Repository navigation
Summary
Release 4.11.0 expands interpreted ANTLR4 grammar coverage and improves the AllStar parser on real grammars-v4 inputs.
- Add an XQuery4 example that renders ANTLR4 grammars in the proposed antlr-ng class-like
grammar { lexer { ... } parser { ... } }layout, with fixtures for combined, lexer-only, and parser-only grammars. - Add scripts to find action-free grammars and run interpreted parses against their descriptor-selected tests. Each
<test>now gets its own grammar selection, entry point, input pattern, and generated tables, including unnamed tests and descriptors with multiple grammar families. - Make
trinterpusedesc.xmlto choose the top-level grammar, resolve transitive lexer imports and cyclic parser imports, preserve delegate token declarations, and support an explicit--grammar-nameoverride. Correct descriptor whitespace and top-level entry-point precedence. - Extend interpreted Unicode property sets for grammars including GQL, JSON5, and graphemes. Consume supplementary Unicode characters as scalar values in lexer transitions and literals, and include the third-party Unicode data license in published output.
- Support lexer
-> morein interpreted parsing, and report expected lexer no-match errors without stack traces while retaining tracebacks for internal or unsupported failures. - Fix Kotlin interpreted parsing by retaining literal aliases for command-bearing lexer rules and allowing the parser to match EOF again where the grammar requires it. Keep
--tokensoutput available when parsing fails. - Add regression fixtures and tests for GQL, Elixir, Kotlin, grammar imports, descriptor selection, Unicode properties, lexer commands, and EOF handling.
Validation
dotnet test -c Release tests/AllStarParserTests/AllStarParserTests.csproj --no-restore: 661 passed.- Release builds of
trparseandtrinterpsucceeded. - Compiled the CPP14 and CPP20 descriptor grammar families independently and verified the CPP14 selection emits only its parser and lexer tables.
- Verified named and unnamed descriptor tests are enumerated correctly and the Unicode data license is present in
trinterppublish output.