@lievclouds/dsh-cortex 0.4.4 → 1.0.0

package/src/host/miniapp/analyze.js

@@ -0,0 +1,1078 @@
+/**
+ * Static analysis of an unpacked mini-program tree.
+ *
+ * Everything here is read-only over the already-unpacked workspace: no network,
+ * no execution of extracted JavaScript, no access to the original WeChat cache.
+ * The analyzer extracts small, high-signal intelligence for the Cortex asset
+ * library:
+ *
+ *   - pages / subpackages from `app-config.json`,
+ *   - business domains and API endpoint candidates from text files,
+ *   - hard-coded secrets (masked in every summary / persisted masked overview),
+ *   - authz-test parameter candidates (query / path / config values).
+ *
+ * `analyzeUnpacked` keeps full secret values in the returned object so the
+ * local asset importer can persist them. `saveOverview` writes `overview.json`
+ * with masked values only plus `extracted.json` with the full values.
+ */
+
+import { closeSync, lstatSync, mkdirSync, openSync, readFileSync, readSync, readdirSync, statSync, writeFileSync } from 'node:fs'
+import { basename, dirname, extname, join, relative, resolve } from 'node:path'
+import { ERROR_CODES, WxapkgError, readPackage } from './wxapkg.js'
+
+/** Deterministic caps for one overview. */
+export const OVERVIEW_LIMITS = Object.freeze({
+  domains: 200,
+  endpoints: 300,
+  secrets: 100,
+  parameters: 100,
+  pages: 500,
+})
+
+const BINARY_PROBE_BYTES = 8 * 1024
+const MAX_ANALYZE_FILE_BYTES = 16 * 1024 * 1024
+const MAX_LISTED_FILES = 5
+const MAX_DOMAIN_EXAMPLES = 3
+const MAX_CONFIG_DEPTH = 8
+const MAX_TEXT_INSIGHT_WARNINGS = 20
+
+/* ------------------------------------------------------------------------- */
+/* Display name extraction                                                   */
+/* ------------------------------------------------------------------------- */
+
+/** Generic page titles that are navigation labels, not the mini-program name. */
+const GENERIC_TITLES = new Set([
+  '首页',
+  '我的',
+  '分类',
+  '购物车',
+  '订单',
+  '搜索',
+  '详情',
+  '活动',
+  '登录',
+  '注册',
+  '预约',
+  '立即预约',
+  '提交订单',
+  '商品详情',
+  '订单详情',
+  '我的订单',
+  '个人中心',
+  '会员中心',
+  '设置',
+  '消息',
+  '通知',
+  '列表',
+  '更多',
+  '加载中',
+  '暂无数据',
+  '我的积分',
+  '积分商城',
+  '会员注册',
+  '会员信息',
+  '我的权益',
+  '商品评价',
+  '评价中心',
+  '免费领券',
+  '运动打卡',
+  '景点',
+  '景点预约',
+  '确认信息',
+  '预约登记',
+  '医生列表',
+  '报告查询',
+  '报告信息',
+  '健康档案',
+  '选择服务',
+  '访客管理',
+  '邀请函',
+  '同行人',
+  '提货码',
+  '物流查询',
+  '我的佣金',
+  '积分分配',
+  '失物招领',
+  '礼品卡',
+  '扫码积分',
+  '拍照积分',
+  '缴费记录',
+  '企业客户预约',
+  '预约详情',
+  '畅享卡',
+  '扫码领券',
+  '分销商品',
+  '分类',
+  '活动',
+  '首页',
+])
+
+const TITLE_CJK_RE = /[\u3400-\u9fff]/
+
+function cleanName(value) {
+  if (typeof value !== 'string') return ''
+  let name = value
+    .replace(/\s+/g, ' ')
+    .replace(/^["'“”‘’《》\s]+|["'“”‘’《》\s]+$/g, '')
+    .trim()
+  // Devtools project names are often percent-encoded in project.private.config.json.
+  if (/%[0-9A-Fa-f]{2}/.test(name)) {
+    try {
+      name = decodeURIComponent(name).replace(/\s+/g, ' ').trim()
+    } catch {
+      /* keep the raw value */
+    }
+  }
+  if (name.length < 2 || name.length > 32) return ''
+  if (/^(?:https?:\/\/|\d{4,}|undefined|null|true|false|none|todo|示例|测试)$/i.test(name)) return ''
+  if (GENERIC_TITLES.has(name)) return ''
+  if (!TITLE_CJK_RE.test(name) && !/^[A-Za-z][A-Za-z0-9 ._-]{1,31}$/.test(name)) return ''
+  return name
+}
+
+function collectTitleCandidates(config, displaySource, acc) {
+  if (!config || typeof config !== 'object') return
+  const subRoots = Array.isArray(config.subPackages)
+    ? config.subPackages
+    : Array.isArray(config.subpackages)
+      ? config.subpackages
+      : []
+  const subRootsList = subRoots
+    .map((entry) => (entry && typeof entry.root === 'string' ? entry.root.replace(/\/$/, '') : ''))
+    .filter((entry) => entry.length > 0)
+  const entryPagePath =
+    typeof config.entryPagePath === 'string' ? config.entryPagePath.replace(/\.html$/, '') : ''
+
+  const globalTitle = config.global && config.global.window && config.global.window.navigationBarTitleText
+  if (typeof globalTitle === 'string' && cleanName(globalTitle)) {
+    acc.push({ title: globalTitle, weight: 3, source: displaySource })
+  }
+
+  const pageMap = config.page && typeof config.page === 'object' ? config.page : null
+  if (!pageMap) return
+  const entries = Object.entries(pageMap)
+  for (let index = 0; index < entries.length; index += 1) {
+    const [pagePath, pageConfig] = entries[index]
+    const title = pageConfig && pageConfig.window && pageConfig.window.navigationBarTitleText
+    if (typeof title !== 'string' || cleanName(title).length === 0) continue
+    const normalizedPage = String(pagePath).replace(/\.html$/, '')
+    const inSubPackage = subRootsList.some((root) => normalizedPage.startsWith(`${root}/`))
+    let weight = inSubPackage ? 1 : 3
+    if (index < 20) weight += 1
+    if (normalizedPage.endsWith('/index') || normalizedPage.endsWith('/index/index')) weight += 1
+    if (entryPagePath && normalizedPage === entryPagePath) weight += 4
+    acc.push({ title, weight, source: displaySource })
+  }
+}
+
+/**
+ * Derive a human display name from package config files. Official app names are
+ * not stored in the WeChat cache in plaintext, so this is a best-effort
+ * extraction from `project.config.json` / app config / page titles, with a
+ * stable heuristic that prefers the entry page and main-package titles.
+ */
+export function deriveAppName({ appConfig = null, projectConfig = null, pluginConfig = null } = {}) {
+  const directCandidates = [
+    ['project.config.json', projectConfig && (projectConfig.projectname || projectConfig.projectName || projectConfig.name)],
+    ['app-config.json', appConfig && (appConfig.name || appConfig.appName || appConfig.nickname || appConfig.brandName || appConfig.shopName || appConfig.mallName)],
+    ['plugin.json', pluginConfig && (pluginConfig.name || pluginConfig.pluginName)],
+  ]
+  for (const [source, value] of directCandidates) {
+    const name = cleanName(value)
+    if (name) return { appName: name, appNameSource: source }
+  }
+
+  const candidates = []
+  collectTitleCandidates(appConfig, 'app-config.json', candidates)
+  collectTitleCandidates(pluginConfig, 'plugin.json', candidates)
+  if (candidates.length === 0) return { appName: '', appNameSource: '' }
+
+  const scores = new Map()
+  for (const candidate of candidates) {
+    const name = cleanName(candidate.title)
+    if (!name) continue
+    const current = scores.get(name) || { name, score: 0, hits: 0 }
+    current.score += candidate.weight
+    current.hits += 1
+    scores.set(name, current)
+  }
+  const ranked = [...scores.values()].sort((left, right) => {
+    if (right.score !== left.score) return right.score - left.score
+    if (right.hits !== left.hits) return right.hits - left.hits
+    return left.name.length - right.name.length
+  })
+  if (ranked.length === 0) return { appName: '', appNameSource: '' }
+  return { appName: ranked[0].name, appNameSource: 'page-title' }
+}
+
+function readJsonFileQuiet(filePath, maxBytes = 2 * 1024 * 1024) {
+  try {
+    const stats = statSync(filePath)
+    if (!stats.isFile() || stats.size > maxBytes) return null
+    const parsed = JSON.parse(readFileSync(filePath, 'utf8'))
+    return parsed && typeof parsed === 'object' && !Array.isArray(parsed) ? parsed : null
+  } catch {
+    return null
+  }
+}
+
+/** Derive the display name from an already unpacked workspace (fast, no scan). */
+export function deriveAppNameFromUnpacked(unpackedPath) {
+  if (typeof unpackedPath !== 'string' || unpackedPath.length === 0) return { appName: '', appNameSource: '' }
+  const root = resolve(unpackedPath)
+  return deriveAppName({
+    appConfig: readJsonFileQuiet(join(root, 'app-config.json')) || readJsonFileQuiet(join(root, 'app.json')),
+    projectConfig:
+      readJsonFileQuiet(join(root, 'project.config.json')) || readJsonFileQuiet(join(root, 'project.private.config.json')),
+    pluginConfig: readJsonFileQuiet(join(root, 'plugin.json')),
+  })
+}
+
+/**
+ * Derive the display name from one encrypted/plain wxapkg file. Used by the
+ * scan phase (one package per app) so the list can show a Chinese name before
+ * the user unpacks anything.
+ */
+export function extractAppNameFromPackage(filePath, appid) {
+  try {
+    const parsed = readPackage(filePath, { appid })
+    const findEntry = (fileName) => {
+      const exact = parsed.files.find((entry) => entry.name === `/${fileName}`)
+      if (exact) return exact
+      return parsed.files.find((entry) => entry.name.endsWith(`/${fileName}`))
+    }
+    const readEntryJson = (fileName) => {
+      const entry = findEntry(fileName)
+      if (!entry || entry.size > 2 * 1024 * 1024) return null
+      try {
+        const parsedJson = JSON.parse(
+          parsed.buffer.subarray(entry.offset, entry.offset + entry.size).toString('utf8'),
+        )
+        return parsedJson && typeof parsedJson === 'object' && !Array.isArray(parsedJson) ? parsedJson : null
+      } catch {
+        return null
+      }
+    }
+    return deriveAppName({
+      appConfig: readEntryJson('app-config.json') || readEntryJson('app.json'),
+      projectConfig: readEntryJson('project.config.json') || readEntryJson('project.private.config.json'),
+      pluginConfig: readEntryJson('plugin.json'),
+    })
+  } catch {
+    return { appName: '', appNameSource: '' }
+  }
+}
+
+/** Workspace metadata files are counted but not mined (they echo file paths). */
+const ANALYSIS_METADATA_FILES = new Set(['manifest.json', 'overview.json', 'extracted.json'])
+
+const HTTP_METHODS = ['GET', 'POST', 'PUT', 'PATCH', 'DELETE', 'OPTIONS', 'HEAD']
+const NOISE_DOMAINS = new Set([
+  'localhost',
+  '127.0.0.1',
+  '0.0.0.0',
+  'w3.org',
+  'www.w3.org',
+  'w3c.org',
+  'www.w3c.org',
+  'schema.org',
+  'www.schema.org',
+  'example.com',
+  'www.example.com',
+])
+const NOISE_DOMAIN_SUFFIXES = ['.w3.org', '.w3c.org', '.schema.org', '.example.com']
+const STATIC_ASSET_RE = /\.(?:js|json|wxml|wxss|wxs|html?|css|png|jpe?g|gif|svg|webp|ico|bmp|ttf|otf|woff2?|eot|mp[34]|wav|amr|map)$/i
+/**
+ * References to bundled page/component/plugin modules (`"/pages/index/index"`,
+ * `"/./components/x/x"`) are not HTTP endpoints; keeping them would drown the
+ * real API candidates on large packages.
+ */
+const MODULE_REFERENCE_RE = /^\/(?:pages|components|__plugin__|miniprogram_npm)\/|^\/\.\//
+const DOMAIN_RE = /https?:\/\/([a-z0-9.-]+\.[a-z]{2,})(?::\d+)?/gi
+const ENDPOINT_RE = /["'](\/[a-zA-Z0-9_\-./{}:]+(?:\?[^"']*)?)["']/g
+const QUERY_PAIR_RE = /^[A-Za-z_][A-Za-z0-9_.-]{0,60}$/
+const PARAM_KEY_RE = /(tenant|user|order|member|card|shop|store|org|account|id)$/i
+const SECRET_KEY_RE = /(secret|password|passwd|pwd|token|api[_-]?key|access[_-]?key|private[_-]?key)$/i
+const PATH_PLACEHOLDER_RE = /^(?:\{([^{}]+)\}|:([A-Za-z_][A-Za-z0-9_]*))$/
+
+const SECRET_PATTERNS = [
+  {
+    re: /AKID[0-9A-Za-z]{13,}/g,
+    name: 'tencent_akid',
+    info_type: 'API_Key',
+    potential_impact: '腾讯云 API 密钥泄露，可能被用于调用云资源或读取数据',
+  },
+  {
+    re: /sk-[A-Za-z0-9]{20,}/g,
+    name: 'openai_api_key',
+    info_type: 'API_Key',
+    potential_impact: '第三方模型 API Key 泄露，可能被滥用并产生费用',
+  },
+  {
+    re: /ghp_[A-Za-z0-9]{20,}/g,
+    name: 'github_token',
+    info_type: 'Token',
+    potential_impact: 'GitHub Token 泄露，可能访问仓库或流水线',
+  },
+  {
+    re: /eyJ[A-Za-z0-9_-]{20,}\.[A-Za-z0-9_-]{10,}/g,
+    name: 'jwt',
+    info_type: 'Token',
+    potential_impact: 'JWT 泄露，可能被重放或用于伪造会话',
+  },
+]
+const ASSIGNMENT_RE =
+  /(api[_-]?key|secret|token|app[_-]?secret|access[_-]?key)\s*[:=]\s*['"]([^'"]{8,120})['"]/gi
+/** Reject minified-code fragments such as `).concat(d,` captured by ASSIGNMENT_RE. */
+const ASSIGNMENT_VALUE_RE = /^[A-Za-z0-9_\-.\/+=:@]{8,120}$/
+/** Full PEM block; multi-line scan runs at content level (bare headers are noise). */
+const PEM_KEY_RE =
+  /-----BEGIN (?:RSA |EC |OPENSSH )?PRIVATE KEY-----[\s\S]{20,}?-----END (?:RSA |EC |OPENSSH )?PRIVATE KEY-----/g
+/** `tenantId:"10001"` / `userId: 42` style config samples in JS/config text. */
+const CONFIG_ASSIGNMENT_RE =
+  /(?:^|[{,;\s])["']?([A-Za-z_$][\w$]{0,60})["']?\s*[:=]\s*(?:"([^"\r\n]{1,120})"|'([^'\r\n]{1,120})'|(\d{1,12}))\s*(?=[,;}\s]|$)/gm
+
+function readPrefix(filePath, bytes) {
+  if (bytes <= 0) return Buffer.alloc(0)
+  const buffer = Buffer.alloc(bytes)
+  const fd = openSync(filePath, 'r')
+  try {
+    const read = readSync(fd, buffer, 0, bytes, 0)
+    return buffer.subarray(0, read)
+  } finally {
+    closeSync(fd)
+  }
+}
+
+function splitLines(text) {
+  if (text.length === 0) return []
+  const normalized = text.replace(/\r\n/g, '\n').replace(/\r/g, '\n')
+  const lines = normalized.split('\n')
+  if (lines.length > 0 && lines[lines.length - 1] === '') lines.pop()
+  return lines
+}
+
+const toPosix = (value) => String(value).split('\\').join('/')
+
+function walkFiles(rootPath) {
+  const files = []
+  const stack = [{ dir: rootPath, rel: '' }]
+  while (stack.length > 0) {
+    const current = stack.pop()
+    let dirents
+    try {
+      dirents = readdirSync(current.dir, { withFileTypes: true })
+    } catch {
+      continue
+    }
+    dirents.sort((left, right) => (left.name < right.name ? -1 : left.name > right.name ? 1 : 0))
+    for (const dirent of dirents) {
+      if (dirent.isSymbolicLink()) continue
+      const absolute = join(current.dir, dirent.name)
+      const rel = current.rel ? `${current.rel}/${dirent.name}` : dirent.name
+      if (dirent.isDirectory()) {
+        stack.push({ dir: absolute, rel })
+        continue
+      }
+      if (!dirent.isFile()) continue
+      let size = 0
+      try {
+        size = statSync(absolute).size
+      } catch {
+        size = 0
+      }
+      files.push({ path: rel, abs: absolute, size })
+    }
+  }
+  files.sort((left, right) => (left.path < right.path ? -1 : left.path > right.path ? 1 : 0))
+  return files
+}
+
+export function maskSecret(value) {
+  const text = typeof value === 'string' ? value : String(value === undefined || value === null ? '' : value)
+  if (text.length <= 8) return '***'
+  return `${text.slice(0, 4)}\u2026${text.slice(-2)}`
+}
+
+function isPlaceholder(value) {
+  const text = String(value || '').trim()
+  if (text.length === 0) return true
+  const lower = text.toLowerCase()
+  if (lower.length < 8) return true
+  if (/^(?:undefined|null|true|false|changeme|password|secret|example|sample|placeholder|todo|none)$/i.test(lower)) {
+    return true
+  }
+  if (/(?:your[_ -]?(?:key|secret|token|appid)|placeholder|example|xxxxx|xxx-xxx)/i.test(lower)) return true
+  if (/^<.*>$/.test(lower) || /^\*.+$/.test(lower)) return true
+  if (/^(.)\1{6,}$/.test(lower)) return true // 'aaaaaaaa', '00000000'
+  if (new Set(lower).size <= 2) return true
+  return false
+}
+
+function infoTypeForKey(key) {
+  const lower = String(key || '').toLowerCase()
+  if (lower.includes('token')) return 'Token'
+  if (lower.includes('secret') || lower.includes('password') || lower.includes('passwd') || lower.includes('pwd')) {
+    return 'Credential'
+  }
+  if (lower.includes('key')) return 'API_Key'
+  return 'Other'
+}
+
+function isNoiseDomain(host) {
+  const lower = String(host || '').toLowerCase()
+  if (NOISE_DOMAINS.has(lower)) return true
+  if (/^127\./.test(lower) || /^10\./.test(lower) || /^192\.168\./.test(lower)) return true
+  if (NOISE_DOMAIN_SUFFIXES.some((suffix) => lower.endsWith(suffix))) return true
+  return false
+}
+
+function formatBytes(bytes) {
+  const value = Number(bytes) || 0
+  if (value < 1024) return `${value} B`
+  if (value < 1024 * 1024) return `${(value / 1024).toFixed(1)} KB`
+  if (value < 1024 * 1024 * 1024) return `${(value / (1024 * 1024)).toFixed(1)} MB`
+  return `${(value / (1024 * 1024 * 1024)).toFixed(2)} GB`
+}
+
+function decodeSafe(value) {
+  try {
+    return decodeURIComponent(String(value).replace(/\+/g, ' '))
+  } catch {
+    return String(value)
+  }
+}
+
+function isSampleValue(value) {
+  if (typeof value === 'number') {
+    if (!Number.isInteger(value) || value < 0) return ''
+    const text = String(value)
+    return text.length <= 12 ? text : ''
+  }
+  if (typeof value !== 'string') return ''
+  const text = value.trim()
+  if (/^\d{1,12}$/.test(text)) return text
+  if (/^wx[0-9a-z]{8,}$/i.test(text)) return text
+  return ''
+}
+
+function addParameter(map, { name, endpoint, location, value, source }) {
+  const paramName = String(name || '').trim()
+  if (paramName.length === 0 || paramName.length > 80) return
+  const targetEndpoint = String(endpoint || '').trim()
+  const targetLocation = String(location || 'query')
+  const key = `${paramName}\u0000${targetEndpoint}\u0000${targetLocation}`
+  let entry = map.get(key)
+  if (!entry) {
+    entry = { name: paramName, endpoint: targetEndpoint, location: targetLocation, values: [], source: String(source || '') }
+    map.set(key, entry)
+  }
+  if (value === undefined || value === null) return
+  const text = String(value).trim()
+  if (text.length === 0 || text.length > 120) return
+  if (entry.values.length >= 5 || entry.values.includes(text)) return
+  entry.values.push(text)
+}
+
+function configParameters(config, sourceName, map) {
+  const visit = (value, depth) => {
+    if (depth > MAX_CONFIG_DEPTH || value === null || typeof value !== 'object') return
+    if (Array.isArray(value)) {
+      for (const item of value) visit(item, depth + 1)
+      return
+    }
+    for (const [key, child] of Object.entries(value)) {
+      if (child !== null && typeof child === 'object') {
+        visit(child, depth + 1)
+        continue
+      }
+      if (!PARAM_KEY_RE.test(key) || SECRET_KEY_RE.test(key)) continue
+      const sample = isSampleValue(child)
+      if (sample.length === 0) continue
+      addParameter(map, { name: key, endpoint: sourceName, location: 'body', value: sample, source: sourceName })
+    }
+  }
+  visit(config, 0)
+}
+
+function scanConfigAssignments(content, sourceName, parameterMap) {
+  CONFIG_ASSIGNMENT_RE.lastIndex = 0
+  let match
+  while ((match = CONFIG_ASSIGNMENT_RE.exec(content)) !== null) {
+    const key = match[1]
+    if (!key || !PARAM_KEY_RE.test(key) || SECRET_KEY_RE.test(key)) continue
+    const rawValue = match[2] !== undefined ? match[2] : match[3] !== undefined ? match[3] : match[4]
+    const sample = isSampleValue(rawValue)
+    if (sample.length === 0) continue
+    addParameter(parameterMap, {
+      name: key,
+      endpoint: sourceName,
+      location: 'body',
+      value: sample,
+      source: sourceName,
+    })
+  }
+}
+
+function nearbyMethod(content, endIndex) {
+  const window = content.slice(endIndex, Math.min(content.length, endIndex + 120))
+  const match = /method\s*[:=]\s*['"]([A-Za-z]{3,7})['"]/i.exec(window)
+  if (!match) return ''
+  const method = match[1].toUpperCase()
+  return HTTP_METHODS.includes(method) ? method : ''
+}
+
+/**
+ * Analyze one unpacked mini-program workspace.
+ *
+ * @param {{unpackedPath:string, appid?:string, version?:string, manifest?:object|null}} input
+ * @returns {object} Overview (see contract §3.6)
+ */
+export function analyzeUnpacked({ unpackedPath, appid = '', version = '', manifest = null } = {}) {
+  const rootPath = typeof unpackedPath === 'string' && unpackedPath.trim().length > 0 ? resolve(unpackedPath) : ''
+  if (!rootPath) throw new WxapkgError(ERROR_CODES.INVALID_PATH, 'unpackedPath is required')
+  let rootStats
+  try {
+    rootStats = lstatSync(rootPath)
+  } catch {
+    throw new WxapkgError(ERROR_CODES.INVALID_PATH, `unpacked directory does not exist: ${rootPath}`)
+  }
+  if (rootStats.isSymbolicLink() || !rootStats.isDirectory()) {
+    throw new WxapkgError(ERROR_CODES.INVALID_PATH, 'unpacked path is not a directory')
+  }
+
+  const warnings = []
+  const files = walkFiles(rootPath)
+  const extensionMap = new Map()
+  const domainMap = new Map()
+  const endpointMap = new Map()
+  const parameterMap = new Map()
+  const secrets = []
+  const secretKeys = new Set()
+
+  let totalBytes = 0
+  let textCount = 0
+  let binaryCount = 0
+  let partialTextFiles = 0
+  let truncatedSecrets = false
+  let appConfig = null
+  let appConfigName = ''
+  let projectConfig = null
+  let pluginConfig = null
+
+  const rememberDomain = (host, url, sourceFile) => {
+    const lower = String(host || '').toLowerCase()
+    if (!lower || isNoiseDomain(lower)) return
+    let entry = domainMap.get(lower)
+    if (!entry) {
+      entry = { host: lower, count: 0, examples: [], files: [] }
+      domainMap.set(lower, entry)
+    }
+    entry.count += 1
+    if (url && entry.examples.length < MAX_DOMAIN_EXAMPLES && !entry.examples.includes(url)) {
+      entry.examples.push(url.length > 300 ? url.slice(0, 300) : url)
+    }
+    if (entry.files.length < MAX_LISTED_FILES && !entry.files.includes(sourceFile)) entry.files.push(sourceFile)
+  }
+
+  const rememberEndpoint = (path, method, sourceFile) => {
+    const key = `${method || ''}\u0000${path}`
+    let entry = endpointMap.get(key)
+    if (!entry) {
+      entry = { path, count: 0, files: [] }
+      if (method) entry.method = method
+      endpointMap.set(key, entry)
+    }
+    entry.count += 1
+    if (entry.files.length < MAX_LISTED_FILES && !entry.files.includes(sourceFile)) entry.files.push(sourceFile)
+  }
+
+  const rememberSecret = (infoType, name, value, sourceFile, line) => {
+    if (secrets.length >= OVERVIEW_LIMITS.secrets) {
+      truncatedSecrets = true
+      return
+    }
+    if (isPlaceholder(value)) return
+    const dedupeKey = `${infoType}\u0000${value}\u0000${sourceFile}\u0000${line}`
+    if (secretKeys.has(dedupeKey)) return
+    secretKeys.add(dedupeKey)
+    secrets.push({
+      info_type: infoType,
+      name,
+      value_masked: maskSecret(value),
+      value,
+      source: sourceFile,
+      line: Number.isFinite(Number(line)) ? Math.trunc(Number(line)) : 0,
+      potential_impact:
+        infoType === 'Token'
+          ? '令牌泄露，可能被重放或用于伪造会话'
+          : infoType === 'API_Key'
+            ? 'API 密钥泄露，可能被用于调用后端或云服务'
+            : infoType === 'Credential'
+              ? '凭据/私钥泄露，可能导致未授权访问'
+              : '硬编码敏感信息泄露',
+    })
+  }
+
+  const scanSecrets = (content, sourceFile) => {
+    const lines = splitLines(content)
+    for (let index = 0; index < lines.length; index += 1) {
+      const line = lines[index]
+      for (const pattern of SECRET_PATTERNS) {
+        pattern.re.lastIndex = 0
+        let match
+        while ((match = pattern.re.exec(line)) !== null) {
+          rememberSecret(pattern.info_type, pattern.name, match[0], sourceFile, index + 1)
+        }
+      }
+      ASSIGNMENT_RE.lastIndex = 0
+      let assignment
+      while ((assignment = ASSIGNMENT_RE.exec(line)) !== null) {
+        if (!ASSIGNMENT_VALUE_RE.test(assignment[2])) continue
+        rememberSecret(infoTypeForKey(assignment[1]), assignment[1].toLowerCase(), assignment[2], sourceFile, index + 1)
+      }
+    }
+    // A real embedded private key may span lines; scan the whole content once.
+    // Bare headers and code fragments that merely mention the markers are not
+    // secrets: the body must look like PEM base64.
+    PEM_KEY_RE.lastIndex = 0
+    let pem
+    while ((pem = PEM_KEY_RE.exec(content)) !== null) {
+      const bodyMatch = /^-----BEGIN (?:RSA |EC |OPENSSH )?PRIVATE KEY-----([\s\S]*?)-----END /.exec(pem[0])
+      const body = bodyMatch ? bodyMatch[1] : ''
+      if (!/^(?:[A-Za-z0-9+/=]|\\[nr]|\s)+$/.test(body)) continue
+      const line = content.slice(0, pem.index).split('\n').length
+      rememberSecret('Credential', 'private_key', pem[0].slice(0, 4096), sourceFile, line)
+    }
+  }
+
+  const scanEndpoints = (content, sourceFile) => {
+    ENDPOINT_RE.lastIndex = 0
+    let match
+    while ((match = ENDPOINT_RE.exec(content)) !== null) {
+      const raw = match[1]
+      const queryIndex = raw.indexOf('?')
+      const rawPath = queryIndex >= 0 ? raw.slice(0, queryIndex) : raw
+      const query = queryIndex >= 0 ? raw.slice(queryIndex + 1) : ''
+      if (!rawPath.startsWith('/')) continue
+      const segments = rawPath.split('/').filter(Boolean)
+      const startsWithApiRoot = /^\/(?:api|wx|miniapp|mp)(?:\/|$)/.test(rawPath)
+      if (segments.length < 2 && !startsWithApiRoot) continue
+      if (STATIC_ASSET_RE.test(rawPath) || MODULE_REFERENCE_RE.test(rawPath)) continue
+      const method = nearbyMethod(content, match.index + match[0].length)
+      rememberEndpoint(rawPath, method, sourceFile)
+
+      if (query.length > 0) {
+        for (const pair of query.split('&')) {
+          const equals = pair.indexOf('=')
+          const rawName = equals >= 0 ? pair.slice(0, equals) : pair
+          const rawValue = equals >= 0 ? pair.slice(equals + 1) : ''
+          const name = decodeSafe(rawName).trim()
+          if (!QUERY_PAIR_RE.test(name) || SECRET_KEY_RE.test(name)) continue
+          addParameter(parameterMap, {
+            name,
+            endpoint: rawPath,
+            location: 'query',
+            value: rawValue.length > 0 ? decodeSafe(rawValue).trim() : undefined,
+            source: sourceFile,
+          })
+        }
+      }
+
+      for (const segment of segments) {
+        const placeholder = PATH_PLACEHOLDER_RE.exec(segment)
+        if (!placeholder) continue
+        const name = (placeholder[1] || placeholder[2] || '').trim()
+        if (name.length === 0 || SECRET_KEY_RE.test(name)) continue
+        addParameter(parameterMap, { name, endpoint: rawPath, location: 'path', source: sourceFile })
+      }
+    }
+  }
+
+  for (const file of files) {
+    totalBytes += file.size
+    const extension = extname(file.path).toLowerCase() || '(none)'
+    let extensionEntry = extensionMap.get(extension)
+    if (!extensionEntry) {
+      extensionEntry = { ext: extension, count: 0, bytes: 0 }
+      extensionMap.set(extension, extensionEntry)
+    }
+    extensionEntry.count += 1
+    extensionEntry.bytes += file.size
+
+    let probe
+    try {
+      probe = readPrefix(file.abs, Math.min(file.size, BINARY_PROBE_BYTES))
+    } catch {
+      continue
+    }
+    if (probe.includes(0)) {
+      binaryCount += 1
+      continue
+    }
+    textCount += 1
+
+    if (ANALYSIS_METADATA_FILES.has(file.path)) continue
+
+    let content
+    try {
+      const readBytes = Math.min(file.size, MAX_ANALYZE_FILE_BYTES)
+      if (file.size > MAX_ANALYZE_FILE_BYTES) partialTextFiles += 1
+      content = readPrefix(file.abs, readBytes).toString('utf8')
+    } catch {
+      continue
+    }
+
+    if (file.path === 'app-config.json' || file.path === 'app.json') {
+      try {
+        const parsed = JSON.parse(content)
+        if (parsed && typeof parsed === 'object' && !Array.isArray(parsed)) {
+          if (!appConfig) {
+            appConfig = parsed
+            appConfigName = file.path
+          }
+          configParameters(parsed, file.path, parameterMap)
+        }
+      } catch (error) {
+        warnings.push(`could not parse ${file.path}: ${error instanceof Error ? error.message : String(error)}`)
+      }
+    }
+    if (
+      (file.path === 'project.config.json' || file.path === 'project.private.config.json') &&
+      projectConfig === null
+    ) {
+      try {
+        const parsed = JSON.parse(content)
+        if (parsed && typeof parsed === 'object' && !Array.isArray(parsed)) projectConfig = parsed
+      } catch {
+        /* optional metadata */
+      }
+    }
+    if (file.path === 'plugin.json' && pluginConfig === null) {
+      try {
+        const parsed = JSON.parse(content)
+        if (parsed && typeof parsed === 'object' && !Array.isArray(parsed)) pluginConfig = parsed
+      } catch {
+        /* optional metadata */
+      }
+    }
+
+    DOMAIN_RE.lastIndex = 0
+    let domainMatch
+    while ((domainMatch = DOMAIN_RE.exec(content)) !== null) {
+      rememberDomain(domainMatch[1], domainMatch[0], file.path)
+    }
+
+    scanEndpoints(content, file.path)
+    scanSecrets(content, file.path)
+    scanConfigAssignments(content, file.path, parameterMap)
+  }
+
+  if (partialTextFiles > 0) {
+    warnings.push(`${Math.min(partialTextFiles, MAX_TEXT_INSIGHT_WARNINGS)} large text file(s) were only partially analyzed (${formatBytes(MAX_ANALYZE_FILE_BYTES)} cap)`)
+  }
+
+  /* ------------------------------ pages ------------------------------ */
+  let pages = []
+  let subPackages = []
+  if (appConfig) {
+    const rawPages = Array.isArray(appConfig.pages) ? appConfig.pages : []
+    pages = rawPages.filter((page) => typeof page === 'string' && page.trim().length > 0).map((page) => page.trim())
+    if (pages.length > OVERVIEW_LIMITS.pages) {
+      warnings.push(`pages truncated at ${OVERVIEW_LIMITS.pages}`)
+      pages = pages.slice(0, OVERVIEW_LIMITS.pages)
+    }
+    const rawSubPackages = Array.isArray(appConfig.subPackages)
+      ? appConfig.subPackages
+      : Array.isArray(appConfig.subpackages)
+        ? appConfig.subpackages
+        : []
+    subPackages = []
+    for (const item of rawSubPackages.slice(0, 100)) {
+      if (!item || typeof item !== 'object') continue
+      const root = typeof item.root === 'string' ? item.root.trim() : ''
+      const raw = Array.isArray(item.pages) ? item.pages : []
+      const subPages = raw.filter((page) => typeof page === 'string' && page.trim().length > 0).map((page) => page.trim())
+      if (root.length === 0 && subPages.length === 0) continue
+      subPackages.push({ root, pages: subPages.slice(0, OVERVIEW_LIMITS.pages) })
+    }
+  } else {
+    warnings.push('app-config.json was not found; pages/subPackages are empty')
+  }
+
+  /* ------------------------------ domains ------------------------------ */
+  const sortedDomains = [...domainMap.values()].sort((left, right) => {
+    if (right.count !== left.count) return right.count - left.count
+    return left.host < right.host ? -1 : left.host > right.host ? 1 : 0
+  })
+  if (sortedDomains.length > OVERVIEW_LIMITS.domains) {
+    warnings.push(`domains truncated at ${OVERVIEW_LIMITS.domains}`)
+  }
+  const domains = sortedDomains.slice(0, OVERVIEW_LIMITS.domains).map((entry) => ({
+    host: entry.host,
+    count: entry.count,
+    examples: entry.examples.slice(),
+    files: entry.files.slice(),
+  }))
+
+  /* ----------------------------- endpoints ----------------------------- */
+  const sortedEndpoints = [...endpointMap.values()].sort((left, right) => {
+    if (right.count !== left.count) return right.count - left.count
+    if (left.path !== right.path) return left.path < right.path ? -1 : 1
+    return String(left.method || '') < String(right.method || '') ? -1 : 1
+  })
+  if (sortedEndpoints.length > OVERVIEW_LIMITS.endpoints) {
+    warnings.push(`endpoints truncated at ${OVERVIEW_LIMITS.endpoints}`)
+  }
+  const endpoints = sortedEndpoints.slice(0, OVERVIEW_LIMITS.endpoints).map((entry) => {
+    const item = { path: entry.path, count: entry.count, files: entry.files.slice() }
+    if (entry.method) item.method = entry.method
+    return item
+  })
+
+  /* ----------------------------- parameters ----------------------------- */
+  const parameters = [...parameterMap.values()]
+    .map((entry) => {
+      const item = {
+        name: entry.name,
+        endpoint: entry.endpoint,
+        location: entry.location,
+        values: entry.values.slice(0, 5),
+        source: entry.source,
+      }
+      if (entry.values.length > 0) {
+        if (entry.values.every((value) => /^\d+$/.test(value))) item.pattern = 'numeric'
+        else if (entry.values.every((value) => /^wx[0-9a-z]+$/i.test(value))) item.pattern = 'wxid'
+        else item.pattern = 'string'
+      }
+      return item
+    })
+    .sort((left, right) => {
+      if (left.endpoint !== right.endpoint) return left.endpoint < right.endpoint ? -1 : 1
+      if (left.name !== right.name) return left.name < right.name ? -1 : 1
+      return left.location < right.location ? -1 : 1
+    })
+  if (parameters.length > OVERVIEW_LIMITS.parameters) {
+    warnings.push(`parameters truncated at ${OVERVIEW_LIMITS.parameters}`)
+  }
+
+  if (truncatedSecrets) warnings.push(`secrets truncated at ${OVERVIEW_LIMITS.secrets}`)
+
+  /* ------------------------- files / extensions ------------------------- */
+  const byExtension = [...extensionMap.values()]
+    .sort((left, right) => {
+      if (right.count !== left.count) return right.count - left.count
+      return left.ext < right.ext ? -1 : left.ext > right.ext ? 1 : 0
+    })
+    .slice(0, 12)
+
+  const largestFiles = files
+    .slice()
+    .sort((left, right) => {
+      if (right.size !== left.size) return right.size - left.size
+      return left.path < right.path ? -1 : left.path > right.path ? 1 : 0
+    })
+    .slice(0, 10)
+    .map((file) => ({ path: file.path, size: file.size }))
+
+  const contentFileCount = files.filter((file) => !ANALYSIS_METADATA_FILES.has(file.path)).length
+  if (manifest && typeof manifest === 'object' && Array.isArray(manifest.files) && manifest.files.length !== contentFileCount) {
+    warnings.push(`manifest lists ${manifest.files.length} file(s) but the workspace contains ${contentFileCount}`)
+  }
+  const nameInfo = deriveAppName({ appConfig, projectConfig, pluginConfig })
+
+  return {
+    appid: typeof appid === 'string' ? appid : String(appid || ''),
+    version: typeof version === 'string' ? version : String(version || ''),
+    generatedAt: new Date().toISOString(),
+    appName: nameInfo.appName,
+    appNameSource: nameInfo.appNameSource,
+    files: { total: files.length, text: textCount, binary: binaryCount, bytes: totalBytes },
+    byExtension,
+    pages,
+    subPackages,
+    domains,
+    endpoints,
+    secrets: secrets.slice(0, OVERVIEW_LIMITS.secrets),
+    parameters: parameters.slice(0, OVERVIEW_LIMITS.parameters),
+    largestFiles,
+    warnings,
+  }
+}
+
+/**
+ * Render a compact, token-friendly overview. Secret values are NEVER printed:
+ * only `value_masked` is used.
+ *
+ * @param {object} overview
+ * @param {{maxChars?:number}} [options]
+ * @returns {string}
+ */
+export function buildOverviewText(overview, { maxChars = 8000 } = {}) {
+  const data = overview && typeof overview === 'object' ? overview : {}
+  const cap = Number.isFinite(Number(maxChars)) ? Math.max(200, Math.trunc(Number(maxChars))) : 8000
+  const lines = []
+  const push = (line = '') => lines.push(line)
+  const list = (items, render, capItems) => {
+    const visible = items.slice(0, capItems)
+    for (const item of visible) push(render(item))
+    if (items.length > visible.length) push(`… and ${items.length - visible.length} more`)
+  }
+
+  push(`Mini-program: ${data.appName ? `${data.appName} · ` : ''}${data.appid || '(unknown)'}${data.version ? `@${data.version}` : ''}`)
+  if (data.generatedAt) push(`Generated: ${data.generatedAt}`)
+  const files = data.files && typeof data.files === 'object' ? data.files : {}
+  push(
+    `Files: ${Number(files.total) || 0} total (${Number(files.text) || 0} text, ${Number(files.binary) || 0} binary), ${formatBytes(files.bytes)}`,
+  )
+
+  const pages = Array.isArray(data.pages) ? data.pages : []
+  push('')
+  push(`Pages (${pages.length}):`)
+  list(pages, (page) => `- ${page}`, 25)
+
+  const subPackages = Array.isArray(data.subPackages) ? data.subPackages : []
+  if (subPackages.length > 0) {
+    push('')
+    push(`Subpackages (${subPackages.length}):`)
+    list(
+      subPackages,
+      (item) => `- ${item.root || '(root)'}: ${Array.isArray(item.pages) ? item.pages.length : 0} page(s)`,
+      10,
+    )
+  }
+
+  const domains = Array.isArray(data.domains) ? data.domains : []
+  push('')
+  push(`Domains (${domains.length}):`)
+  list(domains, (item) => `- ${item.host} (${Number(item.count) || 0} hit(s))`, 20)
+
+  const endpoints = Array.isArray(data.endpoints) ? data.endpoints : []
+  push('')
+  push(`Endpoint candidates (${endpoints.length}):`)
+  list(endpoints, (item) => `- ${item.method ? `${item.method} ` : ''}${item.path} (${Number(item.count) || 0})`, 30)
+
+  const secrets = Array.isArray(data.secrets) ? data.secrets : []
+  push('')
+  push(`Secrets (${secrets.length}, masked):`)
+  list(
+    secrets,
+    (item) =>
+      `- ${item.info_type || 'Other'} ${item.name || ''} ${item.value_masked || '***'}`.trimEnd() +
+      ` (${item.source || 'unknown'}${item.line ? `:${item.line}` : ''})`,
+    30,
+  )
+
+  const parameters = Array.isArray(data.parameters) ? data.parameters : []
+  push('')
+  push(`Test parameter candidates (${parameters.length}):`)
+  list(
+    parameters,
+    (item) =>
+      `- ${item.name} [${item.location || 'query'}] @ ${item.endpoint || 'unknown'}` +
+      (Array.isArray(item.values) && item.values.length > 0 ? ` values=${item.values.join(',')}` : ''),
+    20,
+  )
+
+  const largest = Array.isArray(data.largestFiles) ? data.largestFiles : []
+  if (largest.length > 0) {
+    push('')
+    push('Largest files:')
+    list(largest, (item) => `- ${item.path} (${formatBytes(item.size)})`, 5)
+  }
+
+  const warnings = Array.isArray(data.warnings) ? data.warnings : []
+  if (warnings.length > 0) {
+    push('')
+    push('Warnings:')
+    list(warnings, (warning) => `- ${warning}`, 10)
+  }
+
+  let text = lines.join('\n')
+  if (text.length > cap) {
+    const suffix = '\n…(truncated)'
+    text = text.slice(0, Math.max(0, cap - suffix.length)) + suffix
+  }
+  return text
+}
+
+function maskedOverview(overview) {
+  const secrets = Array.isArray(overview.secrets)
+    ? overview.secrets.map((secret) => ({
+        info_type: secret && secret.info_type ? secret.info_type : 'Other',
+        name: secret && secret.name ? String(secret.name) : '',
+        value_masked:
+          secret && typeof secret.value_masked === 'string' && secret.value_masked.length > 0
+            ? secret.value_masked
+            : maskSecret(secret && secret.value),
+        source: secret && secret.source ? String(secret.source) : '',
+        line: Number(secret && secret.line) || 0,
+        potential_impact: secret && secret.potential_impact ? String(secret.potential_impact) : '',
+      }))
+    : []
+  return { ...overview, secrets }
+}
+
+function extractedPayload(overview) {
+  const secrets = Array.isArray(overview.secrets)
+    ? overview.secrets.map((secret) => ({
+        info_type: secret && secret.info_type ? secret.info_type : 'Other',
+        name: secret && secret.name ? String(secret.name) : '',
+        value: secret && typeof secret.value === 'string' ? secret.value : String((secret && secret.value) || ''),
+        source: secret && secret.source ? String(secret.source) : '',
+        line: Number(secret && secret.line) || 0,
+        potential_impact: secret && secret.potential_impact ? String(secret.potential_impact) : '',
+      }))
+    : []
+  return {
+    appid: overview.appid || '',
+    version: overview.version || '',
+    generatedAt: overview.generatedAt || '',
+    secrets,
+    endpoints: Array.isArray(overview.endpoints) ? overview.endpoints : [],
+    parameters: Array.isArray(overview.parameters) ? overview.parameters : [],
+    domains: Array.isArray(overview.domains) ? overview.domains : [],
+    pages: Array.isArray(overview.pages) ? overview.pages : [],
+    subPackages: Array.isArray(overview.subPackages) ? overview.subPackages : [],
+  }
+}
+
+/**
+ * Persist the analyzer output inside the unpacked directory.
+ *
+ * `overview.json` contains masked secret values only; `extracted.json` contains
+ * the full values for local asset import.
+ *
+ * @param {string} unpackedPath
+ * @param {object} overview
+ */
+export function saveOverview(unpackedPath, overview) {
+  const root = typeof unpackedPath === 'string' && unpackedPath.trim().length > 0 ? resolve(unpackedPath) : ''
+  if (!root) throw new WxapkgError(ERROR_CODES.INVALID_PATH, 'unpackedPath is required')
+  const data = overview && typeof overview === 'object' ? overview : {}
+  mkdirSync(root, { recursive: true })
+  writeFileSync(join(root, 'overview.json'), JSON.stringify(maskedOverview(data), null, 2), 'utf8')
+  writeFileSync(join(root, 'extracted.json'), JSON.stringify(extractedPayload(data), null, 2), 'utf8')
+}
+
+/**
+ * Read a previously saved `overview.json` (masked secrets only).
+ *
+ * @param {string} unpackedPath
+ * @returns {object|null}
+ */
+export function loadOverview(unpackedPath) {
+  if (typeof unpackedPath !== 'string' || unpackedPath.trim().length === 0) return null
+  const root = resolve(unpackedPath)
+  const candidates = [join(root, 'overview.json')]
+  if (basename(root) === 'unpacked') candidates.push(join(dirname(root), 'overview.json'))
+  for (const candidate of candidates) {
+    try {
+      const parsed = JSON.parse(readFileSync(candidate, 'utf8'))
+      if (parsed && typeof parsed === 'object' && !Array.isArray(parsed)) return parsed
+    } catch {
+      /* try the next location */
+    }
+  }
+  return null
+}