Fix snippet generation failure in some cases with Chinese characters

[edgeinfinity1]

Fixes #0000

Changes proposed in this pull request:
In some uncertain cases, uploaded text files containing Chinese(and possibly other special characters) will fail to generate snippets. I tried to track the procedure and found that the preg_split \R will result in some corrupted characters in $lines.
The reason is not clear to me, but by changing it to /\r\n|\n|\r/ it works well as intended.

Reviewers should focus on:
I'm not sure why this fix could work, since the new expression should be equal with \R.

Screenshot
Before fix:

After fix:

Confirmed

• Frontend changes: tested on a local Flarum installation.
• Backend changes: tests are green (run composer test).

Required changes:

• Related Flarum core extension PR's: (Omit this section if irrelevant)

[Copilot]

Pull Request Overview

This PR fixes snippet generation failures for text files containing Chinese characters and other special characters. The issue was caused by the \R regex pattern in preg_split corrupting characters during line splitting, which prevented proper snippet generation for affected files.

• Replaces the \R regex pattern with explicit line break patterns (/\r\n|\n|\r/) in the text preview formatter
• Resolves character corruption issues that prevented snippet generation for files with Chinese characters

---

_{Tip: Customize your code reviews with copilot-instructions.md. Create the file or learn how to get started.}

[clarkwinkelmann]

Could this be a case where mb_split should be used?

I don't see why a regular split by newline characters could break a multibyte UTF string, unless some Unicode characters use one of the newline codes as their second byte?

I'm not sure where the full pattern definition can be found, but if my duck.ai answer is correct it seems like \R also matches \x0B, \f, \u{2028} and \u{2029}. Could one of those codes specifically be causing issue?

[clarkwinkelmann]

Also, while I think about it, can we get some example text that causes issues in case someone ever writes tests for this?

[edgeinfinity1]

Could this be a case where mb_split should be used?

I tested it just now, and mb_split('\R', ...) works fine. Should I apply this to the PR?

Also, while I think about it, can we get some example text that causes issues in case someone ever writes tests for this?

Here's an example to reproduce the issue:
ftbe.txt

[imorland]

Could this be a case where mb_split should be used?

I tested it just now, and mb_split('\R', ...) works fine. Should I apply this to the PR?

Also, while I think about it, can we get some example text that causes issues in case someone ever writes tests for this?

Here's an example to reproduce the issue: ftbe.txt

Yes, please use mb_split - once that's done this is good to merge. Thanks for your contribution!

[edgeinfinity1]

Changed as requested.