issue with mdbook-i18n-helpers 3.0 in persian string

[moaminsharifi]

PR: #1469

Wired behive in persian texts with <span dir=ltr> ... </span>
as I showed in #1439 how is important to have <span dir=ltr> ... </span> in translation file

with normalizing with version 3

#: src/SUMMARY.md:92
msgid "Double Frees in Modern C++"
msgstr "آزاد سازی مضاعف در<span dir=ltr>C++</span> مدرن"

https://github.com/google/comprehensive-rust/blob/main/po/fa.po#L302

become to:

#: src/SUMMARY.md:92
#, fuzzy
msgid "Double Frees in Modern C++"
msgstr ""
"آزاد سازی مضاعف در\n"
"\n"
"C++\n"
"\n"
"مدرن"

https://github.com/moaminsharifi/comprehensive-rust/blob/normalize-fa.po-file-with-mdbook-i18n-helpers-0.3.0/po/fa.po#L299

after seening this I darf my PRs (#1353, #1469, #1470)

[mgeisler]

Hi @moaminsharifi, thanks for tracking this down!

You're right that there is a problem here... I had not anticipated the messages containing HTML when writing the normalization tool! The basis of the normalization is an assumption that we can run mdbook-xgettext on the translation and get the same result as what we started with.

I just created google/mdbook-i18n-helpers#125 to track fixing this.

As for your case: have you translated "many" comments and strings in the code blocks? If not, then I suggest not using mdbook-i18n-normalize on the fa.po file. Instead, run mdbook-xgettext and msgmerge to refresh the translation. This will also remove all the code blocks from your translation since they're no longer present in the messages.pot file.

However, this is of course not ideal since you'll be prevented from normalizing the file in the future. I don't think we'll make major changes that require normalization going forward, but it's hard to predict the future 🙂