@soimy/dingtalk 3.1.4 → 3.3.0

package/src/message-utils.ts

@@ -1,4 +1,101 @@
-import type { DingTalkInboundMessage, MessageContent, SendMessageOptions } from "./types";
+import type { DingTalkInboundMessage, MessageContent, QuotedInfo, SendMessageOptions } from "./types";
+
+interface DingTalkDocMeta {
+  spaceId: string;
+  fileId: string;
+}
+
+function parseBizCustomActionUrl(url: string | undefined): DingTalkDocMeta | null {
+  if (!url || typeof url !== "string") {
+    return null;
+  }
+
+  const queryIndex = url.indexOf("?");
+  if (queryIndex < 0 || queryIndex === url.length - 1) {
+    return null;
+  }
+
+  try {
+    const params = new URLSearchParams(url.slice(queryIndex + 1));
+    const route = params.get("route");
+    const type = params.get("type");
+    const spaceId = params.get("spaceId");
+    const fileId = params.get("fileId");
+    if (route !== "previewDentry" || type !== "file" || !spaceId || !fileId) {
+      return null;
+    }
+    return { spaceId, fileId };
+  } catch {
+    return null;
+  }
+}
+
+function extractRichTextQuoteParts(
+  richText: Array<Record<string, unknown>> | undefined,
+): { summary: string; pictureDownloadCode?: string; pictureDownloadCodes?: string[] } | null {
+  if (!Array.isArray(richText) || richText.length === 0) {
+    return null;
+  }
+
+  const textParts: string[] = [];
+  const pictureDownloadCodes: string[] = [];
+
+  for (const part of richText) {
+    const partType =
+      typeof part.msgType === "string"
+        ? part.msgType
+        : typeof part.type === "string"
+          ? part.type
+          : undefined;
+    const textValue =
+      typeof part.content === "string"
+        ? part.content
+        : typeof part.text === "string"
+          ? part.text
+          : undefined;
+
+    if ((partType === "text" || partType === undefined) && textValue) {
+      textParts.push(textValue);
+      continue;
+    }
+    if (partType === "emoji" && textValue) {
+      textParts.push(textValue);
+      continue;
+    }
+    if (partType === "picture") {
+      textParts.push("[图片]");
+      if (typeof part.downloadCode === "string" && part.downloadCode.trim()) {
+        pictureDownloadCodes.push(part.downloadCode.trim());
+      }
+      continue;
+    }
+    if (partType === "at") {
+      const atName =
+        typeof part.atName === "string"
+          ? part.atName
+          : typeof textValue === "string"
+            ? textValue
+            : "某人";
+      textParts.push(`@${atName}`);
+      continue;
+    }
+    if (textValue) {
+      textParts.push(textValue);
+    }
+  }
+
+  const summary = textParts.join("").trim();
+  const uniquePictureDownloadCodes = [...new Set(pictureDownloadCodes)];
+  const pictureDownloadCode = uniquePictureDownloadCodes[0];
+  if (!summary && !pictureDownloadCode) {
+    return null;
+  }
+  return {
+    summary,
+    pictureDownloadCode,
+    pictureDownloadCodes: uniquePictureDownloadCodes.length > 0 ? uniquePictureDownloadCodes : undefined,
+  };
+}
 
 /**
  * Auto-detect markdown usage and derive message title.
@@ -24,76 +121,199 @@ export function detectMarkdownAndExtractTitle(
   return { useMarkdown, title };
 }
 
+function isMarkdownTableSeparator(line: string): boolean {
+  const normalized = line.trim();
+  if (!normalized.includes("-")) {
+    return false;
+  }
+  const cells = normalized
+    .replace(/^\|/, "")
+    .replace(/\|$/, "")
+    .split("|")
+    .map((cell) => cell.trim());
+  return cells.length > 0 && cells.every((cell) => /^:?-{3,}:?$/.test(cell));
+}
+
+function isMarkdownTableRow(line: string): boolean {
+  const trimmed = line.trim();
+  return trimmed.includes("|") && !trimmed.startsWith("```");
+}
+
+function parseMarkdownTableRow(line: string): string[] {
+  return line
+    .trim()
+    .replace(/^\|/, "")
+    .replace(/\|$/, "")
+    .split("|")
+    .map((cell) => cell.trim());
+}
+
+function renderMarkdownTable(lines: string[]): string {
+  const rows = lines.map(parseMarkdownTableRow).filter((cells) => cells.length > 0);
+  return rows.map((cells) => cells.join(" | ")).join("\n");
+}
+
+export function convertMarkdownTablesToPlainText(text: string): string {
+  const lines = text.split("\n");
+  const output: string[] = [];
+  let index = 0;
+  let inCodeFence = false;
+
+  while (index < lines.length) {
+    const line = lines[index] || "";
+    if (line.trim().startsWith("```")) {
+      inCodeFence = !inCodeFence;
+      output.push(line);
+      index += 1;
+      continue;
+    }
+
+    if (
+      !inCodeFence &&
+      index + 1 < lines.length &&
+      isMarkdownTableRow(line) &&
+      isMarkdownTableSeparator(lines[index + 1] || "")
+    ) {
+      const tableLines = [line];
+      index += 2;
+      while (index < lines.length && isMarkdownTableRow(lines[index] || "")) {
+        tableLines.push(lines[index] || "");
+        index += 1;
+      }
+      output.push(renderMarkdownTable(tableLines));
+      continue;
+    }
+
+    output.push(line);
+    index += 1;
+  }
+
+  return output.join("\n");
+}
+
 export function extractMessageContent(data: DingTalkInboundMessage): MessageContent {
   const msgtype = data.msgtype || "text";
 
-  // Normalize quote/reply metadata into a readable text prefix so the agent can understand message context.
-  const formatQuotedContent = (): string => {
-    const textField = data.text as any;
+  const formatQuotedContent = (): QuotedInfo | null => {
+    const textField = data.text;
 
     if (textField?.isReplyMsg && textField?.repliedMsg) {
       const repliedMsg = textField.repliedMsg;
-      const content = repliedMsg?.content;
+      const repliedMsgType = repliedMsg.msgType;
+      const content = repliedMsg.content;
+
+      if (repliedMsgType === "text" && content?.text?.trim()) {
+        return { prefix: `[引用消息: "${content.text.trim()}"]\n\n` };
+      }
 
-      if (content?.text) {
-        const quoteText = content.text.trim();
-        if (quoteText) {
-          return `[引用消息: "${quoteText}"]\n\n`;
+      if (repliedMsgType === "picture" && content?.downloadCode) {
+        return {
+          prefix: "[引用图片]\n\n",
+          mediaDownloadCode: content.downloadCode,
+          mediaType: "image",
+        };
+      }
+
+      if (repliedMsgType === "richText") {
+        const richTextQuote = extractRichTextQuoteParts(content?.richText);
+        if (richTextQuote) {
+          const quoteImageCount = richTextQuote.pictureDownloadCodes?.length || 0;
+          const prefix =
+            richTextQuote.summary && richTextQuote.summary !== "[图片]"
+              ? `[引用消息: "${richTextQuote.summary}"]${quoteImageCount > 1 ? ` [含${quoteImageCount}张引用图片]` : ""}\n\n`
+              : "[引用图片]\n\n";
+          return {
+            prefix,
+            mediaDownloadCode: richTextQuote.pictureDownloadCode,
+            mediaType: richTextQuote.pictureDownloadCode ? "image" : undefined,
+          };
         }
       }
 
-      if (content?.richText && Array.isArray(content.richText)) {
-        const textParts: string[] = [];
-        for (const part of content.richText) {
-          if (part.msgType === "text" && part.content) {
-            textParts.push(part.content);
-          } else if (part.msgType === "emoji" || part.type === "emoji") {
-            textParts.push(part.content || "[表情]");
-          } else if (part.msgType === "picture" || part.type === "picture") {
-            textParts.push("[图片]");
-          } else if (part.msgType === "at" || part.type === "at") {
-            textParts.push(`@${part.content || part.atName || "某人"}`);
-          } else if (part.text) {
-            textParts.push(part.text);
-          }
+      if (repliedMsgType === "unknownMsgType") {
+        return {
+          prefix: "[引用文件]\n\n",
+          isQuotedFile: true,
+          fileCreatedAt: repliedMsg.createdAt,
+          msgId: repliedMsg.msgId,
+        };
+      }
+
+      if (repliedMsgType === "interactiveCard") {
+        const isBotCard = repliedMsg.senderId === data.chatbotUserId;
+        if (isBotCard) {
+          return {
+            prefix: "[引用了机器人的回复]\n\n",
+            isQuotedCard: true,
+            cardCreatedAt: repliedMsg.createdAt,
+            processQueryKey: data.originalProcessQueryKey,
+            msgId: repliedMsg.msgId,
+          };
         }
-        const quoteText = textParts.join("").trim();
-        if (quoteText) {
-          return `[引用消息: "${quoteText}"]\n\n`;
+
+        return {
+          prefix: "[引用了钉钉文档]\n\n",
+          isQuotedDocCard: true,
+          fileCreatedAt: repliedMsg.createdAt,
+          msgId: repliedMsg.msgId,
+        };
+      }
+
+      // Has msgType but not one we handle — generic fallback.
+      if (repliedMsgType) {
+        const idPart = repliedMsg.msgId ? `，原消息ID: ${repliedMsg.msgId}` : "";
+        return { prefix: `[引用消息不可见: msgType=${repliedMsgType}${idPart}]\n\n` };
+      }
+
+      // No msgType — backward compat: extract text or richText from content.
+      if (content?.text?.trim()) {
+        return { prefix: `[引用消息: "${content.text.trim()}"]\n\n` };
+      }
+
+      if (content?.richText && Array.isArray(content.richText)) {
+        const richTextQuote = extractRichTextQuoteParts(content.richText);
+        if (richTextQuote?.summary) {
+          return {
+            prefix: `[引用消息: "${richTextQuote.summary}"]\n\n`,
+            mediaDownloadCode: richTextQuote.pictureDownloadCode,
+            mediaType: richTextQuote.pictureDownloadCode ? "image" : undefined,
+          };
         }
       }
     }
 
-    // Some clients only send originalMsgId for rich media reply messages.
     if (textField?.isReplyMsg && !textField?.repliedMsg && data.originalMsgId) {
-      return `[这是一条引用消息，原消息ID: ${data.originalMsgId}]\n\n`;
+      return {
+        prefix: `[这是一条引用消息，原消息ID: ${data.originalMsgId}]\n\n`,
+        msgId: data.originalMsgId,
+      };
     }
 
     if (data.quoteMessage) {
       const quoteText = data.quoteMessage.text?.content?.trim() || "";
       if (quoteText) {
-        return `[引用消息: "${quoteText}"]\n\n`;
+        return { prefix: `[引用消息: "${quoteText}"]\n\n` };
       }
     }
 
     if (data.content?.quoteContent) {
-      return `[引用消息: "${data.content.quoteContent}"]\n\n`;
+      return { prefix: `[引用消息: "${data.content.quoteContent}"]\n\n` };
     }
 
-    return "";
+    return null;
   };
 
-  const quotedPrefix = formatQuotedContent();
+  const quoted = formatQuotedContent();
+  const quotedPrefix = quoted?.prefix || "";
 
-  // Unified extraction by DingTalk msgtype for downstream routing/agent processing.
   if (msgtype === "text") {
-    return { text: quotedPrefix + (data.text?.content?.trim() || ""), messageType: "text" };
+    return { text: quotedPrefix + (data.text?.content?.trim() || ""), messageType: "text", quoted: quoted ?? undefined };
   }
 
   if (msgtype === "richText") {
     const richTextParts = data.content?.richText || [];
     let text = "";
-    let pictureDownloadCode: string | undefined;
+    const pictureDownloadCodes: string[] = [];
     // Keep first image downloadCode while preserving readable text and @mention parts.
     for (const part of richTextParts) {
       if (part.text && (part.type === "text" || part.type === undefined)) {
@@ -102,16 +322,21 @@ export function extractMessageContent(data: DingTalkInboundMessage): MessageCont
       if (part.type === "at" && part.atName) {
         text += `@${part.atName} `;
       }
-      if (part.type === "picture" && part.downloadCode && !pictureDownloadCode) {
-        pictureDownloadCode = part.downloadCode;
+      if (part.type === "picture" && part.downloadCode) {
+        pictureDownloadCodes.push(part.downloadCode);
       }
     }
+    const uniquePictureDownloadCodes = [...new Set(pictureDownloadCodes)];
+    const pictureDownloadCode = uniquePictureDownloadCodes[0];
     return {
       text:
         quotedPrefix + (text.trim() || (pictureDownloadCode ? "<media:image>" : "[富文本消息]")),
       mediaPath: pictureDownloadCode,
+      mediaPaths: uniquePictureDownloadCodes.length > 0 ? uniquePictureDownloadCodes : undefined,
       mediaType: pictureDownloadCode ? "image" : undefined,
+      mediaTypes: uniquePictureDownloadCodes.length > 0 ? uniquePictureDownloadCodes.map(() => "image") : undefined,
       messageType: "richText",
+      quoted: quoted ?? undefined,
     };
   }
 
@@ -151,6 +376,43 @@ export function extractMessageContent(data: DingTalkInboundMessage): MessageCont
     };
   }
 
-  // Fallback: preserve unknown msgtype as readable marker.
-  return { text: data.text?.content?.trim() || `[${msgtype}消息]`, messageType: msgtype };
+  if (msgtype === "interactiveCard") {
+    const docMeta = parseBizCustomActionUrl(data.content?.biz_custom_action_url);
+    if (docMeta) {
+      return {
+        text: "[钉钉文档]\n\n",
+        messageType: "interactiveCardFile",
+        docSpaceId: docMeta.spaceId,
+        docFileId: docMeta.fileId,
+        quoted: quoted ?? undefined,
+      };
+    }
+    return {
+      text: data.text?.content?.trim() || "[interactiveCard消息]",
+      messageType: msgtype,
+      quoted: quoted ?? undefined,
+    };
+  }
+  if (msgtype === "chatRecord") {
+    const content = data.content as Record<string, unknown> | undefined;
+    const summary = typeof content?.summary === "string" ? content.summary.trim() : "";
+    const rawRecord = content?.chatRecord;
+    if (
+      summary === "[]" ||
+      (typeof rawRecord === "string" && rawRecord.trim() === "[]") ||
+      (Array.isArray(rawRecord) && rawRecord.length === 0)
+    ) {
+      return {
+        text: "[系统提示] 没有读到引用记录（chatRecord 为空）。请改用逐条转发、复制原文，或重新转发非空聊天记录。",
+        messageType: "chatRecord",
+        quoted: quoted ?? undefined,
+      };
+    }
+    if (summary) {
+      return { text: `[聊天记录摘要] ${summary}`, messageType: "chatRecord", quoted: quoted ?? undefined };
+    }
+    return { text: "[chatRecord消息: 无可读内容]", messageType: "chatRecord", quoted: quoted ?? undefined };
+  }
+
+  return { text: data.text?.content?.trim() || `[${msgtype}消息]`, messageType: msgtype, quoted: quoted ?? undefined };
 }

package/src/onboarding.ts

@@ -2,6 +2,7 @@ import type { OpenClawConfig, ChannelOnboardingAdapter, WizardPrompter } from "o
 import { DEFAULT_ACCOUNT_ID, normalizeAccountId, formatDocsLink } from "openclaw/plugin-sdk";
 import type { DingTalkConfig, DingTalkChannelConfig } from "./types.js";
 import { listDingTalkAccountIds, resolveDingTalkAccount } from "./types.js";
+import { DEFAULT_JOURNAL_TTL_DAYS } from "./quote-journal.js";
 
 const channel = "dingtalk" as const;
 
@@ -109,6 +110,11 @@ function applyAccountConfig(params: {
     ...(typeof input.maxReconnectCycles === "number"
       ? { maxReconnectCycles: input.maxReconnectCycles }
       : {}),
+    ...(typeof input.useConnectionManager === "boolean"
+      ? { useConnectionManager: input.useConnectionManager }
+      : {}),
+    ...(typeof input.mediaMaxMb === "number" ? { mediaMaxMb: input.mediaMaxMb } : {}),
+    ...(typeof input.journalTTLDays === "number" ? { journalTTLDays: input.journalTTLDays } : {}),
   };
 
   if (useDefault) {
@@ -271,10 +277,10 @@ export const dingtalkOnboardingAdapter: ChannelOnboardingAdapter = {
         String(
           await prompter.text({
             message: "Card Template Key (content field name)",
-            placeholder: "msgContent",
-            initialValue: resolved.cardTemplateKey ?? "msgContent",
+            placeholder: "content",
+            initialValue: resolved.cardTemplateKey ?? "content",
           }),
-        ).trim() || "msgContent";
+        ).trim() || "content";
 
       messageType = "card";
     }
@@ -298,6 +304,14 @@ export const dingtalkOnboardingAdapter: ChannelOnboardingAdapter = {
       allowFrom = parsed.length > 0 ? parsed : undefined;
     }
 
+    const mediaUrlAllowlistEntry = await prompter.text({
+      message: "Media URL allowlist (comma-separated host/IP/CIDR, optional)",
+      placeholder: "cdn.example.com, 192.168.1.23, 10.0.0.0/8",
+      initialValue: (resolved.mediaUrlAllowlist || []).join(", ") || undefined,
+    });
+    const mediaUrlAllowlistParsed = parseList(String(mediaUrlAllowlistEntry ?? ""));
+    const mediaUrlAllowlist = mediaUrlAllowlistParsed.length > 0 ? mediaUrlAllowlistParsed : undefined;
+
     const groupPolicyValue = await prompter.select({
       message: "Group message policy",
       options: [
@@ -336,6 +350,71 @@ export const dingtalkOnboardingAdapter: ChannelOnboardingAdapter = {
       maxReconnectCycles = Number.isInteger(parsedCycles) && parsedCycles > 0 ? parsedCycles : 10;
     }
 
+    let mediaMaxMb: number | undefined;
+    const wantsMediaMax = await prompter.confirm({
+      message: "Configure inbound media max size in MB? (optional)",
+      initialValue: typeof resolved.mediaMaxMb === "number",
+    });
+    if (wantsMediaMax) {
+      const parsedMediaMax = Number(
+        String(
+          await prompter.text({
+            message: "Max inbound media size (MB)",
+            placeholder: "20",
+            initialValue:
+              typeof resolved.mediaMaxMb === "number" ? String(resolved.mediaMaxMb) : "20",
+            validate: (value) => {
+              const raw = String(value ?? "").trim();
+              const num = Number(raw);
+              if (!raw) {
+                return "Required";
+              }
+              if (!Number.isInteger(num) || num < 1) {
+                return "Must be an integer >= 1";
+              }
+              return undefined;
+            },
+          }),
+        ).trim(),
+      );
+      mediaMaxMb = Number.isInteger(parsedMediaMax) && parsedMediaMax > 0 ? parsedMediaMax : 20;
+    }
+
+    let journalTTLDays: number | undefined;
+    const wantsJournalTTL = await prompter.confirm({
+      message: "Configure quote journal retention in days?",
+      initialValue: typeof resolved.journalTTLDays === "number",
+    });
+    if (wantsJournalTTL) {
+      const parsedJournalTTL = Number(
+        String(
+          await prompter.text({
+            message: "Quote journal retention days",
+            placeholder: String(DEFAULT_JOURNAL_TTL_DAYS),
+            initialValue:
+              typeof resolved.journalTTLDays === "number"
+                ? String(resolved.journalTTLDays)
+                : String(DEFAULT_JOURNAL_TTL_DAYS),
+            validate: (value) => {
+              const raw = String(value ?? "").trim();
+              const num = Number(raw);
+              if (!raw) {
+                return "Required";
+              }
+              if (!Number.isInteger(num) || num < 1) {
+                return "Must be an integer >= 1";
+              }
+              return undefined;
+            },
+          }),
+        ).trim(),
+      );
+      journalTTLDays =
+        Number.isInteger(parsedJournalTTL) && parsedJournalTTL > 0
+          ? parsedJournalTTL
+          : DEFAULT_JOURNAL_TTL_DAYS;
+    }
+
     const next = applyAccountConfig({
       cfg,
       accountId,
@@ -348,10 +427,13 @@ export const dingtalkOnboardingAdapter: ChannelOnboardingAdapter = {
         dmPolicy: dmPolicyValue as "open" | "allowlist",
         groupPolicy: groupPolicyValue as "open" | "allowlist",
         allowFrom,
+        mediaUrlAllowlist,
         messageType,
         cardTemplateId,
         cardTemplateKey,
         maxReconnectCycles,
+        mediaMaxMb,
+        journalTTLDays,
       },
     });
 

package/src/peer-id-registry.ts

@@ -8,7 +8,13 @@
  * can be delivered correctly.
  */
 
+import { readFileSync, readdirSync } from "fs";
+import * as os from "os";
+import { join } from "path";
+import { getLogger } from "./logger-context";
+
 const peerIdMap = new Map<string, string>();
+let preloaded = false;
 
 /**
  * Register an original peer ID, keyed by its lowercased form.
@@ -20,14 +26,31 @@ export function registerPeerId(originalId: string): void {
   peerIdMap.set(originalId.toLowerCase(), originalId);
 }
 
+function maybeRegisterDingTalkGroupPeerId(value: unknown): void {
+  // DingTalk group openConversationId values are base64-like and consistently start with "cid".
+  // User IDs in DM contexts do not follow this pattern and do not require case restoration.
+  if (typeof value === "string" && value.startsWith("cid")) {
+    registerPeerId(value);
+  }
+}
+
 /**
  * Resolve a possibly-lowercased peer ID back to its original casing.
+ *
+ * If registry is empty at first outbound call (for example, cron/delivery queue
+ * fires before inbound callbacks), perform a one-time lazy preload from sessions.
+ *
  * Returns the original if found, otherwise returns the input as-is.
  */
 export function resolveOriginalPeerId(id: string): string {
   if (!id) {
     return id;
   }
+
+  if (!preloaded) {
+    preloadPeerIdsFromSessions();
+  }
+
   return peerIdMap.get(id.toLowerCase()) || id;
 }
 
@@ -36,4 +59,83 @@ export function resolveOriginalPeerId(id: string): string {
  */
 export function clearPeerIdRegistry(): void {
   peerIdMap.clear();
+  preloaded = false;
+}
+
+/**
+ * Preload known peer IDs from all agents' sessions.json files.
+ *
+ * Safe to call repeatedly:
+ * - without explicit homeDir: runs once, then no-ops;
+ * - with explicit homeDir: always runs (useful for tests).
+ */
+export function preloadPeerIdsFromSessions(homeDir?: string): void {
+  if (!homeDir && preloaded) {
+    return;
+  }
+
+  const home = homeDir || os.homedir();
+  const agentsDir = join(home, ".openclaw", "agents");
+  const log = getLogger();
+  let preloadCompleted = false;
+
+  try {
+    const agentDirs = readdirSync(agentsDir, { withFileTypes: true })
+      .filter((entry) => entry.isDirectory())
+      .map((entry) => entry.name);
+
+    for (const agentName of agentDirs) {
+      const sessionsPath = join(agentsDir, agentName, "sessions", "sessions.json");
+
+      try {
+        const raw = readFileSync(sessionsPath, "utf-8");
+        const parsed = JSON.parse(raw);
+        if (!parsed || typeof parsed !== "object") {
+          continue;
+        }
+
+        for (const session of Object.values(parsed as Record<string, unknown>)) {
+          if (!session || typeof session !== "object") {
+            continue;
+          }
+
+          const sessionRecord = session as Record<string, unknown>;
+          maybeRegisterDingTalkGroupPeerId(sessionRecord.lastTo);
+
+          const origin = sessionRecord.origin;
+          if (!origin || typeof origin !== "object") {
+            continue;
+          }
+
+          const originRecord = origin as Record<string, unknown>;
+          maybeRegisterDingTalkGroupPeerId(originRecord.from);
+          maybeRegisterDingTalkGroupPeerId(originRecord.to);
+        }
+      } catch (err) {
+        // sessions.json may be missing or malformed for some agents; ignore per file.
+        log?.debug?.(
+          `[DingTalk][PeerIdRegistry] Failed to parse preload sessions file ${sessionsPath}: ${
+            err instanceof Error ? err.message : String(err)
+          }`,
+        );
+      }
+    }
+    preloadCompleted = true;
+  } catch (err) {
+    const errorCode = (err as NodeJS.ErrnoException | undefined)?.code;
+    if (errorCode === "ENOENT") {
+      // agents directory may not exist yet; this is a stable no-op state.
+      preloadCompleted = true;
+    } else {
+      log?.debug?.(
+        `[DingTalk][PeerIdRegistry] Failed to scan preload directory ${agentsDir}: ${
+          err instanceof Error ? err.message : String(err)
+        }`,
+      );
+    }
+  }
+
+  if (!homeDir && preloadCompleted) {
+    preloaded = true;
+  }
 }