Unexpected translations on plugin changelog pages

[SergeyBiryukov]

Reported by @cybr on ​Polyglots:

I noticed that my changelog has automated translations based on earlier translations.

I’m OK with this, however, as you can see on ​https://nl.wordpress.org/plugins/autodescription/changelog/, the class names I describe between backticks are also translated. This might cause fatal errors or just plain confusion :(

My request is the following:
Could it be so that code doesn’t get auto-translated, or more specific, inline code :)

Generally, only the whole strings should be translated, but it looks like on plugin changelog pages translations are currently applied for standalone words too, which leads to unexpected results:

• AutoBeschijving_Generate::the_url() should be AutoDescription_Generate::the_url().
• "The Canonieke URL within Posts and Paginas are now correct again" is neither Dutch nor English.

The strings appear to come from the ​readme translation:

• Page → Pagina
• Canonical → Canonieke
• Description → Beschijving

[ivanrf]

Similar issue: ​https://make.wordpress.org/polyglots/2016/01/08/i-found-a-bug-while-translating-the-front/

[Cybr]

As promised, the regex to match all PHP and JavaScript comments, which are therefore safe to translate:

$regex = "/(\#|(\/)?(\*|\/)[^\r\n])(.+?(?=\*\/)|.*)(\*\/)?/igm";

// Explained:
$regex = "
/(              // Group 1
\#             // 1: Grab #
|(\/)?(\*|\/)  // 1: Or, if a slash is found: /* or //
[^\r\n]        // 1: Grab everything including line breaks
)              // End group 1
(              // Group 2
.+?(?=\*\/)    // 2: Grab everything, until */ on same line
|.*            // 2: Or continue to new line
)              // End group 2
(\*\/)?        // Group 3: Find final */ if new line
/igm";

Edit: Fair caution, this script also grabs hyperlinks after the double slash.

[ocean90]

Fixed in [dotorg11414].