fix: Gettext formatter merges duplicate plural entries

[garikkh]

Description

Fixes #2346

When using po-gettext formatter, plural calls with identical strings but different variables:

<div>
  {plural(count, {
    one: "one book",
    other: "many books",
  })}
</div>
<div>
  {plural(anotherCount, {
    one: "one book",
    other: "many books",
  })}
</div>

generated duplicate msgid entries in PO files, which violates PO format rules.

#. js-lingui:icu=%7BanotherCount%2C+plural%2C+one+%7Bone+book%7D+other+%7Bmany+books%7D%7D&pluralize_on=anotherCount
#: src/App.tsx:68
msgid "one book"
msgid_plural "many books"
msgstr[0] "one book"
msgstr[1] "many books"

#. js-lingui:icu=%7Bcount%2C+plural%2C+one+%7Bone+book%7D+other+%7Bmany+books%7D%7D&pluralize_on=count
#: src/App.tsx:61
msgid "one book"
msgid_plural "many books"
msgstr[0] "one book"
msgstr[1] "many books"

Added pre-processing step that merges duplicate plural catalog entries before serialization.

1. Groups catalog entries by their plural case content (ignoring variable names)
2. Merges entries with identical plural strings into a single catalog entry
3. Combines all source location references from merged entries
4. Prevents duplicate msgid/msgid_plural entries in generated PO files

With this change the PO now looks like:

#. js-lingui:icu=%7BanotherCount%2C+plural%2C+one+%7Bone+book%7D+other+%7Bmany+books%7D%7D&pluralize_on=anotherCount
#: src/App.tsx:61
#: src/App.tsx:67
msgid "one book"
msgid_plural "many books"
msgstr[0] "one book"
msgstr[1] "many books"

Types of changes

• Bugfix (non-breaking change which fixes an issue)
• New feature (non-breaking change which adds functionality)
• Breaking change (fix or feature that would cause existing functionality to not work as expected)
• Documentation update
• Examples update

Fixes #2346

Checklist

• I have read the CONTRIBUTING and CODE_OF_CONDUCT docs
• I have added tests that prove my fix is effective or that my feature works
• I have added the necessary documentation (if appropriate) No relevant docs

[vercel]

@garikkh is attempting to deploy a commit to the Crowdin Team on Vercel.

A member of the Team first needs to authorize it.

[codecov]

Codecov Report

❌ Patch coverage is 91.76471% with 7 lines in your changes missing coverage. Please review.
✅ Project coverage is 76.72%. Comparing base (6bb8983) to head (d46d3a1).
⚠️ Report is 228 commits behind head on main.

Files with missing lines	Patch %	Lines
packages/format-po-gettext/src/po-gettext.ts	91.76%	4 Missing and 3 partials ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #2347      +/-   ##
==========================================
- Coverage   77.05%   76.72%   -0.33%     
==========================================
  Files          84       99      +15     
  Lines        2157     2733     +576     
  Branches      555      707     +152     
==========================================
+ Hits         1662     2097     +435     
- Misses        382      509     +127     
- Partials      113      127      +14     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.
• 📦 JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.

[garikkh]

Ah, didn't work properly, need to continue working on it

[timofei-iatsenko]

I think implementation is suboptimal, instead of introducing another type of comment, simply print the same comments many times

#. js-lingui:icu=%7BanotherCount%2C+plural%2C+one+%7Bone+book%7D+other+%7Bmany+books%7D%7D&pluralize_on=anotherCount
#. js-lingui:icu=%7Bcount%2C+plural%2C+one+%7Bone+book%7D+other+%7Bmany+books%7D%7D&pluralize_on=count
#: src/App.tsx:61
msgid "one book"
msgid_plural "many books"
msgstr[0] "one book"
msgstr[1] "many books"

When deserializing just take to consideration that there could be multiple comments, flat map 1 message + 1 comment and use existing methods to process that.

[garikkh]

I considered that approach, but it seemed messier to me when you have many duplicates. In our codebase, we've only been using lingui a few months, but already have cases with ~7 duplicates. So now you'd have a PO entry with 7 ICU comments instead of 1 really long comment.

[garikkh]

instead of introducing another type of comment

Also wanted to note, this isn't introducing another comment - it's just using the icu comment. the ICU comment is a URLSearchParams, so all I did is add another key here.

  const contextComment = item.extractedComments
    .find((comment) => comment.startsWith(ctxPrefix))
    ?.substring(ctxPrefix.length)
  const ctx = new URLSearchParams(contextComment)

This solution is a little more complex, but IMO is more elegant/consistent

[garikkh]

bump @timofei-iatsenko