SPEC = {
+  name: "deepseek",
+  displayName: "DeepSeek",
+  androidPackage: "com.deepseek.chat",
+  loginUrl: "https://chat.deepseek.com/",
+  cookieDomains: ["chat.deepseek.com", ".deepseek.com"],
+  rateLimits: { perMinute: 30, minIntervalMs: 1500 },
+
+  validateCookie,
+  listConversations,
+  listMessages,
+};
+
+module.exports = {
+  SPEC,
+  // exported for tests
+  _internal: { _toMs, _normalizeRole, _buildContent, BASE, CONV_PAGE_PATH, MSG_PATH },
+};

package/lib/adapters/ai-chat-history/vendors/doubao.js

@@ -0,0 +1,255 @@
+/**
+ * Doubao / 豆包 (ByteDance text AI) vendor adapter — Phase 10.2(+) scaffold.
+ *
+ * Doubao is ByteDance's flagship text AI assistant — sibling to Dreamina
+ * (image/video) but on a separate domain and surface. Treated as the 9th
+ * AIChatHistory vendor.
+ *
+ * Reference: docs/design/Adapter_AIChat_History.md §6.9 (added 2026-05-20).
+ *
+ *   - login        https://www.doubao.com/chat/
+ *   - convs        POST /samantha/conversation/list
+ *                  body: { cursor: "<opaque>", count: N }
+ *                  response: { data: { conversation_list: [...], cursor, has_more } }
+ *   - msgs         POST /samantha/conversation/<id>/message/list
+ *                  body: { conversation_id, cursor, count }
+ *                  response: { data: { message_list: [...], cursor, has_more } }
+ *   - user info    POST /samantha/user/info
+ *
+ * Phase 10.2(+) scope (this file): scaffold only.
+ *   - SPEC declares endpoint shape + rate limits + cookie domain
+ *   - validateCookie / listConversations / listMessages parse the documented
+ *     response shape but the exact field names are TBD until Phase 10.4
+ *     real-account fixture pin
+ *   - response shape uses defensive `_extractList` fallback to absorb minor
+ *     field-name drift without breaking sync
+ *
+ * Phase 10.4 will pin field names from real-account har capture and add the
+ * remaining response variants (search-result attachments, image gen inside
+ * Doubao, voice replies, etc).
+ */
+
+"use strict";
+
+const BASE = "https://www.doubao.com";
+const CONV_LIST_PATH = "/samantha/conversation/list";
+const MSG_LIST_PATH = (id) =>
+  `/samantha/conversation/${encodeURIComponent(id)}/message/list`;
+const USER_INFO_PATH = "/samantha/user/info";
+
+const DEFAULT_PAGE_SIZE = 30;
+
+function _ensureClient(ctx) {
+  if (!ctx || !ctx.httpClient) {
+    throw new Error("doubao: ctx.httpClient required (AIChatHistoryAdapter must wire one)");
+  }
+  return ctx.httpClient;
+}
+
+async function validateCookie(ctx) {
+  const client = _ensureClient(ctx);
+  try {
+    const data = await client.postJson(
+      BASE + USER_INFO_PATH,
+      {},
+      { session: ctx.session },
+    );
+    if (data && (data.code === 0 || data.code === "0") && data.data) {
+      return { ok: true, userId: data.data.user_id || data.data.uid };
+    }
+    return { ok: false, reason: "UNEXPECTED_RESPONSE_SHAPE" };
+  } catch (err) {
+    return { ok: false, reason: err.code || err.message };
+  }
+}
+
+/**
+ * Yield one RawConversation per remote chat session, newest first.
+ * Pagination uses an opaque `cursor` returned by the previous page plus
+ * `has_more` boolean (Doubao does not use offset / page numbers).
+ */
+async function *listConversations(ctx, opts = {}) {
+  const client = _ensureClient(ctx);
+  const limit = Number.isFinite(opts.pageSize) ? opts.pageSize : DEFAULT_PAGE_SIZE;
+  const sinceTs =
+    opts.since && opts.since.lastUpdatedAt ? Number(opts.since.lastUpdatedAt) : 0;
+
+  let cursor = "";
+  let safety = 0;
+  while (true) {
+    safety++;
+    if (safety > 200) return; // hard cap
+
+    const body = { count: limit };
+    if (cursor) body.cursor = cursor;
+
+    const data = await client.postJson(BASE + CONV_LIST_PATH, body, {
+      session: ctx.session,
+    });
+    const list = _extractConvList(data);
+    if (list.length === 0) return;
+
+    let stopped = false;
+    for (const c of list) {
+      const updatedAt = _toMs(c.last_message_time || c.update_time || c.create_time);
+      if (sinceTs && updatedAt <= sinceTs) {
+        stopped = true;
+        break;
+      }
+      yield {
+        vendor: "doubao",
+        originalId: String(c.conversation_id || c.id),
+        title: c.name || c.title || undefined,
+        modelName: c.bot_name || c.model || undefined,
+        createdAt: _toMs(c.create_time),
+        updatedAt,
+        messageCount: c.message_count || undefined,
+        archived: Boolean(c.archived || c.deleted),
+        extra: { botId: c.bot_id, botName: c.bot_name },
+      };
+    }
+    if (stopped) return;
+
+    const meta = (data && data.data) || {};
+    const hasMore = meta.has_more === true || meta.hasMore === true;
+    const nextCursor = meta.cursor || meta.next_cursor || "";
+    if (!hasMore || !nextCursor) return;
+    cursor = nextCursor;
+  }
+}
+
+/**
+ * Yield each message in a conversation in chronological order.
+ * Doubao paginates messages too — historical conversations can have
+ * thousands of turns.
+ */
+async function *listMessages(ctx, conversationId, _opts = {}) {
+  const client = _ensureClient(ctx);
+  let cursor = "";
+  let safety = 0;
+  const collected = [];
+
+  while (true) {
+    safety++;
+    if (safety > 200) break;
+
+    const body = { conversation_id: String(conversationId), count: 100 };
+    if (cursor) body.cursor = cursor;
+
+    const data = await client.postJson(BASE + MSG_LIST_PATH(conversationId), body, {
+      session: ctx.session,
+    });
+    const list = _extractMsgList(data);
+    if (list.length === 0) break;
+
+    for (const m of list) collected.push(m);
+
+    const meta = (data && data.data) || {};
+    const hasMore = meta.has_more === true || meta.hasMore === true;
+    const nextCursor = meta.cursor || meta.next_cursor || "";
+    if (!hasMore || !nextCursor) break;
+    cursor = nextCursor;
+  }
+
+  // Newest-first → reverse to chronological
+  collected.sort((a, b) => _toMs(a.create_time) - _toMs(b.create_time));
+
+  for (const m of collected) {
+    yield {
+      vendor: "doubao",
+      originalId: String(m.id || m.message_id),
+      conversationId: String(conversationId),
+      role: _normalizeRole(m.sender_type || m.role),
+      content: _buildContent(m),
+      createdAt: _toMs(m.create_time),
+      parentMessageId: m.parent_id ? String(m.parent_id) : undefined,
+      modelName: m.model || undefined,
+      extra: {
+        botId: m.bot_id,
+        thinking: m.thinking || undefined,
+      },
+    };
+  }
+}
+
+function _extractConvList(data) {
+  if (!data || !data.data) return [];
+  if (Array.isArray(data.data.conversation_list)) return data.data.conversation_list;
+  if (Array.isArray(data.data.conversations)) return data.data.conversations;
+  if (Array.isArray(data.data.list)) return data.data.list;
+  return [];
+}
+
+function _extractMsgList(data) {
+  if (!data || !data.data) return [];
+  if (Array.isArray(data.data.message_list)) return data.data.message_list;
+  if (Array.isArray(data.data.messages)) return data.data.messages;
+  if (Array.isArray(data.data.list)) return data.data.list;
+  return [];
+}
+
+function _normalizeRole(r) {
+  // Doubao uses sender_type "USER" / "ASSISTANT" / "SYSTEM" or numeric codes.
+  if (r === 1 || r === "1" || r === "USER" || r === "user") return "user";
+  if (r === 2 || r === "2" || r === "ASSISTANT" || r === "assistant") return "assistant";
+  if (r === 3 || r === "3" || r === "SYSTEM" || r === "system") return "system";
+  return r || "assistant";
+}
+
+function _buildContent(m) {
+  // Doubao messages can carry text + attachments (images uploaded by user,
+  // search refs, code blocks). For scaffold we keep text + attachment slot —
+  // Phase 10.4 will refine after fixture pin.
+  const text = m.content || m.text || m.message_content || "";
+  const out = { text };
+  if (Array.isArray(m.attachments) && m.attachments.length > 0) {
+    out.attachments = m.attachments
+      .map((a) => ({
+        type: a.kind === "image" ? "image" : "file",
+        filename: a.name || a.file_name,
+        url: a.url || a.download_url,
+        size: a.size,
+        mimeType: a.mime_type,
+      }))
+      .filter((a) => a.url || a.filename);
+  }
+  return out;
+}
+
+function _toMs(t) {
+  if (typeof t === "number") return t > 1e12 ? t : t * 1000;
+  if (typeof t === "string") {
+    const n = Number(t);
+    if (Number.isFinite(n)) return n > 1e12 ? n : n * 1000;
+    const d = Date.parse(t);
+    return Number.isFinite(d) ? d : 0;
+  }
+  return 0;
+}
+
+const SPEC = {
+  name: "doubao",
+  displayName: "豆包 Doubao",
+  androidPackage: "com.larus.nova",
+  loginUrl: "https://www.doubao.com/chat/",
+  cookieDomains: ["www.doubao.com", ".doubao.com"],
+  rateLimits: { perMinute: 20, minIntervalMs: 2000 },
+
+  validateCookie,
+  listConversations,
+  listMessages,
+};
+
+module.exports = {
+  SPEC,
+  _internal: {
+    _toMs,
+    _normalizeRole,
+    _buildContent,
+    _extractConvList,
+    _extractMsgList,
+    BASE,
+    CONV_LIST_PATH,
+    MSG_LIST_PATH,
+  },
+};