191 lines
10 KiB
Python
191 lines
10 KiB
Python
|
#!/usr/bin/env python3
|
|||
|
|
# 档案 19 §C6（B4：安全扫描分级扩展）+ 档案 18 v3 收尾（B2：编排器自愈补齐 picker）
|
|||
|
|
# 用法: python3 patch-b4-b2.py /opt/dshs
|
|||
|
|
import io, os, sys
|
|||
|
|
root = sys.argv[1] if len(sys.argv) > 1 else '.'
|
|||
|
|
def rd(p): return io.open(os.path.join(root,p), encoding='utf-8').read()
|
|||
|
|
def wr(p,s): io.open(os.path.join(root,p),'w',encoding='utf-8',newline='').write(s)
|
|||
|
|
def sub(p, old, new, cnt=1):
|
|||
|
|
s = rd(p); assert old in s, f'anchor missing in {p}: {old[:80]}'
|
|||
|
|
wr(p, s.replace(old,new,cnt))
|
|||
|
|
|
|||
|
|
# ───────────────────────── B4：security-scan.ts 分级扩展 ─────────────────────────
|
|||
|
|
p = 'src/web/security-scan.ts'
|
|||
|
|
s = rd(p)
|
|||
|
|
s = s.replace("""/**
|
|||
|
|
* Shared upload security scan (安全检测) for skill and business-plugin uploads.
|
|||
|
|
* Heuristic scan over text files only — a hit means an *obviously* malicious or
|
|||
|
|
* privilege-escalating pattern, so the upload is rejected rather than admitted.
|
|||
|
|
* @module dshs/web/security-scan
|
|||
|
|
*/""",
|
|||
|
|
"""/**
|
|||
|
|
* Shared upload security scan (安全检测) for skill and business-plugin uploads.
|
|||
|
|
*
|
|||
|
|
* 分级（档案 19 §C6，2026-09-11）：
|
|||
|
|
* · **P0（阻断）** —— 明显恶意/提权模式：命中即拒绝上传（HTTP 400），与历史行为一致；
|
|||
|
|
* · **P1（告警）** —— 可疑但常见的模式（动态执行、外联、敏感 env、长 base64、隐藏文件/`.git`）：
|
|||
|
|
* **不阻断**，作为 findings 返回并写日志，供管理员在上传时人工复核。
|
|||
|
|
* 之前只有"命中即 400"，缺少分级与整改提示；现在 `scanDir` 返回 P1 findings（调用方可忽略）。
|
|||
|
|
* @module dshs/web/security-scan
|
|||
|
|
*/""", 1)
|
|||
|
|
|
|||
|
|
s = s.replace("""const SCAN_TEXT_EXT = new Set([
|
|||
|
|
'.js', '.mjs', '.cjs', '.ts', '.tsx', '.jsx', '.sh', '.bash', '.py', '.json', '.md', '.txt', '.yml', '.yaml', '.html', '.css',
|
|||
|
|
])""",
|
|||
|
|
"""const SCAN_TEXT_EXT = new Set([
|
|||
|
|
'.js', '.mjs', '.cjs', '.ts', '.tsx', '.jsx', '.sh', '.bash', '.zsh', '.py', '.json', '.md', '.txt',
|
|||
|
|
'.yml', '.yaml', '.html', '.css', '.toml', '.ini', '.conf', '.env', '.cfg',
|
|||
|
|
// 2026-09-11 扩展：非 JS 生态的常见脚本/配置（上传包里同样可能藏恶意指令）
|
|||
|
|
'.rs', '.go', '.java', '.php', '.rb', '.pl', '.lua', '.sql', '.ps1', '.psm1', '.bat', '.cmd',
|
|||
|
|
])
|
|||
|
|
|
|||
|
|
/** 扫描体积上限（旧值 2MB 会漏掉较大脚本；提到 8MB，仍跳过二进制/超大文件）。 */
|
|||
|
|
const SCAN_MAX_BYTES = 8 * 1024 * 1024""", 1)
|
|||
|
|
|
|||
|
|
# 原 7 条 → BLOCK（P0），并补充明显恶意的模式
|
|||
|
|
s = s.replace("const DANGEROUS_PATTERNS: Array<{ re: RegExp; why: string }> = [",
|
|||
|
|
"/** P0：明显恶意/提权 → 直接阻断上传。 */\nconst BLOCK_PATTERNS: Array<{ re: RegExp; why: string }> = [", 1)
|
|||
|
|
s = s.replace(""" { re: /\\.credentials\\.yaml/, why: '读取实例会话密钥' },
|
|||
|
|
]""",
|
|||
|
|
""" { re: /\\.credentials\\.yaml/, why: '读取实例会话密钥' },
|
|||
|
|
// 2026-09-11 新增（档案 17 §S/M 对齐）：仍是"明显恶意"档，故归 P0
|
|||
|
|
{ re: /\\bnc\\s+-[a-z]*e[a-z]*\\s+\\S+\\s+\\d+/, why: '反弹 shell（nc -e）' },
|
|||
|
|
{ re: /\\/dev\\/tcp\\/\\d{1,3}(\\.\\d{1,3}){3}\\/\\d+/, why: '反弹 shell（/dev/tcp）' },
|
|||
|
|
{ re: /(?:base64\\s+-d|b64decode|atob)\\s*[\\s\\S]{0,40}?\\|\\s*(?:sh|bash)\\b/, why: 'base64 解码后直接执行' },
|
|||
|
|
{ re: /\\bchmod\\s+(?:\\+s|4[0-7]{3}|6[0-7]{3})\\b[^\\n]{0,40}(?:\\/usr\\/bin|\\/bin)\\//, why: '尝试为系统二进制加 setuid/特权位' },
|
|||
|
|
{ re: /:\\s*\\(\\s*\\)\\s*\\{\\s*:\\s*\\|\\s*:/, why: 'fork 炸弹' },
|
|||
|
|
]
|
|||
|
|
|
|||
|
|
/**
|
|||
|
|
* P1：可疑但常见 → **不阻断**，作为 findings 上报（管理员人工复核）。
|
|||
|
|
* 说明书见档案 19 §C6 与档案 17 §S/M。
|
|||
|
|
*/
|
|||
|
|
const WARN_RULES: Array<{ id: string; re: RegExp; why: string }> = [
|
|||
|
|
{ id: 'dynamic-exec', re: /\\b(?:child_process|execSync|execFileSync|spawnSync|require\\('child_process'\\))/, why: '动态执行子进程（需确认目标命令可信）' },
|
|||
|
|
{ id: 'vm-eval', re: /\\b(?:new\\s+Function\\s*\\(|require\\('vm'\\)|from 'node:vm'|eval\\s*\\()/, why: '动态求值 vm/eval/new Function' },
|
|||
|
|
{ id: 'sensitive-env', re: /process\\.env\\s*[.\\[]\\s*['"]?(?:DEEPSEEK_API_KEY|DSH_[A-Z_]+|[A-Z_]*TOKEN|[A-Z_]*SECRET|[A-Z_]*KEY)/, why: '读取敏感环境变量' },
|
|||
|
|
{ id: 'long-base64', re: /[A-Za-z0-9+/]{600,}={0,2}/, why: '超长 base64 载荷（可能是混淆/内嵌二进制）' },
|
|||
|
|
{ id: 'network-egress', re: /https?:\\/\\/(?!localhost|127\\.0\\.0\\.1)[a-z0-9.-]+\\.[a-z]{2,}/i, why: '外部网络访问（需确认域名可信；配合出网护栏）' },
|
|||
|
|
{ id: 'hidden-or-git', re: /(?:\\.git\\/|(?:^|\\/)\\.[a-z][a-z0-9_-]*\\/)/i, why: '包含隐藏目录/.git（可能携带仓库元数据或绕过审查）' },
|
|||
|
|
]
|
|||
|
|
|
|||
|
|
/** 一条扫描发现（P1 告警）。 */
|
|||
|
|
export interface ScanFinding { rule: string; why: string; file: string }""", 1)
|
|||
|
|
|
|||
|
|
# scanDir：跳过二进制、用新上限、收集 P1 findings 并返回
|
|||
|
|
s = s.replace("""/** Recursively scan text files under `root` for dangerous patterns. */
|
|||
|
|
export function scanDir(root: string, rel = ''): void {""",
|
|||
|
|
"""/**
|
|||
|
|
* Recursively scan text files under `root`.
|
|||
|
|
* P0 命中 → 抛 400（阻断上传）；P1 命中 → 收集并**返回**（调用方可记录/展示，不影响上传）。
|
|||
|
|
*/
|
|||
|
|
export function scanDir(root: string, rel = '', findings: ScanFinding[] = []): ScanFinding[] {""", 1)
|
|||
|
|
s = s.replace(""" if (st.isDirectory()) {
|
|||
|
|
scanDir(abs, relPath)
|
|||
|
|
continue
|
|||
|
|
}""",
|
|||
|
|
""" if (st.isDirectory()) {
|
|||
|
|
scanDir(abs, relPath, findings)
|
|||
|
|
continue
|
|||
|
|
}""", 1)
|
|||
|
|
s = s.replace(""" const ext = dot >= 0 ? base.slice(dot).toLowerCase() : ''
|
|||
|
|
if (!SCAN_TEXT_EXT.has(ext)) continue
|
|||
|
|
if (st.size > 2 * 1024 * 1024) continue // skip huge files (unlikely to be source)
|
|||
|
|
let text: string
|
|||
|
|
try {
|
|||
|
|
text = readFileSync(abs, 'utf8')
|
|||
|
|
} catch {
|
|||
|
|
continue
|
|||
|
|
}
|
|||
|
|
for (const p of DANGEROUS_PATTERNS) {
|
|||
|
|
if (p.re.test(text)) {
|
|||
|
|
throw httpError(400, `安全检测未通过：${p.why}（${relPath}）`)
|
|||
|
|
}
|
|||
|
|
}
|
|||
|
|
}
|
|||
|
|
}""",
|
|||
|
|
""" const ext = dot >= 0 ? base.slice(dot).toLowerCase() : ''
|
|||
|
|
const hasShebangCandidate = ext === '' // 无扩展名文件也可能是脚本（按内容判 shebang）
|
|||
|
|
if (!SCAN_TEXT_EXT.has(ext) && !hasShebangCandidate) continue
|
|||
|
|
if (st.size > SCAN_MAX_BYTES) continue // 跳过超大文件（不太可能是源码）
|
|||
|
|
let text: string
|
|||
|
|
try {
|
|||
|
|
text = readFileSync(abs, 'utf8')
|
|||
|
|
} catch {
|
|||
|
|
continue
|
|||
|
|
}
|
|||
|
|
// 二进制探测 + shebang 判定（0x00 视为二进制；无扩展名须含 shebang 才继续）
|
|||
|
|
if (text.includes('\\u0000')) continue
|
|||
|
|
if (hasShebangCandidate && !/^#!\\s*\\S/.test(text.slice(0, 64))) continue
|
|||
|
|
for (const p of BLOCK_PATTERNS) {
|
|||
|
|
if (p.re.test(text)) {
|
|||
|
|
throw httpError(400, `安全检测未通过（P0 阻断）：${p.why}（${relPath}）`)
|
|||
|
|
}
|
|||
|
|
}
|
|||
|
|
for (const w of WARN_RULES) {
|
|||
|
|
if (w.re.test(text)) {
|
|||
|
|
findings.push({ rule: w.id, why: w.why, file: relPath })
|
|||
|
|
}
|
|||
|
|
}
|
|||
|
|
}
|
|||
|
|
return findings
|
|||
|
|
}""", 1)
|
|||
|
|
wr(p, s)
|
|||
|
|
|
|||
|
|
# 调用点：记录并回传 P1 findings（加法式，不改变原有阻断行为）
|
|||
|
|
sub('src/web/routes/business-plugins.ts', ' scanDir(unzipDir)',
|
|||
|
|
""" const scanFindings = scanDir(unzipDir)
|
|||
|
|
if (scanFindings.length > 0) {
|
|||
|
|
// P1 告警：不阻断，仅记录（管理员可在响应中看到）
|
|||
|
|
request.log.warn({ findings: scanFindings.slice(0, 20), count: scanFindings.length }, 'upload-scan-warnings')
|
|||
|
|
}""")
|
|||
|
|
sub('src/web/routes/skills.ts', ' scanDir(unzipDir)',
|
|||
|
|
""" const scanFindings = scanDir(unzipDir)
|
|||
|
|
if (scanFindings.length > 0) {
|
|||
|
|
request.log.warn({ findings: scanFindings.slice(0, 20), count: scanFindings.length }, 'upload-scan-warnings')
|
|||
|
|
}""")
|
|||
|
|
|
|||
|
|
# ───────────────────────── B2：编排器自愈补齐 picker ─────────────────────────
|
|||
|
|
p = 'src/supervisor/orchestrator.ts'
|
|||
|
|
s = rd(p)
|
|||
|
|
assert "import { execFileSync, spawn," in s
|
|||
|
|
s = s.replace(""" /** Stop the current main (clean) and respawn it with the same folder/patch. */""",
|
|||
|
|
""" /**
|
|||
|
|
* 档案 18 v3 收尾（第二层自愈）：确保该用户的 profile 具备"受限目录选择器"
|
|||
|
|
* （平台段 + 插件包）—— 复用平台脚本 `ensure-workspace-picker.cjs --user-id <id>`（幂等）。
|
|||
|
|
* 异步 fire-and-forget：不阻塞启动；失败只记日志（第一层由 provisioning 钩子兜底）。
|
|||
|
|
* 覆盖场景：profile 被清空/重建、手工删掉平台段、插件被卸载。
|
|||
|
|
*/
|
|||
|
|
private ensurePickerProfile(userId: string): void {
|
|||
|
|
const script =
|
|||
|
|
process.env.DSH_PICKER_ENSURE_SCRIPT ??
|
|||
|
|
join(process.cwd(), 'poc', 'workspace-scoped-picker', 'ensure-workspace-picker.cjs')
|
|||
|
|
if (!existsSync(script)) return
|
|||
|
|
try {
|
|||
|
|
const child = spawn('node', [script, '--user-id', userId], { detached: true, stdio: 'ignore' })
|
|||
|
|
child.on('error', (err) => {
|
|||
|
|
process.stderr.write(`[picker-ensure] spawn failed: ${err.message}\\n`)
|
|||
|
|
})
|
|||
|
|
child.unref()
|
|||
|
|
} catch (err) {
|
|||
|
|
process.stderr.write(
|
|||
|
|
`[picker-ensure] failed for ${userId}: ${err instanceof Error ? err.message : String(err)}\\n`,
|
|||
|
|
)
|
|||
|
|
}
|
|||
|
|
}
|
|||
|
|
|
|||
|
|
/** Stop the current main (clean) and respawn it with the same folder/patch. */""", 1)
|
|||
|
|
s = s.replace("import { chmodSync, chownSync, mkdirSync, readFileSync, writeFileSync } from 'node:fs'",
|
|||
|
|
"import { chmodSync, chownSync, existsSync, mkdirSync, readFileSync, writeFileSync } from 'node:fs'", 1)
|
|||
|
|
# 调用点：launch 前 + spawn 后
|
|||
|
|
s = s.replace(""" // 显式启动 = 新一轮：清空崩溃计数（用户重新进入后应拿到完整重试预算，档案 20）。
|
|||
|
|
this.resetCrashState(userId)""",
|
|||
|
|
""" // 显式启动 = 新一轮：清空崩溃计数（用户重新进入后应拿到完整重试预算，档案 20）。
|
|||
|
|
this.resetCrashState(userId)
|
|||
|
|
// 档案 18 v3 收尾：启动前补齐"受限目录选择器"（幂等；首登 profile 尚未创建时会自行跳过）。
|
|||
|
|
this.ensurePickerProfile(userId)""", 1)
|
|||
|
|
s = s.replace(""" if (isMain) await this.seedDefaultWorkspace(userId)""",
|
|||
|
|
""" // spawn 成功后异步再补一次：首登时 profile 刚被 dsh 创建，这一次能真正装上（第二层自愈）。
|
|||
|
|
if (isMain) this.ensurePickerProfile(userId)""", 1)
|
|||
|
|
wr(p, s)
|
|||
|
|
print('B4 + B2 已写入')
|