function getCanvasFingerprint() {
    const canvas = document.createElement('canvas');
    const ctx = canvas.getContext('2d');
    ctx.fillStyle = '#f60'; ctx.fillRect(0, 0, 100, 50);
    ctx.fillStyle = '#069'; ctx.font = '16px Arial';
    ctx.fillText('Fingerprint', 5, 30);
    // 添加抗鋸齒、漸變等復(fù)雜渲染
    const gradient = ctx.createLinearGradient(0, 0, 100, 0);
    gradient.addColorStop(0, 'rgba(255,0,0,0.5)');
    ctx.fillStyle = gradient; ctx.fillRect(0, 50, 100, 50);
    return canvas.toDataURL(); // 返回圖像數(shù)據(jù)哈希
}

function getWebGLFingerprint() {
    const gl = document.createElement('canvas').getContext('webgl');
    if (!gl) return null;
    const debugInfo = gl.getExtension('WEBGL_debug_renderer_info');
    const vendor = gl.getParameter(debugInfo.UNMASKED_VENDOR_WEBGL);
    const renderer = gl.getParameter(debugInfo.UNMASKED_RENDERER_WEBGL);
    const maxAnisotropy = gl.getParameter(gl.MAX_TEXTURE_MAX_ANISOTROPY_EXT);
    // 執(zhí)行復(fù)雜著色器，提取精度、性能特征
    return hash(vendor + renderer + maxAnisotropy + ...); 
}

function getAudioFingerprint() {
    const context = new AudioContext();
    const oscillator = context.createOscillator();
    const analyser = context.createAnalyser();
    oscillator.connect(analyser);
    analyser.connect(context.destination);
    oscillator.start();
    // 采集聲卡DAC輸出波形，分析頻率響應(yīng)曲線
    const dataArray = new Float32Array(analyser.frequencyBinCount);
    analyser.getFloatFrequencyData(dataArray);
    oscillator.stop();
    return hash(dataArray.join(','));
}

# 偽代碼：設(shè)備指紋生成與關(guān)聯(lián)
def generate_device_fingerprint(request):
    features = {
        'canvas': get_canvas_hash(request),
        'webgl': get_webgl_info(request),
        'fonts': get_font_list(),  # 枚舉已安裝字體
        'screen': (screen.width, screen.height, colorDepth),
        'timezone': get_timezone_offset(),
        'audio': get_audio_fp(),
        'plugins': get_plugin_list(),
        'hardware_concurrency': navigator.hardwareConcurrency,
        # ... 其他特征
    }
    # 特征向量化與標(biāo)準(zhǔn)化
    feature_vector = normalize_vector(features)
    # 使用局部敏感哈希(LSH)或SimHash生成指紋ID
    device_id = simhash(feature_vector)  
    # 關(guān)聯(lián)歷史訂單與設(shè)備
    redis.sadd(f"device:{device_id}:orders", current_order_id)
    return device_id

// Flink 實(shí)時(shí)計(jì)算訂單頻率 (偽代碼)
DataStream<OrderEvent> orderEvents = ... // 從Kafka讀取訂單流
DataStream<Alert> alerts = orderEvents
    .keyBy(OrderEvent::getDeviceFingerprint) // 按設(shè)備ID分組
    .window(TumblingProcessingTimeWindows.of(Time.minutes(5))) // 5分鐘滾動(dòng)窗口
    .process(new ProcessWindowFunction<OrderEvent, Alert, String, TimeWindow>() {
        @Override
        public void process(String deviceId, Context ctx, Iterable<OrderEvent> orders, Collector<Alert> out) {
            int count = 0;
            for (OrderEvent order : orders) count++;
            if (count > THRESHOLD) { // 閾值判定
                out.collect(new Alert(deviceId, "高頻訂單", count));
            }
        }
    });

from sklearn.ensemble import IsolationForest
# 特征X: [訂單間隔, 鼠標(biāo)移動(dòng)距離, 頁(yè)面跳轉(zhuǎn)深度, IP風(fēng)險(xiǎn)分...]
clf = IsolationForest(n_estimators=100, contamination=0.01)
clf.fit(X_train)  
anomalies = clf.predict(X_test)  # 返回-1表示異常

# 使用LSTM建模用戶(hù)點(diǎn)擊序列
model = Sequential([
    Embedding(input_dim=VOCAB_SIZE, output_dim=64),  # 頁(yè)面ID嵌入
    LSTM(128, return_sequences=True),
    Dropout(0.3),
    LSTM(64),
    Dense(1, activation='sigmoid')  # 異常概率
])
model.compile(loss='binary_crossentropy', optimizer='adam')
# 訓(xùn)練數(shù)據(jù): 正常/惡意用戶(hù)會(huì)話序列

-- 限流Lua腳本：令牌桶算法實(shí)現(xiàn)
local key = KEYS[1]        -- 限流key (e.g., "rate_limit:device:abcd")
local capacity = tonumber(ARGV[1]) -- 桶容量
local rate = tonumber(ARGV[2])    -- 令牌生成速率 (個(gè)/秒)
local now = tonumber(ARGV[3])     -- 當(dāng)前時(shí)間戳
local requested = tonumber(ARGV[4]) -- 請(qǐng)求令牌數(shù)

local last_time = redis.call("hget", key, "last_time")
local tokens = redis.call("hget", key, "tokens")

-- 初始化桶
if not last_time then
    last_time = now
    tokens = capacity
else
    -- 計(jì)算新增令牌
    local delta = math.floor((now - last_time) * rate)
    tokens = math.min(capacity, tokens + delta)
    last_time = now
end

-- 判斷是否允許請(qǐng)求
if tokens >= requested then
    tokens = tokens - requested
    redis.call("hmset", key, "last_time", last_time, "tokens", tokens)
    return 1  -- 允許
else
    redis.call("hmset", key, "last_time", last_time, "tokens", tokens)
    return 0  -- 拒絕
end

// 動(dòng)態(tài)風(fēng)控規(guī)則示例 (Drools語(yǔ)法)
rule "High frequency order from new device"
    when
        $o : OrderEvent()
        $count : Number(intValue > 10) from accumulate(
            OrderEvent(deviceId == $o.getDeviceId(), 
                       this after[0s, 5m] $o),
            count(1)
        )
        Device(deviceId == $o.getDeviceId(), firstSeen > now - 24h) // 新設(shè)備
    then
        insert(new BlockAction($o.getDeviceId(), "HIGH_RISK_DEVICE"));
end

Unsupported markdown: listUnsupported markdown: listUnsupported markdown: list返回設(shè)備ID/風(fēng)險(xiǎn)分返回行為風(fēng)險(xiǎn)分Unsupported markdown: list查詢(xún)限流狀態(tài)是否放行Unsupported markdown: listUnsupported markdown: list客戶(hù)端API網(wǎng)關(guān)設(shè)備指紋服務(wù)行為分析引擎規(guī)則引擎Redis限流器