mooncat-browser 0.1.0

package/browser-op/web/humanize.js

@@ -0,0 +1,522 @@
+// -*- coding: utf-8 -*-
+//
+// web/humanize — 拟人化行为注入（路由无关，全走 operate）
+//
+// 设计：随机触发，不是每次 action 都注入。每个 operate 有概率走 humanize 序列，
+// 概率按 riskLevel 调。避免"每次必注入"本身变成新的固定模式。
+//
+// 嵌入方式：operate 顶部作为隐式中间件，调用方无感：
+//   if (ph.humanize && 触发) await humanize.beforeAction(ph, action, params)
+//   执行 action
+//   if (ph.humanize && 触发) await humanize.afterAction(ph, action)
+//
+// beforeAction 末尾对有 selector 的 action 必归位（reposition）：
+//   平滑滚回目标 + 贝塞尔移到目标中心 + focus。
+//   保证目标 action 命中（不被注入的行为搞砸）。
+//
+// 依赖：./browser（operate）
+
+'use strict'
+
+// 注：不直接 require('./browser')，避免循环依赖。
+// operate 函数由 createHumanizer 调用时从外部注入，或用懒加载获取。
+let _operate = null
+function getOperate () {
+  if (_operate) return _operate
+  // 懒加载：第一次调用原子行为时才 require，此时 browser.js 已加载完
+  _operate = require('./browser').operate
+  return _operate
+}
+let _scroll = null
+function getScroll () {
+  // scroll.js 只依赖 browser, 不依赖 humanize, 安全 require
+  if (_scroll) return _scroll
+  _scroll = require('./scroll')
+  return _scroll
+}
+function operate (pageHandle, params) {
+  return getOperate()(pageHandle, params)
+}
+
+// ─── 风险等级配置（影响延迟倍率、序列长度、触发概率、滚动频率）───
+const RISK_LEVELS = {
+  low: {
+    delayScale: 0.6,
+    sequenceLen: [1, 3],     // 序列长度范围
+    triggerProb: 0.35,       // 每个 operate 触发 humanize 的概率
+    scrollProb: 0.3,         // 序列里含滚动的概率
+    wanderProb: 0.25
+  },
+  medium: {
+    delayScale: 1.0,
+    sequenceLen: [2, 4],
+    triggerProb: 0.55,
+    scrollProb: 0.5,
+    wanderProb: 0.4
+  },
+  high: {
+    delayScale: 1.8,
+    sequenceLen: [3, 6],
+    triggerProb: 0.75,
+    scrollProb: 0.7,
+    wanderProb: 0.6
+  }
+}
+
+// 高风控域名（open/newTab 检测到自动切 high）
+const HIGH_RISK_DOMAINS = [
+  'taobao.com', 'tmall.com', 'jd.com', '1688.com', 'pinduoduo.com',
+  'yangkeduo.com', 'xiaohongshu.com', 'xhslink.com', 'weidian.com',
+  'douyin.com', 'kuaishou.com', 'bilibili.com', 'weibo.com',
+  'alipay.com', 'bank', 'icbc', 'cmbchina', '12306.cn'
+]
+
+// ─── 工具：高斯分布随机数 ───
+function gaussian (mean, std) {
+  // Box-Muller
+  let u = 0; let v = 0
+  while (u === 0) u = Math.random()
+  while (v === 0) v = Math.random()
+  return mean + std * Math.sqrt(-2.0 * Math.log(u)) * Math.cos(2.0 * Math.PI * v)
+}
+
+const sleep = (ms) => new Promise((r) => setTimeout(r, Math.max(0, ms)))
+const randomBetween = (min, max) => min + Math.random() * (max - min)
+const randomInt = (min, max) => Math.floor(randomBetween(min, max + 1))
+const pick = (arr) => arr[Math.floor(Math.random() * arr.length)]
+
+// ─── 贝塞尔曲线点（复用 behavior.js 的逻辑，这里独立实现避免循环依赖）───
+function bezier (p0, p1, p2, p3, t) {
+  const u = 1 - t
+  return u * u * u * p0 + 3 * u * u * t * p1 + 3 * u * t * t * p2 + t * t * t * p3
+}
+
+function bezierTrajectory (from, to, steps) {
+  const pts = []
+  const mx = (from.x + to.x) / 2 + (Math.random() - 0.5) * Math.abs(to.x - from.x) * 0.6
+  const my = (from.y + to.y) / 2 + (Math.random() - 0.5) * Math.abs(to.y - from.y) * 0.6
+  const c1 = { x: from.x + (mx - from.x) * 0.3, y: from.y + (my - from.y) * 0.3 }
+  const c2 = { x: to.x + (mx - to.x) * 0.3, y: to.y + (my - to.y) * 0.3 }
+  for (let i = 0; i <= steps; i++) {
+    const t = i / steps
+    pts.push({
+      x: bezier(from.x, c1.x, c2.x, to.x, t) + (Math.random() - 0.5) * 2,
+      y: bezier(from.y, c1.y, c2.y, to.y, t) + (Math.random() - 0.5) * 2
+    })
+  }
+  return pts
+}
+
+// 模块级：每个 pageHandle 记录上次鼠标位置（贝塞尔轨迹的起点）
+const _lastMousePos = new WeakMap()
+function getLastMouse (pageHandle) {
+  return _lastMousePos.get(pageHandle) || { x: 200 + Math.random() * 400, y: 200 + Math.random() * 300 }
+}
+function setLastMouse (pageHandle, pos) {
+  _lastMousePos.set(pageHandle, pos)
+}
+
+// ─── 原子行为（每个独立、可组合、可重复）───
+
+// 停顿"阅读"
+async function read (pageHandle, scale) {
+  await sleep(gaussian(2500, 1200) * scale)
+}
+
+// 短暂走神
+async function idle (pageHandle, scale) {
+  await sleep(gaussian(1200, 600) * scale)
+}
+
+// 平滑滚动（随机方向、随机距离）
+async function scrollSmooth (pageHandle, scale, direction) {
+  const dir = direction || (Math.random() < 0.8 ? 'down' : 'up')
+  const pixels = randomInt(120, 480)
+  const dy = dir === 'down' ? pixels : -pixels
+  // 调用封装的连续滚动 (scrollBy 内部分步不跳变)
+  await getScroll().scrollBy(pageHandle, dy).catch(() => {})
+  await sleep(gaussian(500, 200) * scale)
+}
+
+// 滚到页面某比例位置 (调封装连续滚动, 不跳变)
+async function scrollToRatio (pageHandle, scale) {
+  const ratio = pick([0.25, 0.4, 0.55, 0.7, 0.85])
+  // 先算出目标 Y, 再用封装的 scrollToY 连续滚
+  const targetY = await operate(pageHandle, { _skipHumanize: true,
+    action: 'evaluate',
+    source: '(r) => Math.max(0, (document.body.scrollHeight - window.innerHeight) * r)',
+    args: ratio
+  }).catch(() => 0)
+  await getScroll().scrollToY(pageHandle, targetY || 0).catch(() => {})
+  await sleep(gaussian(800, 300) * scale)
+}
+
+// 鼠标贝塞尔漫游到随机点（不和任何元素交互）
+// 鼠标贝塞尔漫游: 随机坐标 → elementFromPoint 找到元素 → 高亮 → 移到元素中心
+// 这是"随机选一个元素并围绕它操作"的拟人行为
+async function mouseWander (pageHandle, scale) {
+  const from = getLastMouse(pageHandle)
+  // 1. 随机坐标 → 找到该坐标下的元素 (模拟人随意看某处)
+  const randX = randomBetween(100, 1100)
+  const randY = randomBetween(100, 700)
+  const elInfo = await operate(pageHandle, { _skipHumanize: true,
+    action: 'evaluate',
+    source: '(x, y) => { const el = document.elementFromPoint(x, y); if (!el) return null; const r = el.getBoundingClientRect(); if (window.__beeVR) window.__beeVR.visualMark(el); return { x: Math.round(r.x + r.width/2), y: Math.round(r.y + r.height/2), tag: el.tagName, text: (el.innerText||"").trim().slice(0,20) } }',
+    args: [randX, randY]
+  }).catch(() => null)
+  const target = (elInfo && (elInfo.value || elInfo)) || { x: randX, y: randY }
+  // (高亮已在上面同一 evaluate 内对真实 el 调 visualMark, 不再回 Node 侧传坐标)
+  // 3. 贝塞尔移到元素中心
+  const to = { x: target.x || randX, y: target.y || randY }
+  const steps = randomInt(12, 22)
+  const pts = bezierTrajectory(from, to, steps)
+  for (const pt of pts) {
+    await operate(pageHandle, { _skipHumanize: true, action: 'mouseMove', x: Math.round(pt.x), y: Math.round(pt.y) }).catch(() => {})
+    await sleep(randomBetween(8, 22))
+  }
+  setLastMouse(pageHandle, to)
+  await sleep(gaussian(300, 150) * scale)
+}
+
+// 悬停某个可交互元素（不点击）—— 随机选页面上一个 button/link/input
+async function hoverRandom (pageHandle, scale) {
+  // 用 evaluate 找一个随机可交互元素的中心点
+  const pos = await operate(pageHandle, { _skipHumanize: true,
+    action: 'evaluate',
+    source: '() => { const els = document.querySelectorAll("a, button, input, [role=button]"); const arr = [...els].filter(e => { const r = e.getBoundingClientRect(); return r.width > 0 && r.height > 0 && r.top > 0 && r.top < window.innerHeight - 50 }); if (!arr.length) return null; const el = arr[Math.floor(Math.random()*arr.length)]; const r = el.getBoundingClientRect(); if (window.__beeVR) window.__beeVR.visualMark(el); return { x: Math.round(r.x + r.width/2), y: Math.round(r.y + r.height/2) } }'
+  }).catch(() => null)
+  const target = (pos && (pos.value || pos)) || null
+  if (!target || target.x == null) return
+  // (高亮已在上面同一 evaluate 内对真实 el 调 visualMark)
+  const from = getLastMouse(pageHandle)
+  const steps = randomInt(10, 18)
+  const pts = bezierTrajectory(from, target, steps)
+  for (const pt of pts) {
+    await operate(pageHandle, { _skipHumanize: true, action: 'mouseMove', x: Math.round(pt.x), y: Math.round(pt.y) }).catch(() => {})
+    await sleep(randomBetween(10, 25))
+  }
+  setLastMouse(pageHandle, target)
+  await sleep(gaussian(400, 200) * scale)
+}
+
+// 键盘翻页: 原生 PageDown/Space 不可控 (页面可拦截、距离不定),
+// 改用 ScrollController.scrollBy 模拟 (可控连续滚动)
+async function scrollKey (pageHandle, scale) {
+  const delta = randomInt(300, 600) * (Math.random() < 0.85 ? 1 : -1)
+  await getScroll().scrollBy(pageHandle, delta).catch(() => {})
+  await sleep(gaussian(600, 250) * scale)
+}
+
+// 原地鼠标小抖动
+async function mouseJitter (pageHandle, scale) {
+  const base = getLastMouse(pageHandle)
+  for (let i = 0; i < randomInt(2, 5); i++) {
+    await operate(pageHandle, { _skipHumanize: true,
+      action: 'mouseMove',
+      x: Math.round(base.x + (Math.random() - 0.5) * 20),
+      y: Math.round(base.y + (Math.random() - 0.5) * 20)
+    }).catch(() => {})
+    await sleep(randomBetween(30, 80))
+  }
+  await sleep(gaussian(200, 100) * scale)
+}
+
+// Tab 键跳几下焦点
+async function tabNav (pageHandle, scale) {
+  const count = randomInt(1, 3)
+  for (let i = 0; i < count; i++) {
+    await operate(pageHandle, { _skipHumanize: true, action: 'press', selector: 'body', key: 'Tab' }).catch(() => {})
+    await sleep(gaussian(250, 100) * scale)
+  }
+}
+
+// noop（什么都不做，制造"空洞"）
+async function noop (pageHandle, scale) {
+  await sleep(randomBetween(100, 400))
+}
+
+// ─── 原子行为池（按类别，供随机组合器抽取）───
+const ATOMS = {
+  // 注意力类
+  attention: [
+    { fn: read, weight: 3 },
+    { fn: idle, weight: 1 }
+  ],
+  // 滚动类
+  scroll: [
+    { fn: scrollSmooth, weight: 4 },
+    { fn: scrollToRatio, weight: 2 },
+    { fn: scrollKey, weight: 1 }
+  ],
+  // 鼠标类
+  mouse: [
+    { fn: mouseWander, weight: 3 },
+    { fn: hoverRandom, weight: 2 },
+    { fn: mouseJitter, weight: 1 }
+  ],
+  // 键盘/导航类
+  nav: [
+    { fn: tabNav, weight: 1 }
+  ],
+  // 空洞
+  meta: [
+    { fn: noop, weight: 2 }
+  ]
+}
+
+// 按权重从池里抽一个
+function weightedPick (pool) {
+  const total = pool.reduce((s, a) => s + a.weight, 0)
+  let r = Math.random() * total
+  for (const item of pool) {
+    r -= item.weight
+    if (r <= 0) return item.fn
+  }
+  return pool[0].fn
+}
+
+// ─── 随机序列生成器（按约束规则）───
+// 约束：导航后必有 read+scroll；交互前可选瞄准；偶发分心；总长度有界
+function buildSequence (phase, riskCfg) {
+  const [minLen, maxLen] = riskCfg.sequenceLen
+  const targetLen = randomInt(minLen, maxLen)
+  const seq = []
+
+  if (phase === 'navigate') {
+    // 导航后适应序列：必 read + 多次 scroll + 偶发 mouse
+    seq.push(read)
+    const scrollCount = randomInt(1, 3)
+    for (let i = 0; i < scrollCount; i++) seq.push(weightedPick(ATOMS.scroll))
+    if (Math.random() < riskCfg.wanderProb) seq.push(weightedPick(ATOMS.mouse))
+    // 可能再来一段
+    while (seq.length < targetLen && Math.random() < 0.5) {
+      seq.push(weightedPick([...ATOMS.attention, ...ATOMS.scroll, ...ATOMS.mouse, ...ATOMS.meta]))
+    }
+  } else if (phase === 'interact') {
+    // 交互前瞄准序列：0-2 个（mouse/hover/idle/noop）
+    const aimLen = randomInt(0, Math.min(2, targetLen))
+    for (let i = 0; i < aimLen; i++) {
+      const pool = Math.random() < 0.6 ? ATOMS.mouse : [...ATOMS.attention, ...ATOMS.meta]
+      seq.push(weightedPick(pool))
+    }
+  } else if (phase === 'after-interact') {
+    // 交互后确认序列：0-2 个（read 短 / scroll / noop）
+    const len = randomInt(0, Math.min(2, targetLen))
+    for (let i = 0; i < len; i++) {
+      const pool = [...ATOMS.attention, ...ATOMS.scroll, ...ATOMS.meta]
+      seq.push(weightedPick(pool))
+    }
+  } else {
+    // 默认：全池随机
+    const all = [...ATOMS.attention, ...ATOMS.scroll, ...ATOMS.mouse, ...ATOMS.meta]
+    for (let i = 0; i < targetLen; i++) seq.push(weightedPick(all))
+  }
+
+  // 偶发分心：10% 概率在随机位置插入 idle
+  if (seq.length > 0 && Math.random() < 0.1) {
+    seq.splice(randomInt(0, seq.length), 0, idle)
+  }
+
+  return seq
+}
+
+// ─── 归位：保证目标 action 命中 ───
+async function reposition (pageHandle, selector, scale) {
+  if (!selector) return
+  // 1. 连续滚到目标元素 (调封装的 scrollToElement, 不跳变)
+  await getScroll().scrollToElement(pageHandle, selector, { block: 'center' }).catch(() => {})
+  await sleep(gaussian(500, 200) * scale)
+
+  // 2. 拿目标 bbox，贝塞尔移到中心（带随机偏移）
+  const boxR = await operate(pageHandle, { _skipHumanize: true, action: 'boundingBox', selector }).catch(() => null)
+  const box = boxR && (boxR.value || boxR)
+  if (box && box.width != null) {
+    const cx = box.x + box.width * (0.4 + Math.random() * 0.2)
+    const cy = box.y + box.height * (0.4 + Math.random() * 0.2)
+    const from = getLastMouse(pageHandle)
+    const steps = randomInt(8, 16)
+    const pts = bezierTrajectory(from, { x: cx, y: cy }, steps)
+    for (const pt of pts) {
+      await operate(pageHandle, { _skipHumanize: true, action: 'mouseMove', x: Math.round(pt.x), y: Math.round(pt.y) }).catch(() => {})
+      await sleep(randomBetween(10, 25))
+    }
+    setLastMouse(pageHandle, { x: cx, y: cy })
+  }
+
+  // 3. focus 目标
+  await operate(pageHandle, { _skipHumanize: true, action: 'focus', selector }).catch(() => {})
+}
+
+// ─── 触发判定（随机）───
+function shouldTrigger (riskCfg) {
+  return Math.random() < riskCfg.triggerProb
+}
+
+// ─── 主接口：beforeAction / afterAction ───
+
+/**
+ * 在 operate 执行 action 之前调用（随机触发）。
+ * @param {object} pageHandle
+ * @param {string} action
+ * @param {object} params  operate 的原始 params（用于拿 selector 归位）
+ * @param {object} riskCfg RISK_LEVELS[level]
+ */
+async function beforeAction (pageHandle, action, params, riskCfg) {
+  const scale = riskCfg.delayScale
+  const isNavigate = ['goto', 'goBack', 'goForward', 'reload'].includes(action)
+  const isInteract = ['click', 'fill', 'type', 'press', 'hover', 'dblclick', 'check', 'uncheck', 'selectOption', 'dragTo', 'setInputFiles'].includes(action)
+
+  // 导航类：beforeAction 不注入（导航还没发生），归位交给 afterAction 的适应序列
+  if (isNavigate) return
+
+  // 触发判定
+  if (!shouldTrigger(riskCfg)) {
+    // 即使不触发完整序列，交互类也要保证最小归位（鼠标到目标）
+    if (isInteract && params.selector) {
+      await reposition(pageHandle, params.selector, scale)
+    }
+    return
+  }
+
+  // 注入序列
+  const seq = buildSequence(isInteract ? 'interact' : 'default', riskCfg)
+  for (const atom of seq) {
+    await atom(pageHandle, scale)
+  }
+
+  // 交互类必归位（序列末尾）
+  if (isInteract && params.selector) {
+    await reposition(pageHandle, params.selector, scale)
+  }
+}
+
+/**
+ * 在 operate 执行 action 之后调用（随机触发）。
+ * @param {object} pageHandle
+ * @param {string} action
+ * @param {object} riskCfg
+ */
+async function afterAction (pageHandle, action, riskCfg) {
+  const scale = riskCfg.delayScale
+  const isNavigate = ['goto', 'goBack', 'goForward', 'reload'].includes(action)
+
+  // 导航类：必注入完整适应序列（人到新页面先打量）
+  if (isNavigate) {
+    const seq = buildSequence('navigate', riskCfg)
+    for (const atom of seq) {
+      await atom(pageHandle, scale)
+    }
+    return
+  }
+
+  // 其余：随机触发
+  if (!shouldTrigger(riskCfg)) return
+  const seq = buildSequence('after-interact', riskCfg)
+  for (const atom of seq) {
+    await atom(pageHandle, scale)
+  }
+}
+
+// ─── 工厂：createHumanizer ───
+/**
+ * 创建 humanizer 实例。
+ * @param {object} options
+ * @param {string} [options.riskLevel='medium']  'low'|'medium'|'high'
+ * @returns {{beforeAction, afterAction, riskLevel, riskCfg}}
+ */
+function createHumanizer (options = {}) {
+  const level = options.riskLevel || 'medium'
+  const riskCfg = RISK_LEVELS[level] || RISK_LEVELS.medium
+  // 未知 level 回退到 medium（riskCfg 和 riskLevel 都呑予 medium，不泄漏原值）
+  const safeLevel = RISK_LEVELS[level] ? level : 'medium'
+  return {
+    riskLevel: safeLevel,
+    riskCfg,
+    beforeAction: (ph, action, params) => beforeAction(ph, action, params, riskCfg),
+    afterAction: (ph, action) => afterAction(ph, action, riskCfg)
+  }
+}
+
+// ─── 自由调用的拟人接口 (采集/evaluate 场景) ───
+/**
+ * 随机拟人行为 + 归位到目标。一个调用搞定采集前的拟人。
+ *
+ * 自由调用场景 (不依赖 operate 封装): 调用方明确知道目标 (要点击/采集的元素或位置),
+ * 传给 humanAct, 它随机注入拟人行为 (滚动/鼠标漫游/阅读停顿) 后, 归位到目标。
+ *
+ * @param {pageHandle} pageHandle
+ * @param {object} [opts]
+ * @param {string} [opts.target]     归位目标 (CSS selector), 拟人后滚+移到它
+ * @param {{x:number,y:number}} [opts.targetXY]  归位目标 (坐标), 与 target 二选一
+ * @param {string} [opts.phase='default']  'default'|'navigate'|'interact'|'after'  组合策略
+ *   - default/navigate/interact: 拟人后归位到目标 (动作前用)
+ *   - after: 拟人后随机收尾, 不归位 (动作已执行完, 收尾拟人)
+ * @param {string} [opts.riskLevel='medium'] 风控等级
+ * @param {boolean} [opts.reposition=true]  拟人后归位到目标 (保证后续动作命中)
+ * @returns {Promise<void>}
+ */
+async function humanAct (pageHandle, opts = {}) {
+  const riskCfg = RISK_LEVELS[opts.riskLevel] || RISK_LEVELS.medium
+  const scale = riskCfg.delayScale
+  const phase = opts.phase || 'default'
+  const target = opts.target || null
+  const targetXY = opts.targetXY || null
+
+  // 高亮真实目标 (单源 resolver): 让用户看到拟人行为围绕什么展开
+  // target=selector 走 resolver 解析真实元素再标记; targetXY=坐标直接标记
+  const interact = require('./interact')
+  if (target) {
+    try { await interact.highlightSelector(pageHandle, target, { ms: 0 }) } catch {}
+  } else if (targetXY) {
+    try { await interact.highlight(pageHandle, targetXY, { ms: 0 }) } catch {}
+  }
+
+  // 1. 随机拟人行为序列 (buildSequence 随机组合原子)
+  if (shouldTrigger(riskCfg)) {
+    const seqPhase = phase === 'navigate' ? 'navigate' : (phase === 'after' ? 'after-interact' : 'default')
+    const seq = buildSequence(seqPhase, riskCfg)
+    for (const atom of seq) {
+      await atom(pageHandle, scale)
+    }
+  }
+
+  // 2. 归位到目标 (after 阶段不归位: 动作已执行完, 随机收尾即可)
+  if (opts.reposition !== false && phase !== 'after') {
+    if (target) {
+      await reposition(pageHandle, target, scale)
+    } else if (targetXY) {
+      // 坐标归位: 贝塞尔移到目标坐标
+      const from = getLastMouse(pageHandle)
+      const steps = randomInt(8, 16)
+      const pts = bezierTrajectory(from, targetXY, steps)
+      for (const pt of pts) {
+        await operate(pageHandle, { _skipHumanize: true, action: 'mouseMove', x: Math.round(pt.x), y: Math.round(pt.y) }).catch(() => {})
+        await sleep(randomBetween(10, 25))
+      }
+      setLastMouse(pageHandle, targetXY)
+    }
+  }
+}
+
+// 检测 url 是否命中高风控域名
+function detectRiskLevel (url) {
+  if (!url) return 'medium'
+  const lower = url.toLowerCase()
+  for (const domain of HIGH_RISK_DOMAINS) {
+    if (lower.includes(domain)) return 'high'
+  }
+  return 'medium'
+}
+
+module.exports = {
+  createHumanizer,
+  humanAct,
+  detectRiskLevel,
+  RISK_LEVELS,
+  HIGH_RISK_DOMAINS,
+  // 导出原子行为供测试/自定义组合
+  ATOMS,
+  // 导出工具
+  gaussian,
+  bezierTrajectory
+}