这是本节的多页打印视图。 点击此处打印.

返回本页常规视图.

Troubleshooting Assistance

How to get manage WebDriver problems.

It is not always obvious the root cause of errors in Selenium.

  1. The most common Selenium-related error is a result of poor synchronization. Read about Waiting Strategies. If you aren’t sure if it is a synchronization strategy you can try temporarily hard coding a large sleep where you see the issue, and you’ll know if adding an explicit wait can help.

  2. Note that many errors that get reported to the project are actually caused by issues in the underlying drivers that Selenium sends the commands to. You can rule out a driver problem by executing the command in multiple browsers.

  3. If you have questions about how to do things, check out the Support options for ways get assistance.

  4. If you think you’ve found a problem with Selenium code, go ahead and file a Bug Report on GitHub.

1 - 理解常见的异常

如何处理Selenium代码中的各种问题.

无效选择器的异常 (Invalid Selector Exception)

某些时候难以获得正确的CSS以及XPath选择器。

潜在原因

您尝试使用的CSS或XPath选择器包含无效字符或无效查询。

可行方案

通过验证器服务运行选择器：

或者使用浏览器扩展程序来获取已知的良好值：

没有这样元素的异常 (No Such Element Exception)

在您尝试找到该元素的当前时刻无法定位元素。

潜在原因

  • 您在错误的位置寻找元素 (也许以前的操作不成功)
  • 您在错误的时间寻找元素 (该元素尚未显示在 DOM 中)
  • 自您编写代码以来定位器已变更

可行方案

  • 确保您位于期望的页面上，并且代码中的前置操作已正确完成
  • 确保您使用的是正确的 等待策略
  • 使用浏览器的devtools控制台更新定位器或使用浏览器扩展程序，例如:

过时元素引用的异常 (Stale Element Reference Exception)

当成功定位到元素时， WebDriver会为其设置一个引用ID作为标记， 如果由于上下文环境发生变化， 导致之前元素的位置发生了变化或者无法找到了， WebDriver并不会自动重新定位， 任何使用之前元素所做的操作将报错该异常。

常见因素

以下情况可能发生此异常:

  • 您已刷新页面，或者页面的 DOM 已动态更改。
  • 您已导航到其他页面。
  • 您已切换到另一个窗口，或者进入/移出某个 frame / iframe。

常见方案

DOM已变更

当页面刷新或页面上的项目各处移动时， 页面上仍然有一个具有所需定位器的元素， 它只是不再被正在使用的元素对象访问， 并且必须重新定位该元素才能再次使用。

这往往通过以下两种方式之一完成：

  • 每次使用时都要重新定位元素。 尽管有可能元素在定位和使用元素之间的微秒内， 发生变化的可能性很小。 缺点是这不是最有效的方法， 尤其是在 Remote Grid上运行时。

  • 用另一个存储定位器的对象包装 Web 元素，并缓存定位的 Selenium 元素。 对该包装对象执行操作时，您可以尝试使用之前找到的缓存对象， 如果它是发生了变化，则可以捕获异常， 使用存储的定位器重新定位元素，并重试该方法。 这样效率更高，但如果您使用的定位器在页面更改后引用了不同的元素（而不是您想要的元素），则可能会导致问题。

上下文已变更

元素对象是针对特定的上下文存储的， 因此如果您切换到不同的上下文， 比如不同的 Window 或不同的 frame 或 iframe 元素引用仍然有效， 但暂时无法访问。在这种情况下， 重新定位元素无济于事，因为它在当前上下文中不存在。

要解决此问题，您需要确保在使用该元素之前切换回正确的上下文。

页面已变更

这种情况发生在您不仅更改了上下文， 而且导航到另一个页面并破坏了元素所在的上下文。 您无法仅从当前上下文重新定位它， 也无法切换回元素有效的活动上下文。 如果这是您的错误原因， 您必须回到正确的位置并重新定位元素。

2 - Logging Selenium commands

Getting information about Selenium execution.

Turning on logging is a valuable way to get extra information that might help you determine why you might be having a problem.

Getting a logger

Java logs are typically created per class. You can work with the default logger to work with all loggers. To filter out specific classes, see Filtering

Get the root logger:

    Logger logger = Logger.getLogger("");
Check code on GitHub

Java Logging is not exactly straightforward, and if you are just looking for an easy way to look at the important Selenium logs, take a look at the Selenium Logger project

Python logs are typically created per module. You can match all submodules by referencing the top level module. So to work with all loggers in selenium module, you can do this:

    logger = logging.getLogger('selenium')
Check code on GitHub 
.NET does not currently have a Logging implementation

If you want to see as much debugging as possible in all the classes, you can turn on debugging globally in Ruby by setting $DEBUG = true.

For more fine-tuned control, Ruby Selenium created its own Logger class to wrap the default Logger class. This implementation provides some interesting additional features. Obtain the logger directly from the #loggerclass method on the Selenium::WebDriver module:

      logger = Selenium::WebDriver.logger
Check code on GitHub 
const logging = require('selenium-webdriver/lib/logging')
logger = logging.getLogger('webdriver')

Logger level

Logger level helps to filter out logs based on their severity.

Java has 7 logger levels: SEVERE, WARNING, INFO, CONFIG, FINE, FINER, and FINEST. The default is INFO.

You have to change both the level of the logger and the level of the handlers on the root logger:

    logger.setLevel(Level.FINE);
    Arrays.stream(logger.getHandlers()).forEach(handler -> {
      handler.setLevel(Level.FINE);
    });
Check code on GitHub

Python has 6 logger levels: CRITICAL, ERROR, WARNING, INFO, DEBUG, and NOTSET. The default is WARNING

To change the level of the logger:

    logger.setLevel(logging.DEBUG)
Check code on GitHub

Things get complicated when you use PyTest, though. By default, PyTest hides logging unless the test fails. You need to set 3 things to get PyTest to display logs on passing tests.

To always output logs with PyTest you need to run with additional arguments. First, -s to prevent PyTest from capturing the console. Second, -p no:logging, which allows you to override the default PyTest logging settings so logs can be displayed regardless of errors.

So you need to set these flags in your IDE, or run PyTest on command line like:

pytest -s -p no:logging

Finally, since you turned off logging in the arguments above, you now need to add configuration to turn it back on:

logging.basicConfig(level=logging.WARN)

.NET does not currently have a Logging implementation

Ruby logger has 5 logger levels: :debug, :info, :warn, :error, :fatal. As of Selenium v4.9.1, The default is :info.

To change the level of the logger:

      logger.level = :debug
Check code on GitHub

JavaScript has 9 logger levels: OFF, SEVERE, WARNING, INFO, DEBUG, FINE, FINER, FINEST, ALL. The default is OFF.

To change the level of the logger:

logger.setLevel(logging.Level.INFO)

Actionable items

Things are logged as warnings if they are something the user needs to take action on. This is often used for deprecations. For various reasons, Selenium project does not follow standard Semantic Versioning practices. Our policy is to mark things as deprecated for 3 releases and then remove them, so deprecations may be logged as warnings.

Java logs actionable content at logger level WARN

Example:

May 08, 2023 9:23:38 PM dev.selenium.troubleshooting.LoggingTest logging
WARNING: this is a warning

Python logs actionable content at logger level — WARNING Details about deprecations are logged at this level.

Example:

WARNING  selenium:test_logging.py:23 this is a warning

.NET does not currently have a Logging implementation

Ruby logs actionable content at logger level — :warn. Details about deprecations are logged at this level.

For example:

2023-05-08 20:53:13 WARN Selenium [:example_id] this is a warning

Because these items can get annoying, we’ve provided an easy way to turn them off, see filtering section below.

Useful information

This is the default level where Selenium logs things that users should be aware of but do not need to take actions on. This might reference a new method or direct users to more information about something

Java logs useful information at logger level INFO

Example:

May 08, 2023 9:23:38 PM dev.selenium.troubleshooting.LoggingTest logging
INFO: this is useful information

Python logs useful information at logger level — INFO

Example:

INFO     selenium:test_logging.py:22 this is useful information

.NET does not currently have a Logging implementation

Ruby logs useful information at logger level — :info.

Example:

2023-05-08 20:53:13 INFO Selenium [:example_id] this is useful information

Logs useful information at level: INFO

Debugging Details

The debug log level is used for information that may be needed for diagnosing issues and troubleshooting problems.

Java logs most debug content at logger level FINE

Example:

May 08, 2023 9:23:38 PM dev.selenium.troubleshooting.LoggingTest logging
FINE: this is detailed debug information

Python logs debugging details at logger level — DEBUG

Example:

DEBUG    selenium:test_logging.py:24 this is detailed debug information

.NET does not currently have a Logging implementation

Ruby only provides one level for debugging, so all details are at logger level — :debug.

Example:

2023-05-08 20:53:13 DEBUG Selenium [:example_id] this is detailed debug information

Logs debugging details at level: FINER and FINEST

Logger output

Logs can be displayed in the console or stored in a file. Different languages have different defaults.

By default all logs are sent to System.err. To direct output to a file, you need to add a handler:

    Handler handler = new FileHandler("selenium.xml");
    logger.addHandler(handler);
Check code on GitHub

By default all logs are sent to sys.stderr. To direct output somewhere else, you need to add a handler with either a StreamHandler or a FileHandler:

    log_path = "selenium.log"
    handler = logging.FileHandler(log_path)
    logger.addHandler(handler)
Check code on GitHub 
.NET does not currently have a Logging implementation

By default, logs are sent to the console in stdout.
To store the logs in a file:

      logger.output = file_name
Check code on GitHub

JavaScript does not currently support sending output to a file.

To send logs to console output:

logging.installConsoleHandler()

Logger filtering

Java logging is managed on a per class level, so instead of using the root logger (Logger.getLogger("")), set the level you want to use on a per-class basis:

    Logger.getLogger(RemoteWebDriver.class.getName()).setLevel(Level.FINEST);
    Logger.getLogger(SeleniumManager.class.getName()).setLevel(Level.SEVERE);
Check code on GitHub
Because logging is managed by module, instead of working with just "selenium", you can specify different levels for different modules: 
    logging.getLogger('selenium.webdriver.remote').setLevel(logging.WARN)
    logging.getLogger('selenium.webdriver.common').setLevel(logging.DEBUG)
Check code on GitHub 
.NET does not currently have a Logging implementation

Ruby’s logger allows you to opt in (“allow”) or opt out (“ignore”) of log messages based on their IDs. Everything that Selenium logs includes an ID. You can also turn on or off all deprecation notices by using :deprecations.

These methods accept one or more symbols or an array of symbols:

      logger.ignore(:jwp_caps, :logger_info)
Check code on GitHub

or

      logger.allow(%i[selenium_manager example_id])
Check code on GitHub