mirror of
https://github.com/morluto/rea
synced 2026-10-26 06:16:17 -04:00
54646e8dc4
Pair a preceding low surrogate only with its adjacent high surrogate. Unpaired code units must not make an earlier letter become the token's neighbor. Keep UTF-16 offsets and source strings unchanged at the shared delimiter owner used by rule discovery, copy checks, and replacement. Cover valid astral letters, digits, emoji, and lone-surrogate controls, plus create/read behavior through a CLI-started environment MCP server. Validation: regression-first run had 4 expected failures; after the fix, 437 selected domain, application, and server-boundary tests passed.