如何用Python的Selenium处理Shadow Root并在Adobe快速动作上传文件？

我想在https://quick-actions.express.adobe.com/remove-background实现文件上传，执行以下代码时出现错误：

driver.get("https://quick-actions.express.adobe.com/remove-background")
WebDriverWait(driver, 60).until(EC.presence_of_element_located((By.TAG_NAME, "body")))
shadow_host = driver.find_element(By.TAG_NAME, "qa-standalone-app")
shadow_root = shadow_host.shadow_root
file_input = shadow_root.find_element(By.TAG_NAME, "input")

错误信息：
selenium.common.exceptions.InvalidArgumentException: Message: invalid argument: invalid locator

尝试直接定位文件输入框也失败：

file_input = WebDriverWait(driver, 10).until(
     EC.presence_of_element_located((By.XPATH, '//input[@type="file"]'))
)
file_input.send_keys(file_path)

尝试多层Shadow Root定位时，在第四步失败：

shadow_root1 = driver.find_element(By.CSS_SELECTOR, "qa-standalone-app").shadow_root
shadow_root2 = shadow_root1.find_element(By.CSS_SELECTOR, "standalone-quick-action").shadow_root
shadow_root3 = shadow_root2.find_element(By.CSS_SELECTOR, "qa-app-root").shadow_root
shadow_root4 = shadow_root3.find_element(By.TAG_NAME, "qa-app").shadow_root # 此处失败
shadow_root5 = shadow_root4.find_element(By.CSS_SELECTOR, "qa-remove-background-editor").shadow_root
shadow_root6 = shadow_root5.find_element(By.CSS_SELECTOR, "qa-workspace").shadow_root
shadow_root7 = shadow_root6.find_element(By.CSS_SELECTOR, "qa-file-upload").shadow_root

补充页面元素结构截图：

可行解决方案 #

方法1：用JavaScript穿透多层Shadow DOM定位 #

该网站嵌套多层Shadow DOM，常规Selenium定位容易踩坑，直接通过执行JS遍历所有Shadow Root找到文件输入框：

driver.get("https://quick-actions.express.adobe.com/remove-background")
# 等待页面完全加载
WebDriverWait(driver, 30).until(lambda d: d.execute_script('return document.readyState') == 'complete')

# 执行JS查找file input
file_input = driver.execute_script("""
function findFileInput(root) {
    const input = root.querySelector('input[type="file"]');
    if (input) return input;
    // 遍历所有带shadow root的元素
    const shadowHosts = root.querySelectorAll('*');
    for (const host of shadowHosts) {
        if (host.shadowRoot) {
            const result = findFileInput(host.shadowRoot);
            if (result) return result;
        }
    }
    return null;
}
return findFileInput(document);
""")

# 上传文件
file_input.send_keys("你的本地文件绝对路径")

方法2：修正多层Shadow Root定位逻辑 #

之前定位失败大概率是元素未加载完全或选择器使用不当，给每一步添加等待确保元素就绪：

from selenium.webdriver.support import expected_conditions as EC

driver.get("https://quick-actions.express.adobe.com/remove-background")

# 第一层Shadow Root
shadow_host1 = WebDriverWait(driver, 20).until(
    EC.presence_of_element_located((By.CSS_SELECTOR, "qa-standalone-app"))
)
shadow_root1 = shadow_host1.shadow_root

# 第二层Shadow Root
shadow_host2 = WebDriverWait(shadow_root1, 20).until(
    EC.presence_of_element_located((By.CSS_SELECTOR, "standalone-quick-action"))
)
shadow_root2 = shadow_host2.shadow_root

# 第三层Shadow Root
shadow_host3 = WebDriverWait(shadow_root2, 20).until(
    EC.presence_of_element_located((By.CSS_SELECTOR, "qa-app-root"))
)
shadow_root3 = shadow_host3.shadow_root

# 第四层Shadow Root（改用CSS_SELECTOR更准确）
shadow_host4 = WebDriverWait(shadow_root3, 20).until(
    EC.presence_of_element_located((By.CSS_SELECTOR, "qa-app"))
)
shadow_root4 = shadow_host4.shadow_root

# 第五层Shadow Root
shadow_host5 = WebDriverWait(shadow_root4, 20).until(
    EC.presence_of_element_located((By.CSS_SELECTOR, "qa-remove-background-editor"))
)
shadow_root5 = shadow_host5.shadow_root

# 第六层Shadow Root
shadow_host6 = WebDriverWait(shadow_root5, 20).until(
    EC.presence_of_element_located((By.CSS_SELECTOR, "qa-workspace"))
)
shadow_root6 = shadow_host6.shadow_root

# 第七层Shadow Root
shadow_host7 = WebDriverWait(shadow_root6, 20).until(
    EC.presence_of_element_located((By.CSS_SELECTOR, "qa-file-upload"))
)
shadow_root7 = shadow_host7.shadow_root

# 定位文件输入框并上传
file_input = WebDriverWait(shadow_root7, 20).until(
    EC.presence_of_element_located((By.CSS_SELECTOR, 'input[type="file"]'))
)
file_input.send_keys("你的本地文件绝对路径")

注意点 #

• 确保使用Selenium 4及以上版本，低版本不支持Shadow DOM操作
• 代码中的文件路径必须是本地绝对路径，不能用相对路径
• 若页面加载慢，可适当延长等待时间（比如把20改成30）

内容的提问来源于stack exchange，提问作者LA_