An MWE to show the limitations of headless chrome when downloading a PDF from a new tab link. The options and preferences used to configure chrome were taken from this chromium thread. These comments are old, but it's the only one to seem to have gotten things working with capybara.

Install the required gems and drivers.

I'm a python person, so I use:

• python3 -m http.server

Visit http://localhost:8000 and try out both links. Both should point to our sample PDF, one in a new tab and one not.

Run the ruby code that uses headless chrome / selenium to download this:

• ruby download_pdf.rb

By default this will attempt to download the new tab PDF. It won't fail, but there won't be any files listed in the DOWNLOAD_PATH at the end of the script.

Edit download_pdf.rb to find id: "normal_link" on line 9, then run ruby download_pdf.rb again. This time the DOWNLOAD_PATH should have a PDF file in it, which you can go in and open.

Restore download_pdf.rb to its original state, so it's looking for id: "newtab_link". Then open scrapppy.rb and comment out line 18, where we're specifying that chrome run in --headless mode. Save and then run ruby download_pdf.rb again. Chrome will start in headed mode, then successfully download the new tab PDF.

• capybara (2.18.0)
• selenium-webdriver (3.141.0)
• ChromeDriver 2.44.609545