v1.6.0

What's Changed

Bug Fixes

• Correct hash semantics for Item, State, and Token (#38)

  • Item now hashes by value (rule + position tuple), so independently constructed equal items share a hash and satisfy ==.
  • State hash is derived from its item-set, making equal states interchangeable in dictionaries.
  • Token hash incorporates the concrete subclass, preventing different token types at the same position from colliding.

• Fixed-point FIRST / FOLLOW computation (#36)

  • Replaced the recursive re-entry-guard implementation of calc_first / calc_follow with a worklist fixed-point algorithm.
  • Grammars with mutual recursion now produce correct FIRST and FOLLOW sets; the previous implementation could return incomplete results.
  • Rule.calc_first and Rule.calc_follow are retained as thin wrappers for backward compatibility.

Improvements

• Deterministic LR(0) state construction (#37)
  • The parser now builds its LR(0) automaton in a stable, input-order-independent way.
  • Repeated Parser constructions on the same grammar always produce identical parse tables, which aids caching, reproducibility, and debugging.

Documentation

• English docstrings and algorithm comments (#34)
  • All public classes and methods now carry English docstrings (Lexer, Parser, Token, PlareException, LexingError, ParserError, ParsingError).
  • Inline comments explain the SLR(1) construction phases (FIRST, FOLLOW, LR(0) automaton, ACTION/GOTO table, conflict resolution).

Tests

• Parser safety-net suite (#35) — covers a wide range of grammars including operator precedence, left/right recursion, ε-rules, and expected SLR(1) conflicts (xfail).
• FIRST / FOLLOW unit tests (#36) — 5 tests covering mutual recursion termination, ε-propagation, nullable chains, and the Dragon Book expression grammar.
• Determinism tests (#37) — verifies that repeated parser construction yields identical state IDs and table entries; marked slow for optional skipping.
• Hash semantics tests (#38) — 5 tests pinning the corrected hash/equality contracts for Item, State, and Token.

Full Changelog: v1.5.0...v1.6.0

v1.5.0

Officially support Python 3.13.

v1.4.1

• Fix bug in printing matched string with escapes

v1.4.0

• Hide EOF token generation from lexer

v1.3.0

• Support multi token generation from single match in lexer

v1.2.0

• Handle empty rule correctly

v1.1.0

• Mark Plare as typed

v1.0.0

• Release of v1.0.0
• Lexer and SLR(1) Parser implementation with Python