@wenbin_wb/dsh-bridge 2.8.7 → 2.9.0

package/lib/platform/dsh-storage.js

@@ -0,0 +1,117 @@
+// DSH 私有存储读取（workspace.json / session_projcache.json 兜底）
+// 自 conversation-bridge.js 拆出。策略：内存服务（workspaceRegistry）优先，
+// 仅当内存服务缺失时才落盘读 DSH 存储；文件不存在时安全返回空值。
+// 注意：不再以 ctx._mock 作为跳过依据——测试通过提供内存服务或注入 DSH_HOME 保持隔离。
+import { existsSync, readFileSync } from 'node:fs'
+import { join, basename } from 'node:path'
+import { homedir } from 'node:os'
+
+export function getArchivedSessionIds(ctx) {
+  const archived = new Set()
+  // 1. 尝试从 ctx.workspaceRegistry 内存服务读取
+  try {
+    const list = ctx?.workspaceRegistry?.archivedSessionIds
+    if (Array.isArray(list)) {
+      for (const id of list) {
+        if (id) archived.add(String(id))
+      }
+      return archived
+    }
+  } catch { /* ignore */ }
+
+  // 2. 尝试从 DSH workspace 存储文件（$DSH_HOME/storages/workspace.json）读取兜底
+  {
+    try {
+      const home = process.env.DSH_HOME || join(homedir(), '.dsh')
+      const wsFile = join(home, 'storages', 'workspace.json')
+      if (existsSync(wsFile)) {
+        const data = JSON.parse(readFileSync(wsFile, 'utf8'))
+        const fileArchived = data?.global?.archivedSessionIds
+        if (Array.isArray(fileArchived)) {
+          for (const id of fileArchived) {
+            if (id) archived.add(String(id))
+          }
+        }
+      }
+    } catch { /* ignore */ }
+  }
+
+  return archived
+}
+
+/**
+ * 安全探测 ctx 上的非 inject 属性。
+ * 宿主 cordis 上下文对插件未在 inject 中声明的属性读取会直接抛错
+ * （'cannot get property "x" without inject'），因此内存注入点（测试夹具用）
+ * 必须经 try/catch 探测，绝不能让异常外溢到命令路径。
+ */
+function peekCtxProperty(ctx, key) {
+  if (!ctx) return undefined
+  try { return ctx[key] } catch { return undefined }
+}
+
+/** 读取 DSH 官方持久化会话缓存元数据（标题、是否空白、创建时间等） */
+export function getSessionProjCache(ctx) {
+  const injected = peekCtxProperty(ctx, 'sessionProjCache')
+  if (injected) return injected
+  try {
+    const home = process.env.DSH_HOME || join(homedir(), '.dsh')
+    const cacheFile = join(home, 'storages', 'session_projcache.json')
+    if (existsSync(cacheFile)) {
+      const data = JSON.parse(readFileSync(cacheFile, 'utf8'))
+      return data?.tables?.sessions || {}
+    }
+  } catch { /* ignore */ }
+  return {}
+}
+
+/** 读取 DSH 官方注册的工作区列表及各自绑定的 sessionIds 列表 */
+export async function getRegisteredWorkspaces(ctx) {
+  const workspaces = []
+
+  // 优先从内存服务获取
+  if (ctx?.workspaceRegistry) {
+    try {
+      const list = await ctx.workspaceRegistry.list?.()
+      if (Array.isArray(list)) {
+        for (const w of list) {
+          if (w && w.path) {
+            workspaces.push({
+              id: w.id || w.path,
+              path: w.path,
+              title: w.title || basename(w.path),
+              sessionIds: Array.isArray(w.sessionIds) ? [...w.sessionIds] : [],
+            })
+          }
+        }
+        return workspaces
+      }
+    } catch { /* ignore */ }
+  }
+
+  // 兜底从 workspace.json 存储文件读取
+  {
+    try {
+      const home = process.env.DSH_HOME || join(homedir(), '.dsh')
+      const wsFile = join(home, 'storages', 'workspace.json')
+      if (existsSync(wsFile)) {
+        const data = JSON.parse(readFileSync(wsFile, 'utf8'))
+        const wsIds = data?.global?.workspaceIds || Object.keys(data?.tables?.workspaces || {})
+        const table = data?.tables?.workspaces || {}
+        for (const wId of wsIds) {
+          const ws = table[wId]
+          if (ws && ws.path) {
+            workspaces.push({
+              id: wId,
+              path: ws.path,
+              title: ws.title || basename(ws.path),
+              sessionIds: Array.isArray(ws.sessionIds) ? [...ws.sessionIds] : [],
+            })
+          }
+        }
+      }
+    } catch { /* ignore */ }
+  }
+
+  return workspaces
+}

package/lib/platform/index.js

@@ -1,10 +1,10 @@
-// dsh-bridge 平台抽象层统一导出
-//
-// 提供多平台 IM 接入的基础设施：
-//   Platform             平台适配器基类（协议/连接/登录/收发消息）
-//   ConversationBridge   平台无关会话桥（白名单/会话/审批/命令/digest）
-//   PlatformManager      多平台注册与状态聚合
-
-export { Platform } from './base.js'
-export { ConversationBridge, conversationBridgeHelpers, BRIDGE_MARK, textOfAssistantMessage } from './conversation-bridge.js'
-export { PlatformManager } from './manager.js'
+// dsh-bridge 平台抽象层统一导出
+//
+// 提供多平台 IM 接入的基础设施：
+//   Platform             平台适配器基类（协议/连接/登录/收发消息）
+//   ConversationBridge   平台无关会话桥（白名单/会话/审批/命令/digest）
+//   PlatformManager      多平台注册与状态聚合
+
+export { Platform } from './base.js'
+export { ConversationBridge, conversationBridgeHelpers, textOfAssistantMessage } from './conversation-bridge.js'
+export { PlatformManager } from './manager.js'

package/lib/platform/message-split.js

@@ -0,0 +1,191 @@
+// 出站消息分块与 SEND_FILE 指令解析（平台无关纯函数）
+// 自 conversation-bridge.js 拆出：按平台 maxMessageChars 分块、保留 fenced code block、
+// [SEND_FILE: ...] 显式指令提取与路径解析。
+import { statSync } from 'node:fs'
+import { isAbsolute, normalize, resolve } from 'node:path'
+
+const FENCE_RE = /^```([^\n`]*)\s*$/
+
+function normalizeMarkdownBlocks(content) {
+  const lines = content.split('\n')
+  const out = []
+  let blankRun = 0
+  let inCode = false
+  for (const raw of lines) {
+    const line = raw.replace(/\s+$/, '')
+    if (FENCE_RE.test(line.trim())) {
+      inCode = !inCode
+      out.push(line)
+      blankRun = 0
+      continue
+    }
+    if (inCode) {
+      out.push(line)
+      continue
+    }
+    if (!line.trim()) {
+      blankRun += 1
+      if (blankRun <= 1) out.push('')
+      continue
+    }
+    blankRun = 0
+    out.push(line)
+  }
+  return out.join('\n').trim()
+}
+
+function splitMarkdownBlocks(content) {
+  const blocks = []
+  let current = []
+  let inCode = false
+  const flush = () => {
+    const block = current.join('\n').trim()
+    if (block) blocks.push(block)
+    current = []
+  }
+  for (const raw of content.split('\n')) {
+    const line = raw.replace(/\s+$/, '')
+    if (FENCE_RE.test(line.trim())) {
+      if (!inCode && current.length) flush()
+      current.push(line)
+      inCode = !inCode
+      if (!inCode) flush()
+      continue
+    }
+    if (inCode) {
+      current.push(line)
+      continue
+    }
+    if (!line.trim()) {
+      flush()
+      continue
+    }
+    current.push(line)
+  }
+  flush()
+  return blocks
+}
+
+function hardSplit(text, max) {
+  const chunks = []
+  let rest = text
+  while (rest.length > max) {
+    chunks.push(rest.slice(0, max))
+    rest = rest.slice(max)
+  }
+  if (rest) chunks.push(rest)
+  return chunks
+}
+
+function packBlocks(blocks, max) {
+  const units = []
+  let current = ''
+  for (const block of blocks) {
+    const candidate = current ? `${current}\n\n${block}` : block
+    if (candidate.length <= max) {
+      current = candidate
+      continue
+    }
+    if (current) units.push(current)
+    if (block.length <= max) {
+      current = block
+    } else {
+      units.push(...hardSplit(block, max))
+      current = ''
+    }
+  }
+  if (current) units.push(current)
+  return units
+}
+
+export function splitForIM(content, max = 2000) {
+  // 安全检查：防止畸形输入导致无限循环或崩溃
+  if (typeof content !== 'string' || content.length === 0) return []
+  if (content.length > 1_000_000) {
+    content = content.slice(0, 1_000_000) + '\n\n[已截断：内容过长]'
+  }
+  const normalized = normalizeMarkdownBlocks(content)
+  if (!normalized) return []
+  if (normalized.length <= max) return [normalized]
+  return packBlocks(splitMarkdownBlocks(normalized), max)
+}
+
+// 协议标记过滤：模型偶尔会把工具调用语法（如 <｜｜DSML｜｜tool_calls>…）当正文
+// 输出，这些是协议内容，转发到 IM 只会是乱码噪音。整块移除 tool_calls 段落，
+// 再剥掉残余的 DSML 标签；剩余正文照常发送。
+export function stripProtocolMarkup(text) {
+  if (!text || typeof text !== 'string' || !text.includes('DSML')) return text
+  let out = text.replace(/<[^<>]*DSML[^<>]*tool_calls[^<>]*>[\s\S]*?<[^<>]*DSML[^<>]*tool_calls[^<>]*>/g, '')
+  out = out.replace(/<[^<>]*DSML[^<>]*>/g, '')
+  return out
+}
+
+export function textOfAssistantMessage(message) {
+  const raw = (message.content ?? [])
+    .filter((block) => block?.type === 'text')
+    .map((block) => block.text)
+    .join('\n')
+  return stripProtocolMarkup(raw)
+}
+
+/**
+ * 尝试将任意路径（绝对或相对当前工作区）解析为真实存在的本地文件绝对路径
+ */
+export function resolveFilePath(rawPath, cwd = process.cwd()) {
+  if (typeof rawPath !== 'string') return null
+  let p = rawPath.trim()
+    .replace(/^["'`]|["'`]$/g, '')
+    .replace(/^file:\/\/\/?/, '')
+    .replace(/^[📁📄📦\s]+/, '')
+  if (!p) return null
+  // 排除 HTTP/HTTPS 网址
+  if (/^https?:\/\//i.test(p)) return null
+  const resolved = isAbsolute(p) ? normalize(p) : resolve(cwd, p)
+  try {
+    if (statSync(resolved).isFile()) {
+      return resolved
+    }
+  } catch {}
+  return null
+}
+
+/**
+ * 提取并过滤文本中的 [SEND_FILE: <path>] 显式发送指令
+ * 由 AI 根据用户意图显式决定何时向用户发送文件附件，杜绝底层盲目扫描与误发。
+ * @param {string} text - 原始助手回复文本
+ * @param {string} cwd - 会话当前工作目录
+ * @returns {{ cleanText: string, files: string[] }}
+ */
+export function extractAndStripSendFileDirectives(text, cwd = process.cwd()) {
+  if (typeof text !== 'string' || !text.trim()) {
+    return { cleanText: text || '', files: [] }
+  }
+
+  const files = []
+  const directiveRegex = /\[(?:SEND_FILE|SEND-FILE|send_file|send-file|SEND_MEDIA|send_media):\s*[`"']?([^\]`"'\r\n]+?)[`"']?\s*\]/gi
+
+  let m
+  const re = new RegExp(directiveRegex)
+  while ((m = re.exec(text)) !== null) {
+    const rawPath = m[1].trim()
+    const resolved = resolveFilePath(rawPath, cwd)
+    if (resolved && !files.includes(resolved)) {
+      files.push(resolved)
+    }
+  }
+
+  // 从聊天正文中彻底剔除控制指令（保持 IM 聊天气泡的干净整洁）
+  const cleanText = text.replace(directiveRegex, '').replace(/\n{3,}/g, '\n\n').trim()
+
+  return { cleanText, files }
+}
+
+/**
+ * 提取文本中的产物文件路径（基于显式指令）
+ */
+export function extractFilePathsFromText(text, cwd = process.cwd()) {
+  return extractAndStripSendFileDirectives(text, cwd).files
+}
+
+// ---------------------------------------------------------------------------
+// digest 摘要