[noexabwuy]

Problem

The server core and the SA-MP client use single-byte Windows codepages (for example windows-1251), while V8 uses Unicode. omp-node passes strings as raw UTF-8, so all non-ASCII text is broken in both directions:

• JS to server: SafeToString sends UTF-8, and the client shows mojibake ("Привет" = Привет). Affects sendClientMessage, dialogs, text draws, etc.
• server to JS: NewFromUtf8 turns codepage bytes into U+FFFD before any script runs. Affects playerText, playerCommandText, dialog input, Player.GetName, etc.

There is no workaround in JS: V8 can't emit a single byte ≥ 0x80, and U+FFFD can't be recovered.

Solution

New option node.codepage (default utf-8, i.e. no conversion):

"node": { "resources": ["resources"], "codepage": "windows-1251" }

• Values: utf-8, windows-1250 … windows-1258 (case-insensitive, cp1251 / 1251 also accepted). An unknown value logs a warning and leaves the conversion off.
• Characters missing from the codepage (e.g. emoji) become ?.
• console.log is converted too, so the server log stays in one encoding.
• The conversion is table-driven (tables are generated by tools/generate_codepages.js) and uses no platform APIs, so it behaves the same on every OS. Multi-byte codepages are out of scope.
• Only game strings are converted; paths, resource names, config values and API names stay UTF-8.

Testing

• CI builds: Windows x64/x86, Linux x64/x86
• Manual test with "codepage": "windows-1251": Cyrillic dialogs, player input, console.log