@chainlesschain/personal-data-hub 0.1.0 → 0.2.1

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

@@ -0,0 +1,193 @@
+/**
+ * 通义千问 (Aliyun Tongyi) vendor adapter — Phase 10.2 wiring.
+ *
+ * Reference: docs/design/Adapter_AIChat_History.md §6.3
+ *   - login  https://tongyi.aliyun.com/
+ *   - convs  POST /dialog/conversation/list
+ *   - msgs   POST /dialog/conversation/messages
+ *   - X-Csrf-Token header required (from XSRF-TOKEN cookie)
+ *   - 阿里通用 anti-bot — conservative rateLimits.
+ *
+ * Endpoints + payload shapes are reverse-engineered and SUBJECT TO DRIFT.
+ * Tests use the same fixture-fetch pattern as DeepSeek + Kimi so wire breakage
+ * surfaces immediately when a real cookie is reconnected.
+ */
+
+"use strict";
+
+const BASE = "https://tongyi.aliyun.com";
+const CONV_LIST_PATH = "/dialog/conversation/list";
+const MSG_LIST_PATH = "/dialog/conversation/messages";
+const USER_INFO_PATH = "/api/user/info";
+
+const DEFAULT_PAGE_SIZE = 30;
+
+function _ensureClient(ctx) {
+  if (!ctx || !ctx.httpClient) {
+    throw new Error("tongyi: ctx.httpClient required");
+  }
+  return ctx.httpClient;
+}
+
+function _csrfHeader(session) {
+  // Tongyi reads CSRF from a cookie + echoes it back as a header.
+  const value = session && (session.get("XSRF-TOKEN") || session.get("_csrf"));
+  return value ? { "X-Csrf-Token": value, "X-Xsrf-Token": value } : {};
+}
+
+async function validateCookie(ctx) {
+  const client = _ensureClient(ctx);
+  try {
+    const data = await client.getJson(BASE + USER_INFO_PATH, {
+      session: ctx.session,
+      headers: _csrfHeader(ctx.session),
+    });
+    if (data && (data.success || data.code === 200) && (data.data || data.userId)) {
+      const userId = (data.data && (data.data.userId || data.data.uid)) || data.userId;
+      return { ok: true, userId };
+    }
+    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 pageNum = 1;
+  while (true) {
+    const body = { pageNum, pageSize };
+    const data = await client.postJson(BASE + CONV_LIST_PATH, body, {
+      session: ctx.session,
+      headers: _csrfHeader(ctx.session),
+    });
+    const items = _extractList(data);
+    if (items.length === 0) return;
+
+    let stopped = false;
+    for (const c of items) {
+      const updatedAt = _toMs(c.gmtModified || c.updatedAt || c.gmtCreate);
+      if (sinceTs && updatedAt <= sinceTs) {
+        stopped = true;
+        break;
+      }
+      yield {
+        vendor: "tongyi",
+        originalId: String(c.sessionId || c.id),
+        title: c.summary || c.firstQuery || c.title || undefined,
+        modelName: c.modelName || c.model || undefined,
+        createdAt: _toMs(c.gmtCreate || c.createdAt),
+        updatedAt,
+        messageCount: c.messageCount || undefined,
+        archived: Boolean(c.archived),
+        extra: { agentName: c.agentName || c.botName },
+      };
+    }
+    if (stopped) return;
+    if (items.length < pageSize) return;
+    pageNum++;
+    if (pageNum > 200) return; // safety
+  }
+}
+
+async function *listMessages(ctx, conversationId, _opts = {}) {
+  const client = _ensureClient(ctx);
+  const body = { sessionId: String(conversationId), parentMsgId: "" };
+  const data = await client.postJson(BASE + MSG_LIST_PATH, body, {
+    session: ctx.session,
+    headers: _csrfHeader(ctx.session),
+  });
+  const msgs = _extractList(data);
+
+  // Tongyi returns messages mixed user/bot in their own order; we sort
+  // ascending by createTime for stable chronological yield.
+  msgs.sort((a, b) => _toMs(a.createTime || a.gmtCreate) - _toMs(b.createTime || b.gmtCreate));
+
+  for (const m of msgs) {
+    yield {
+      vendor: "tongyi",
+      originalId: String(m.msgId || m.id),
+      conversationId: String(conversationId),
+      role: _normalizeRole(m.senderType || m.role || (m.contentType === "user" ? "user" : "assistant")),
+      content: _buildContent(m),
+      createdAt: _toMs(m.createTime || m.gmtCreate),
+      parentMessageId: m.parentMsgId ? String(m.parentMsgId) : undefined,
+      modelName: m.modelName || m.model || undefined,
+      extra: m.feedback ? { feedback: m.feedback } : undefined,
+    };
+  }
+}
+
+function _extractList(data) {
+  if (!data) return [];
+  if (Array.isArray(data.data)) return data.data;
+  if (data.data && Array.isArray(data.data.list)) return data.data.list;
+  if (data.data && Array.isArray(data.data.records)) return data.data.records;
+  if (Array.isArray(data.list)) return data.list;
+  return [];
+}
+
+function _normalizeRole(r) {
+  if (r === "user" || r === "USER" || r === 1) return "user";
+  if (r === "assistant" || r === "ASSISTANT" || r === "bot" || r === 2) return "assistant";
+  if (r === "system" || r === "SYSTEM" || r === 0) return "system";
+  return r ? String(r).toLowerCase() : "assistant";
+}
+
+function _buildContent(m) {
+  // Tongyi messages have content fanned across `contents[].content` for
+  // multi-segment replies (image + text). Flatten to single text.
+  const segments = Array.isArray(m.contents) ? m.contents : null;
+  const text = segments
+    ? segments.map((s) => s.content || "").filter(Boolean).join("\n")
+    : (m.content || "");
+  const content = { text };
+
+  const files = []
+    .concat(Array.isArray(m.fileList) ? m.fileList : [])
+    .concat(Array.isArray(m.imageList) ? m.imageList : []);
+  if (files.length > 0) {
+    content.attachments = files
+      .map((f) => ({
+        type: (f.type === "image" || /image/i.test(f.fileType || "")) ? "image" : "file",
+        filename: f.fileName || f.name,
+        url: f.url || f.fileUrl,
+        size: f.size,
+        mimeType: f.fileType,
+      }))
+      .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: "tongyi",
+  displayName: "通义千问",
+  androidPackage: "com.aliyun.tongyi",
+  loginUrl: "https://tongyi.aliyun.com/",
+  cookieDomains: ["tongyi.aliyun.com", ".aliyun.com"],
+  rateLimits: { perMinute: 20, minIntervalMs: 2000 }, // Aliyun anti-bot — conservative
+
+  validateCookie,
+  listConversations,
+  listMessages,
+};
+
+module.exports = {
+  SPEC,
+  _internal: { _toMs, _normalizeRole, _buildContent, _csrfHeader, _extractList, BASE },
+};

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

@@ -0,0 +1,202 @@
+/**
+ * 智谱清言 (Zhipu Qingyan) vendor adapter — Phase 10.2 wiring.
+ *
+ * Reference: docs/design/Adapter_AIChat_History.md §6.4
+ *   - login  https://chatglm.cn/
+ *   - convs  GET /chatglm/backend-api/v1/conversation/list
+ *   - msgs   GET /chatglm/backend-api/v1/conversation/<id>
+ *   - cookie chatglm_token (Bearer-style — re-projected from cookie when
+ *            present; otherwise relies on cookie auth alone)
+ *
+ * GLM-4 messages have optional `tool_calls` for web-search injection; we
+ * thread those into RawMessage.extra.toolCalls so schema-map preserves them
+ * across the KG / RAG flow.
+ *
+ * Endpoint shapes are reverse-engineered + SUBJECT TO DRIFT.
+ */
+
+"use strict";
+
+const BASE = "https://chatglm.cn";
+const CONV_LIST_PATH = "/chatglm/backend-api/v1/conversation/list";
+const CONV_DETAIL_PATH = "/chatglm/backend-api/v1/conversation/";
+const USER_INFO_PATH = "/chatglm/backend-api/v1/user/info";
+
+const DEFAULT_PAGE_SIZE = 30;
+
+function _ensureClient(ctx) {
+  if (!ctx || !ctx.httpClient) {
+    throw new Error("zhipu: ctx.httpClient required");
+  }
+  return ctx.httpClient;
+}
+
+function _authHeader(session) {
+  const token = session && session.get("chatglm_token");
+  return token ? { Authorization: `Bearer ${token}` } : {};
+}
+
+async function validateCookie(ctx) {
+  const client = _ensureClient(ctx);
+  try {
+    const data = await client.getJson(BASE + USER_INFO_PATH, {
+      session: ctx.session,
+      headers: _authHeader(ctx.session),
+    });
+    if (data && data.status === 0 && data.result) {
+      return { ok: true, userId: data.result.user_id || data.result.id };
+    }
+    if (data && (data.user_id || data.id)) {
+      return { ok: true, userId: data.user_id || data.id };
+    }
+    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 page = 1;
+  while (true) {
+    const url = new URL(BASE + CONV_LIST_PATH);
+    url.searchParams.set("page", String(page));
+    url.searchParams.set("page_size", String(pageSize));
+
+    const data = await client.getJson(url.toString(), {
+      session: ctx.session,
+      headers: _authHeader(ctx.session),
+    });
+    const list = _extractList(data);
+    if (list.length === 0) return;
+
+    let stopped = false;
+    for (const c of list) {
+      const updatedAt = _toMs(c.update_time || c.updateTime || c.create_time);
+      if (sinceTs && updatedAt <= sinceTs) {
+        stopped = true;
+        break;
+      }
+      yield {
+        vendor: "zhipu",
+        originalId: String(c.conversation_id || c.id),
+        title: c.title || c.summary || undefined,
+        modelName: c.model || c.model_name || undefined,
+        createdAt: _toMs(c.create_time || c.createTime),
+        updatedAt,
+        messageCount: c.message_count || undefined,
+        archived: Boolean(c.archived),
+        extra: { assistantId: c.assistant_id },
+      };
+    }
+    if (stopped) return;
+    if (list.length < pageSize) return;
+    page++;
+    if (page > 200) return;
+  }
+}
+
+async function *listMessages(ctx, conversationId, _opts = {}) {
+  const client = _ensureClient(ctx);
+  const url = BASE + CONV_DETAIL_PATH + encodeURIComponent(conversationId);
+  const data = await client.getJson(url, {
+    session: ctx.session,
+    headers: _authHeader(ctx.session),
+  });
+
+  // GLM detail returns { result: { messages: [...] } } or { messages: [...] }.
+  let msgs = [];
+  if (data && data.result && Array.isArray(data.result.messages)) msgs = data.result.messages;
+  else if (data && Array.isArray(data.messages)) msgs = data.messages;
+  else if (data && data.result && Array.isArray(data.result.history)) msgs = data.result.history;
+
+  msgs.sort((a, b) => _toMs(a.create_time || a.timestamp) - _toMs(b.create_time || b.timestamp));
+
+  for (const m of msgs) {
+    yield {
+      vendor: "zhipu",
+      originalId: String(m.id || m.message_id),
+      conversationId: String(conversationId),
+      role: _normalizeRole(m.role || m.sender),
+      content: _buildContent(m),
+      createdAt: _toMs(m.create_time || m.timestamp),
+      parentMessageId: m.parent_id ? String(m.parent_id) : undefined,
+      modelName: m.model || undefined,
+      extra: {
+        toolCalls: Array.isArray(m.tool_calls) ? m.tool_calls : undefined,
+        searchQuery: m.search_query,
+      },
+    };
+  }
+}
+
+function _extractList(data) {
+  if (!data) return [];
+  if (Array.isArray(data.result)) return data.result;
+  if (data.result && Array.isArray(data.result.list)) return data.result.list;
+  if (data.result && Array.isArray(data.result.conversations)) return data.result.conversations;
+  if (Array.isArray(data.conversations)) return data.conversations;
+  if (Array.isArray(data.list)) return data.list;
+  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 || a.file_name,
+        url: a.url || a.download_url,
+        size: a.size,
+        mimeType: a.mime_type,
+      }))
+      .filter((a) => a.url || a.filename);
+  }
+  if (Array.isArray(m.images) && m.images.length > 0) {
+    content.attachments = (content.attachments || []).concat(
+      m.images.map((i) => ({ type: "image", url: i.url || i, mimeType: "image/png" })),
+    );
+  }
+  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: "zhipu",
+  displayName: "智谱清言",
+  androidPackage: "com.zhipuai.qingyan",
+  loginUrl: "https://chatglm.cn/",
+  cookieDomains: ["chatglm.cn", ".chatglm.cn"],
+  rateLimits: { perMinute: 20, minIntervalMs: 2000 },
+
+  validateCookie,
+  listConversations,
+  listMessages,
+};
+
+module.exports = {
+  SPEC,
+  _internal: { _toMs, _normalizeRole, _buildContent, _authHeader, _extractList, BASE },
+};