esutils

Pure Python character utilities for ECMAScript language tools. Python 3.11+; no runtime dependencies or network calls. This remains a focused partial port of estools/esutils: the code module is provided; upstream ast and keyword APIs are not ported.

python -m pip install esutils

Upgrading from 1.x

Version 2.0 requires Python 3.11 or newer. Choose the unqualified identifier predicates for Unicode 17; the ES5/ES6 names retain explicit Unicode 9 policies. Predicates reject invalid integer scalars with False and reject non-integer inputs, including bool, with TypeError. fromCodePoint returns one Python scalar character and raises ValueError for invalid scalar integers. See the migration notes. The distribution and import name remain esutils; esutils.__version__ is the distribution version string.

Usage

from esutils import code
assert code.UNICODE_VERSION == "17.0.0"
assert code.isIdentifierStart(0x10400)
assert code.isIdentifierStart(0x1C90)
assert code.isIdentifierPart(0x200C)
assert not code.isIdentifierStart(0x200C)
assert not code.isIdentifierPart(0x1F600)
assert code.isWhiteSpace(0xFEFF)
assert not code.isWhiteSpace(0x180E)

Inputs are integer Unicode scalar code points, such as ord(character). Predicates return False for negative integers, values above U+10FFFF and surrogates. Non-integers (including bool) raise TypeError consistently. fromCodePoint returns one Python character for astral inputs; it is not a JavaScript UTF-16 surrogate-pair helper.

The modern policy is explicit, reproducible Unicode 17 acceptance for the ECMAScript 2026 toolchain. ES5/ES6 names preserve the previous package's Unicode 9 policy. Edition names do not imply that a particular Unicode version is required by every implementation of that edition. Character classification does not validate full identifier names, escapes, reserved words or contextual keywords. Python's str.isidentifier() and isspace() use different rules and are not used.

Generated scalar intervals use binary search with an ASCII fast path. Checked-in source files, exact SHA-256 checksums and the offline generator are documented in data provenance. Results are independent of the Python interpreter's Unicode database. Unicode updates require deliberate releases.

See contributor instructions, migration notes and CI results. Maintainers can follow the release instructions.

Performance measurements document the public validation cost and the private ord paths used by the coordinated toolchain. Public API validation and Unicode policies remain unchanged.