@sleep2agi/dsh-anet 0.1.0-preview.30 → 0.1.0-preview.32

package/README.md

@@ -53,6 +53,7 @@ The user token is never printed or persisted; `--no-verify` skips the round-trip
 - If the plugin process dies mid-turn, the task is answered `failed` with "please resend" on the next start rather than being re-run silently.
 - Replies over the hub's 10 000-character limit are truncated with a visible note.
 - Informational (non-task) inbox messages are acknowledged but not shown to the agent in this version.
+- Image attachments on tasks are downloaded with the node token (up to 5 images, 10 MB each): vision models receive image blocks, text-only models receive macOS Vision OCR text instead. Other attachment types are not processed yet.
 
 ## Tests
 

package/README.zh.md

@@ -51,6 +51,7 @@ ANET_USER_TOKEN=utok_... node scripts/connect.mjs \
 - 插件进程在一轮中途死掉，下次启动时对该任务回 `failed` 并提示「请重发」，而不是悄悄重跑。
 - 超过 hub 1 万字上限的回件会截断并附说明。
 - 本版对「消息」类（非任务）收件只确认，不交给 agent。
+- 任务里的图片附件会用节点 token 下载（最多 5 张、单张 ≤10MB）：视觉模型收到 image block，文本模型（如 DeepSeek chat-completions）收到 macOS Vision OCR 文字。其他类型附件暂不处理。
 
 ## 测试
 

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@sleep2agi/dsh-anet",
-  "version": "0.1.0-preview.30",
+  "version": "0.1.0-preview.32",
   "description": "DSH (DeepSeek Harness) plugin: expose one or more DSH sessions as Agent Network (anet.sh / CommHub) nodes.",
   "type": "module",
   "main": "./src/plugin.mjs",

package/src/ocr.swift

@@ -0,0 +1,17 @@
+import Foundation
+import Vision
+import AppKit
+
+let path = CommandLine.arguments[1]
+guard let image = NSImage(contentsOfFile: path),
+      let cg = image.cgImage(forProposedRect: nil, context: nil, hints: nil) else {
+    exit(2)
+}
+let request = VNRecognizeTextRequest()
+request.recognitionLevel = .accurate
+request.usesLanguageCorrection = true
+request.recognitionLanguages = ["zh-Hans", "en-US"]
+let handler = VNImageRequestHandler(cgImage: cg, options: [:])
+try handler.perform([request])
+let lines = (request.results ?? []).compactMap { $0.topCandidates(1).first?.string }
+print(lines.joined(separator: "\n"))

package/src/plugin.mjs

@@ -2,8 +2,11 @@
 // Each node has its own name, hub, token, project directory and dedicated
 // session; tasks for a node run in that node's session so context is preserved.
 import { randomUUID } from "node:crypto";
-import { chmodSync, mkdirSync, readFileSync, unlinkSync, writeFileSync } from "node:fs";
-import { homedir } from "node:os";
+import { execFile } from "node:child_process";
+import { promisify } from "node:util";
+import { fileURLToPath } from "node:url";
+import { accessSync, chmodSync, mkdirSync, readFileSync, unlinkSync, writeFileSync } from "node:fs";
+import { homedir, tmpdir } from "node:os";
 import { basename, dirname, join } from "node:path";
 import { installModelSelection } from "@deepseek-ai/dsh-agent";
 import { createUserMessage } from "@deepseek-ai/dsh-llm";
@@ -17,11 +20,12 @@ import { findNodeBySession, loadNodes, newSessionId, normalizeNode, registryPath
 import { startConfigServer } from "./ui-server.mjs";
 
 export { summarizeTurn } from "./turn.mjs";
+export { attachmentCandidates, collectAttachments };
 
 export const name = "anet";
-export const inject = ["agents", "agentDefaultModel", "sessions", "tools"];
+export const inject = ["agents", "agentDefaultModel", "sessions", "tools", "attachments"];
 
-const VERSION = "dsh-anet@0.1.0-preview.30";
+const VERSION = "dsh-anet@0.1.0-preview.32";
 
 function agentSetup(ctx) {
   const selection = ctx.agentDefaultModel.currentSelection();
@@ -48,10 +52,88 @@ async function ensureAgent(ctx, node) {
   return handle.agent;
 }
 
-async function runNodeTurn(ctx, node, prompt) {
+const IMAGE_MAX_BYTES = 10 * 1024 * 1024;
+const MAX_IMAGES = 5;
+const MIME_BY_EXT = { jpg: "image/jpeg", jpeg: "image/jpeg", png: "image/png", gif: "image/gif", webp: "image/webp" };
+
+const runFile = promisify(execFile);
+const ocrSource = fileURLToPath(new URL("./ocr.swift", import.meta.url));
+
+async function ocrImage(path) {
+  const bin = join(homedir(), ".anet", "dsh", "anet-ocr");
+  try { accessSync(bin); } catch {
+    mkdirSync(dirname(bin), { recursive: true, mode: 0o700 });
+    await runFile("swiftc", ["-O", ocrSource, "-o", bin], { timeout: 120_000 });
+    chmodSync(bin, 0o700);
+  }
+  const { stdout } = await runFile(bin, [path], { timeout: 60_000, maxBuffer: 2 * 1024 * 1024 });
+  return String(stdout).trim();
+}
+
+function modelReadsImages(selection) {
+  const id = `${selection?.provider ?? ""} ${selection?.model ?? ""}`.toLowerCase();
+  if (id.includes("deepseek")) return /vl|vision|multimodal/.test(id);
+  return true;
+}
+
+function attachmentCandidates(task) {
+  const candidates = [];
+  for (const item of task?.meta?.attachments ?? task?.attachments ?? []) {
+    if (item && typeof item === "object") candidates.push({ id: item.file_id ?? item.id, url: item.url, name: item.name, mime: item.mime });
+  }
+  const text = String(task?.content ?? "");
+  const re = /附件\s+(\S+?)（([^）]+)）[\s\S]*?API:\s*GET\s+(\/api\/files\/[A-Za-z0-9_-]+)/gu;
+  for (const match of text.matchAll(re)) candidates.push({ name: match[1], mime: match[2], url: match[3] });
+  return candidates;
+}
+
+async function collectAttachments(ctx, node, task, options = {}) {
+  const blocks = []; const notes = []; let text = "";
+  const inline = options.inlineImages !== false;
+  const read = options.ocr ?? ocrImage;
+  let count = 0;
+  for (const candidate of attachmentCandidates(task)) {
+    if (count >= MAX_IMAGES) break;
+    const mime = String(candidate.mime ?? MIME_BY_EXT[String(candidate.name ?? "").split(".").pop()?.toLowerCase()] ?? "");
+    if (!mime.startsWith("image/")) continue;
+    const url = candidate.url
+      ? (String(candidate.url).startsWith("http") ? String(candidate.url) : `${node.hub}${String(candidate.url).startsWith("/") ? "" : "/"}${candidate.url}`)
+      : (candidate.id ? `${node.hub}/api/files/${encodeURIComponent(candidate.id)}` : "");
+    if (!url) continue;
+    try {
+      const res = await fetch(url, { headers: { Authorization: `Bearer ${node.token}` }, signal: AbortSignal.timeout(30_000) });
+      if (!res.ok) throw new Error(`HTTP ${res.status}`);
+      const data = new Uint8Array(await res.arrayBuffer());
+      if (data.byteLength === 0) throw new Error("empty file");
+      if (data.byteLength > IMAGE_MAX_BYTES) throw new Error("image larger than 10 MB");
+      const tmp = join(tmpdir(), `anet-attach-${randomUUID().slice(0, 8)}.${(candidate.name ?? "img").split(".").pop() || "png"}`);
+      writeFileSync(tmp, data, { mode: 0o600 });
+      let recognized = "";
+      try { recognized = await read(tmp); } catch { /* OCR unavailable */ }
+      try { unlinkSync(tmp); } catch { /* best effort */ }
+      if (inline) {
+        const attachment = await ctx.attachments.saveImage({ data, mediaType: mime, ...(candidate.name ? { name: candidate.name } : {}) });
+        blocks.push({ type: "image", attachment });
+      }
+      if (recognized) text += `${text ? "\n" : ""}[图片文字 ${candidate.name ?? "image"}]\n${recognized}`;
+      notes.push(recognized ? `[附件 ${candidate.name ?? "image"}：已提取文字]` : `[附件 ${candidate.name ?? "image"}${inline ? "" : "：模型不支持图片，未提取到文字"}]`);
+      count++;
+    } catch (error) {
+      notes.push(`[附件 ${candidate.name ?? candidate.id ?? url} 下载失败: ${error?.message ?? error}]`);
+    }
+  }
+  return { blocks, notes, text };
+}
+
+async function runNodeTurn(ctx, node, prompt, task) {
+  const selection = ctx.agentDefaultModel.currentSelection();
+  const attachments = await collectAttachments(ctx, node, task, { inlineImages: modelReadsImages(selection) });
   const agent = await ensureAgent(ctx, node);
   const firstSeq = agent.session.seq;
-  agent.followup(createUserMessage({ content: [{ type: "text", text: prompt }], source: { kind: "user" } }));
+  const content = [{ type: "text", text: prompt }, ...attachments.blocks];
+  if (attachments.text) content.push({ type: "text", text: attachments.text });
+  if (attachments.notes.length > 0) content.push({ type: "text", text: attachments.notes.join("\n") });
+  agent.followup(createUserMessage({ content, source: { kind: "user" } }));
   await agent.whenIdle();
   await ctx.sessions.flush(agent.session);
   const { text, reason, errors } = summarizeTurn(agent.session, firstSeq);
@@ -97,7 +179,7 @@ export function apply(ctx, config) {
     const loop = createCommhubNode({
       client, alias: node.name, ledger: createLedger(ledgerPath), log,
       version: VERSION,
-      runTurn: (prompt) => runNodeTurn(ctx, node, prompt),
+      runTurn: (prompt, task) => runNodeTurn(ctx, node, prompt, task),
     });
     runtime.loop = loop;
     runtimes.set(node.id, runtime);