Attempt UTF-8 encoding in logger formatter

[GabrielMahan]

While running specs with Capybara and Selenium I've been experiencing a stream of log writing failed. "\xC2" from ASCII-8BIT to UTF-8 errors; not sure of all the context, but adding this encoding to the formatter seems to be a fix--any reason not to add this to the formatter?

• [X ] By placing an X in the preceding checkbox, I verify that I have signed the Contributor License Agreement

---

This change is 

[p0deje]

Oh, this is weird. Is it possible for you to narrow down your test to understand where this message gets broken? I think it would be great to understand what the particular issue is.

I think your changes make sense, but they are impossible in pure Ruby - there is no String#try method. You're using Selenium something that requires ActiveSupport extensions which adds #try method. In order to make this PR mergeable, you need to make it work without ActiveSupport.

[luke-hill]

I've been working in this area (Encoding), a lot recently. Can you try the following code.

def utf8?
  Encoding.default_internal == Encoding::UTF_8 &&
    Encoding.default_external == Encoding::UTF_8
end

# Then in your block
msg = msg.force_encoding('UTF-8') if msg.respond_to?(:force_encoding)

Now to the caveats. ASCII-8BIT is a useful encoding char-set because it allows things like german umlauts and other stuff. I actually have the following code in my new API tests I write

# file - env.rb
# Set these to 8bit to allow umlauts
Encoding.default_external = Encoding::ASCII_8BIT
Encoding.default_internal = Encoding::ASCII_8BIT

What is your current Encoding.default_external / internal values, could it be your system is somehow setting up some weird ones?

[p0deje]

I've been trying to reproduce the issue by sending non-ASCII-8bit characters to it, but they are parsed perfectly fine, so I'm not sure if this issue is still present in the recent version of selenium-webdriver.

Since there was no response for a few months, I'm closing this PR. I'll be happy to continue the conversation and re-open if somebody replies.

[codyrobbins]

This is an old issue but I believe the problem described here is still present.

For me it happens when setting Selenium::WebDriver.logger.level to :debug and the page_source method is called. If the page contains a non-ASCII character, such as ×, the logging fails and results in the message:

log writing failed. "\xC3" from ASCII-8BIT to UTF-8

This happens even when Encoding.default_external and Encoding.default_internal are both set to Encoding::UTF_8.

[codyrobbins]

If you google for this problem there are tons of unrelated projects that have had the same issue (cf. elastic/elasticsearch-ruby#71), so it was difficult for me to localize the issue to Selenium in my case. But using force_encoding looks like it seems to be the fix.

[luke-hill]

Hi @codyrobbins -

I have a small gem that will patch this and remove this problem

https://github.com/site-prism/automation_helpers/blob/main/lib/automation_helpers/patches/selenium_logger.rb

To use it simply do

  require 'automation_helpers/patches'
  AutomationHelpers::Patches::SeleniumLogger.new.patch!

The issue I found was actually further down to the issues in RubyMine and that the issues in Ruby 3.2+ are now localised to just rubymine