寻求Java可调用的类浏览器沙箱JS运行时解决方案

Great question—this is a common use case for untrusted script execution, web scraping, or automated testing, and there are solid off-the-shelf solutions plus custom approaches tailored to your Java API requirement. Let’s break this down:

Ready-to-Use Solutions (No Reinventing the Wheel) #

1. Playwright for Java (Highly Recommended) #

Playwright is a modern, robust browser automation tool with full Java bindings, and it checks every box in your requirements out of the box:

• Zero Persistence: Create isolated incognito contexts with disabled caching, cookies, and local storage.
• Strict Timeout Control: Set global or per-script execution limits (e.g., 60 seconds).
• Multi-Point DOM Capture: Pull full page HTML snapshots, extract specific node states, or run custom JS to capture incremental changes.
• Java-First API: Integrate directly into your Java application without extra bridge layers.

Quick Implementation Snippet

import com.microsoft.playwright.*;
import java.util.concurrent.TimeUnit;

public class JsSandboxRunner {
    public static void main(String[] args) {
        try (Playwright playwright = Playwright.create()) {
            // Launch headless Chromium (no visible window) with anti-persistence flags
            Browser browser = playwright.chromium().launch(new BrowserType.LaunchOptions()
                    .setHeadless(true)
                    .setArgs(new String[]{"--disable-cache", "--disable-local-storage"}));
            
            // Create a fully isolated incognito context (no shared state)
            BrowserContext context = browser.newContext(new Browser.NewContextOptions()
                    .setIncognito(true)
                    .setStorageState(null) // Wipe all persisted cookies/storage
                    .setDefaultNavigationTimeout(60000) // 60s global timeout
                    .setDefaultTimeout(60000));
            
            // Spin up a blank page to inject your JS
            Page page = context.newPage();
            
            // Inject and execute your custom JS script
            page.evaluate("""
                // Your JS logic here (supports DOM manipulations and fetch requests)
                document.body.innerHTML = '<div id="status">Script running...</div>';
                // Example async network request
                fetch('https://example.com')
                    .then(res => res.text())
                    .then(data => document.getElementById('status').textContent = 'Fetched data');
            """);
            
            // Capture 6 DOM state samples at 10-second intervals
            for (int i = 0; i < 6; i++) {
                TimeUnit.SECONDS.sleep(10);
                String fullDomSnapshot = page.content(); // Full HTML of the page
                String specificNodeState = page.evaluate("return document.getElementById('status').textContent");
                System.out.println("DOM Sample " + (i+1) + ": " + specificNodeState);
            }
            
            // Cleanup (auto-handled by try-with-resources)
            context.close();
            browser.close();
        } catch (Exception e) {
            e.printStackTrace();
        }
    }
}

2. Selenium with Headless Chrome #

If you’re already familiar with Selenium, this is a reliable alternative. Configure ChromeOptions to disable all persistent storage and enforce timeouts:

• No Persistence: Use --incognito, --disable-cache, and --disable-local-storage flags.
• Timeout Control: Set scriptTimeout and pageLoadTimeout on the WebDriver.
• DOM Capture: Use getPageSource() for full HTML or executeScript() to extract targeted state.

Example Snippet

import org.openqa.selenium.chrome.ChromeDriver;
import org.openqa.selenium.chrome.ChromeOptions;
import java.util.concurrent.TimeUnit;

public class SeleniumJsSandbox {
    public static void main(String[] args) {
        ChromeOptions options = new ChromeOptions();
        options.addArguments("--headless=new", "--incognito", "--disable-cache", "--disable-local-storage");
        ChromeDriver driver = new ChromeDriver(options);
        
        // Enforce 60-second timeouts for scripts and page loads
        driver.manage().timeouts().scriptTimeout(60, TimeUnit.SECONDS);
        driver.manage().timeouts().pageLoadTimeout(60, TimeUnit.SECONDS);
        
        // Navigate to a blank page
        driver.get("about:blank");
        
        // Inject your JS script
        driver.executeScript("""
            document.body.appendChild(document.createElement('p')).textContent = 'Initial state';
            setTimeout(() => document.querySelector('p').textContent = 'Updated state', 15000);
        """);
        
        // Capture multiple DOM samples
        for (int i = 0; i < 6; i++) {
            try { TimeUnit.SECONDS.sleep(10); } catch (InterruptedException e) {}
            String nodeText = (String) driver.executeScript("return document.querySelector('p').textContent");
            System.out.println("Sample " + (i+1) + ": " + nodeText);
        }
        
        driver.quit();
    }
}

Custom Approach (For Full Fine-Grained Control) #

If you need a more lightweight solution (avoid full browser automation overhead), use the Chrome DevTools Protocol (CDP) directly with a Java client like chrome-devtools-java-client. This lets you communicate with a headless Chromium instance at the protocol level:

1. Launch headless Chromium with --remote-debugging-port enabled.
2. Use the CDP client to create a session with disabled storage/cache.
3. Inject JS via the Runtime.evaluate command.
4. Capture DOM state using DOM.getDocument or Page.captureSnapshot at intervals.
5. Use Java’s ScheduledExecutorService to terminate the session after 60 seconds.

Critical Security & Reliability Tips #

• Isolation First: Always use incognito/isolated contexts to prevent cross-session state leakage.
• Resource Limits: Add flags like --max-memory-limit to prevent the sandbox from consuming excessive RAM.
• Error Handling: Wrap script execution and DOM capture in try/catch blocks to handle timeouts or unexpected script behavior.
• Network Control: If needed, Playwright/Selenium let you block specific URLs or mock responses (even though you allowed network requests).

内容的提问来源于stack exchange，提问作者Viktor