lib: correct windows-1252 decoding in TextDecoder

[yashwantbezawada]

Description

Fixes#56542

This PR corrects the Windows-1252 decoding in TextDecoder by disabling the Latin-1 fast path for Windows-1252 encoding.

Problem

The TextDecoder was incorrectly using the Latin-1 fast path (via simdutf's convert_latin1_to_utf8) for Windows-1252 encoding. This caused incorrect decoding of bytes in the 0x80-0x9F range.

The root cause is that Windows-1252 differs from ISO-8859-1 (Latin-1) in this byte range:

• ISO-8859-1: Bytes 0x80-0x9F are undefined/control characters that map directly to Unicode (e.g., 0x92 → U+0092)
• Windows-1252: These bytes map to specific printable characters (e.g., 0x92 → U+2019 RIGHT SINGLE QUOTATION MARK ')

Solution

Disable the Latin-1 fast path for Windows-1252 by setting this[kLatin1FastPath] = false. This forces the decoder to use the ICU converter (getConverter()), which correctly handles Windows-1252 character mappings according to the WHATWG Encoding Standard.

Changes

• Modified lib/internal/encoding.js line 423 to disable Latin-1 fast path
• Added comprehensive test coverage for all 32 affected characters (0x80-0x9F)
• Test includes the specific case from issue TextDecoder incorrectly decodes 0x92 and several other characters for Windows-1252 #56542 and realistic text samples

Test Coverage

The new test file test/parallel/test-whatwg-encoding-custom-windows-1252.js verifies:

1. Specific issue case: byte 0x92 correctly decodes to U+2019 (')
2. All 32 characters in the 0x80-0x9F range according to WHATWG spec
3. Common Windows-1252 encoding aliases (windows-1252, cp1252, x-cp1252)
4. Realistic text samples with mixed special characters

References

• Issue: TextDecoder incorrectly decodes 0x92 and several other characters for Windows-1252 #56542
• WHATWG Encoding Standard: https://encoding.spec.whatwg.org/#windows-1252
• Windows-1252 Wikipedia: https://en.wikipedia.org/wiki/Windows-1252#Codepage_layout

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 88.56%. Comparing base (2271d2d) to head (de6c369).
⚠️ Report is 476 commits behind head on main.

Additional details and impacted files

@@ Coverage Diff @@## main #60737 +/- ##
==========================================
+ Coverage 88.53% 88.56% +0.02% 
==========================================
Files 703 703 Lines 208226 208254 +28 Branches 40145 40172 +27 ==========================================
+ Hits 184352 184437 +85 + Misses 15884 15822 -62 - Partials 7990 7995 +5 

Files with missing lines	Coverage Δ
lib/internal/encoding.js	99.50% <100.00%> (-0.01%)	⬇️

... and 60 files with indirect coverage changes

🚀 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.

[ChALkeR]

decodeLatin1 has no usage outside of this and should be just removed, it was added under a wrong assumption
this PR leaves it present but unused

i noticed this PR just now, but I already filed #60889

Tests here are useful though!

[yashwantbezawada]

Makes sense - #60889 is a more complete fix since it removes the C++ code too.

Happy to open a separate PR to add the Windows-1252 test file after #60889 lands, or I can add it directly to #60889 if that's easier. Let me know what works best.

[ChALkeR]

This can be closed, #61093 landed, which removed the broken codepath.
It also added tests for all single-byte encodings.

[yashwantbezawada]

Closing this - looks like #61118 already landed and fixed the issue. Thanks!

[ChALkeR]

upd: the correct link that fixed that was #61093, i linked a follow-up accidentally