fix(claw): restore spreadsheet attachment badges

2026-08-11 21:09:38 +08:00 · 2026-08-11 21:09:38 +08:00 · 69705518d6
commit 69705518d6
parent f6f4f94245
3 changed files with 139 additions and 37 deletions
--- a/components/canvas/CanvasChatPanel.tsx
+++ b/components/canvas/CanvasChatPanel.tsx
@ -504,6 +504,21 @@ const CLAWD_IMAGE_ALLOWED_EXTENSIONS = new Set([
 ]);
 const CLAWD_TABLE_MAX_FILES = 3;
 const CLAWD_TABLE_MAX_BYTES = 10 * 1024 * 1024;
+
+function isClawdSpreadsheetAttachment(
+  url: string,
+  contentType?: string | null,
+  name?: string | null
+) {
+  const normalizedType = String(contentType || "").toLowerCase();
+  return (
+    normalizedType.includes("spreadsheet") ||
+    normalizedType === "text/csv" ||
+    normalizedType === "text/tab-separated-values" ||
+    /\.(xlsx|csv|tsv)(?:$|[?#])/i.test(String(url || "")) ||
+    /\.(xlsx|csv|tsv)$/i.test(String(name || "").trim())
+  );
+}
 const CLAWD_PROCESSING_HINT =
  "已收到，这个任务稍微繁重一些，处理时间可能会长一点，请耐心等候。";
 const CLAWD_SUBMITTED_COMPOSER_RECOVERY_TTL_MS = 30 * 60 * 1000;
@ -923,38 +938,43 @@ function parseClawdSpreadsheetAttachmentsFromText(
    .filter((part: any) => part?.type === "text" && part?.text)
    .map((part: any) => String(part.text))
    .join("\n");
-  const summaryMatch = text.match(
-    /(?:^|\n)已上传表格：\n((?:- [^\n]+(?:\n|$))+)/m
-  );
-  if (!summaryMatch?.[1]) {
-    return [];
-  }
-
  const attachments: ClawdSpreadsheetAttachment[] = [];
-  for (const line of summaryMatch[1].trim().split("\n")) {
-    const lineMatch = line.match(/^- (.+?)(?:（(.+)）)?$/);
-    const name = String(lineMatch?.[1] || "").trim();
-    const details = String(lineMatch?.[2] || "");
-    if (!name) continue;
+  const sectionPattern =
+    /(?:^|\n)(已上传表格|已上传文件)：\n((?:- [^\n]+(?:\n|$))+)/gm;
+  for (const sectionMatch of text.matchAll(sectionPattern)) {
+    const sectionType = String(sectionMatch[1] || "");
+    const sectionLines = String(sectionMatch[2] || "");
+    for (const line of sectionLines.trim().split("\n")) {
+      const lineMatch = line.match(/^- (.+?)(?:（(.+)）)?$/);
+      const name = String(lineMatch?.[1] || "").trim();
+      const details = String(lineMatch?.[2] || "");
+      if (
+        !name ||
+        (sectionType === "已上传文件" &&
+          !isClawdSpreadsheetAttachment("", "", name))
+      ) {
+        continue;
+      }

-    const sizeMatch = details.match(/([\d.]+)\s*(KB|MB)/i);
-    const sheetCountMatch = details.match(/(\d+)\s*个工作表/);
-    const rowCountMatch = details.match(/(\d+)\s*行/);
-    const sizeValue = Number(sizeMatch?.[1]);
-    const sizeUnit = String(sizeMatch?.[2] || "").toUpperCase();
+      const sizeMatch = details.match(/([\d.]+)\s*(KB|MB)/i);
+      const sheetCountMatch = details.match(/(\d+)\s*个工作表/);
+      const rowCountMatch = details.match(/(\d+)\s*行/);
+      const sizeValue = Number(sizeMatch?.[1]);
+      const sizeUnit = String(sizeMatch?.[2] || "").toUpperCase();

-    attachments.push({
-      name,
-      size:
-        Number.isFinite(sizeValue) && sizeValue > 0
-          ? Math.round(
-              sizeValue * (sizeUnit === "MB" ? 1024 * 1024 : 1024)
-            )
-          : undefined,
-      sheetCount: sheetCountMatch ? Number(sheetCountMatch[1]) : undefined,
-      rowCount: rowCountMatch ? Number(rowCountMatch[1]) : undefined,
-      truncated: details.includes("已截断"),
-    });
+      attachments.push({
+        name,
+        size:
+          Number.isFinite(sizeValue) && sizeValue > 0
+            ? Math.round(
+                sizeValue * (sizeUnit === "MB" ? 1024 * 1024 : 1024)
+              )
+            : undefined,
+        sheetCount: sheetCountMatch ? Number(sheetCountMatch[1]) : undefined,
+        rowCount: rowCountMatch ? Number(rowCountMatch[1]) : undefined,
+        truncated: details.includes("已截断"),
+      });
+    }
  }
  return attachments;
 }
@ -977,11 +997,7 @@ function getClawdSpreadsheetAttachments(
    : Array.isArray(videoMetadata.spreadsheetAttachments)
      ? videoMetadata.spreadsheetAttachments
      : [];
-  if (rawAttachments.length === 0) {
-    return parseClawdSpreadsheetAttachmentsFromText(message);
-  }
-
-  return rawAttachments
+  const normalizedAttachments = rawAttachments
    .map((attachment: any) => {
      const name = String(
        attachment?.name || attachment?.fileName || ""
@ -1015,14 +1031,72 @@ function getClawdSpreadsheetAttachments(
        attachment: ClawdSpreadsheetAttachment | null
      ): attachment is ClawdSpreadsheetAttachment => Boolean(attachment)
    );
+
+  const knownNames = new Set(
+    normalizedAttachments.map((attachment: ClawdSpreadsheetAttachment) =>
+      attachment.name.toLowerCase()
+    )
+  );
+  const inferredAttachments = [
+    ...(((message as any).attachments || []) as any[]),
+    ...((message.parts || []).filter((part: any) => part?.type === "file") as any[]),
+  ];
+  for (const attachment of inferredAttachments) {
+    const name = String(attachment?.name || attachment?.fileName || "").trim();
+    const url = String(attachment?.url || attachment?.data || "").trim();
+    const contentType = String(
+      attachment?.contentType ||
+        attachment?.content_type ||
+        attachment?.mediaType ||
+        ""
+    ).trim();
+    if (
+      !name ||
+      knownNames.has(name.toLowerCase()) ||
+      !isClawdSpreadsheetAttachment(url, contentType, name)
+    ) {
+      continue;
+    }
+    normalizedAttachments.push({ name, contentType: contentType || undefined });
+    knownNames.add(name.toLowerCase());
+  }
+
+  for (const attachment of parseClawdSpreadsheetAttachmentsFromText(message)) {
+    if (knownNames.has(attachment.name.toLowerCase())) continue;
+    normalizedAttachments.push(attachment);
+    knownNames.add(attachment.name.toLowerCase());
+  }
+
+  return normalizedAttachments;
 }

-function stripClawdSpreadsheetSummary(text: string) {
+function stripClawdSpreadsheetSummary(
+  text: string,
+  attachments: ClawdSpreadsheetAttachment[]
+) {
+  const spreadsheetNames = new Set(
+    attachments.map((attachment) => attachment.name.toLowerCase())
+  );
  return text
    .replace(
      /(?:^|\n{2})已上传表格：\n(?:- [^\n]*(?:\n|$))+/g,
      "\n"
    )
+    .replace(
+      /(?:^|\n{2})已上传文件：\n((?:- [^\n]*(?:\n|$))+)/g,
+      (_section, lines: string) => {
+        const remainingLines = String(lines)
+          .trim()
+          .split("\n")
+          .filter((line) => {
+            const name = line.replace(/^- /, "").trim().toLowerCase();
+            return !spreadsheetNames.has(name);
+          });
+        return remainingLines.length > 0
+          ? `\n\n已上传文件：\n${remainingLines.join("\n")}`
+          : "\n";
+      }
+    )
    .replace(/\n{3,}/g, "\n\n")
    .trim();
 }
@ -12233,6 +12307,7 @@ export function CanvasChatPanel({
        id: userMessageId,
        role: "user",
        parts: [{ type: "text", text: optimisticMessageText }],
+        attachments: sanitizedAttachmentsSnapshot,
        createdAt: optimisticCreatedAt,
        metadata: {
          createdAt: optimisticCreatedAt.toISOString(),
@ -15913,7 +15988,10 @@ export function CanvasChatPanel({
                                    .trim();
                                  const text =
                                    spreadsheetAttachments.length > 0
-                                      ? stripClawdSpreadsheetSummary(normalizedText)
+                                      ? stripClawdSpreadsheetSummary(
+                                          normalizedText,
+                                          spreadsheetAttachments
+                                        )
                                      : normalizedText;
                                  if (!text) return;
                                  renderedUserParts.push(
--- a/components/canvas/chat-panel/canvas-chat-session.ts
+++ b/components/canvas/chat-panel/canvas-chat-session.ts
@ -1,4 +1,5 @@
 import type { Attachment, ChatMessage } from "@/lib/types";
+import { isOpenClawChatFileAttachment } from "@/lib/openclaw-chat/media-attachments";

 export const SESSION_TITLE_MAX_LEN = 18;
 export const MAX_CHAT_TEXT_PART_LENGTH = 2000;
@ -80,6 +81,9 @@ export function normalizeChatMediaType(value?: string | null): string {
  if (ALLOWED_CHAT_MEDIA_TYPES.has(normalized)) {
    return normalized;
  }
+  if (isOpenClawChatFileAttachment("", normalized)) {
+    return normalized;
+  }
  return "image/jpeg";
 }

--- a/tests/routes/canvas-chat-helpers.test.ts
+++ b/tests/routes/canvas-chat-helpers.test.ts
@ -235,7 +235,11 @@ import {
  normalizeUploadedAssetId,
  validateCanvasImageUploadFile,
 } from "@/components/canvas/chat-panel/canvas-upload-helpers";
-import type { PickedObjectAttachment } from "@/components/canvas/chat-panel/canvas-chat-session";
+import {
+  normalizeChatMediaType,
+  sanitizeChatAttachment,
+  type PickedObjectAttachment,
+} from "@/components/canvas/chat-panel/canvas-chat-session";
 import type { CanvasAgentContext } from "@/types/canvas/agent-protocol";
 import type { ChatMessage } from "@/lib/types";
 import type { DBMessage } from "@/lib/db/schema";
@ -253,6 +257,22 @@ function dbTextMessage(
 }

 test.describe("canvas chat pure helpers", () => {
+  test("preserves spreadsheet attachment content types", () => {
+    const spreadsheetContentType =
+      "application/vnd.openxmlformats-officedocument.spreadsheetml.sheet";
+
+    expect(normalizeChatMediaType(spreadsheetContentType)).toBe(
+      spreadsheetContentType
+    );
+    expect(
+      sanitizeChatAttachment({
+        name: "测试用.xlsx",
+        url: "https://cdn.example.com/uploads/test.xlsx",
+        contentType: spreadsheetContentType,
+      })?.contentType
+    ).toBe(spreadsheetContentType);
+  });
+
  test("keeps optimistic messages without timestamps after persisted history", () => {
    const messages = [
      {