Split strings by code points when the separator is empty
Closes #7842
StringVar.split() with no separator, or an empty one, compiled to JS str.split(""), which splits UTF-16 code units, so emoji and other characters outside the Basic Multilingual Plane broke into surrogate halves, unlike the other string operations that already work on code points since 0.11.
- A literal empty separator now compiles to
pyStrChars(...), the helper the other code-point operations use. - A separator that is only known at runtime compiles to
pyStrSplit(s, sep), a newutils/statehelper (sep === "" ? pyStrChars(s) : s.split(sep)), so the separator expression is evaluated once and no inline closure is emitted per call. - Non-empty literal separators still use
str.splitand add no import.
Tests: unit tests on the generated JS for split(), split(""), split(" ") and a state-var separator, and on the attached imports (they fail on main); pyStrSplit is covered in packages/reflex-frontend/tests/state.test.js and listed in public_api.test.js. The Unicode Playwright test now covers split().length(), foreach over split() and a runtime empty separator; I couldn't run Playwright locally, so that part relies on CI. pytest tests/units/test_var.py tests/units/vars tests/units/components/core: 561 passed; ruff check/format clean. A reflex-base bugfix news fragment is included.
All Submissions:
- Have you followed the guidelines stated in CONTRIBUTING.md file?
- Have you checked to ensure there aren't any other open Pull Requests for the desired changed?
Type of change
- Bug fix (non-breaking change which fixes an issue)