[Bug]: Auto-linking cannot deal with some cases of line-breaks

[Snuffleupagus]

Attach (recommended) or Link to PDF file

https://opensource.adobe.com/dc-acrobat-sdk-docs/pdfstandards/PDF32000_2008.pdf#page=2

Web browser and its version

N/A

Operating system and its version

N/A

PDF.js version

master, or any version after PR #19110

Is the bug present in the latest PDF.js version?

Yes

Is a browser extension

No

Steps to reproduce the problem

1. Load https://opensource.adobe.com/dc-acrobat-sdk-docs/pdfstandards/PDF32000_2008.pdf#page=2
2. Note that an inferred link contains: http://www.adobe.com/devnet/pdf/

What is the expected behavior?

That the "full" link, i.e. http://www.adobe.com/devnet/pdf/pdf_reference.html, should be found.

Note: Acrobat Reader is able to detect the full URL.

What went wrong?

The line-break is causing the URL to be truncated.

Link to a viewer

No response

Additional context

No response

[ryzokuken]

This is because of the structure of that PDF document where the two spans that contain these two parts of the link are separated in the DOM. It could be made to work by a change in the logic where we flatten the content of a page but that would directly conflict with the test for https://github.com/mozilla/pdf.js/blob/master/test/pdfs/bug1019475_2.pdf where we test that the broken link isn't detected.

[nicolo-ribaudo]

We need to be careful about this, since there are many valid cases of URLs ending at the end of a line that might look like it's a multiline URL:

Mozilla: www.mozilla.org/en-US/
Google: www.google.com

Here should not detect www.mozilla.org/en-US/Google as a URL

[Snuffleupagus]

Given the risk of false positives mentioned above, is this still something that we want to try and fix or should we wait and see how common this problem is first?