giles c919d9a0d7 scheme: Phase 1 parser — R7RS lexical reader + 62 tests [consumes-lex] ...

lib/scheme/parser.sx — reader for R7RS-small lexical syntax:
- numbers (int/float/exp)
- booleans #t / #f / #true / #false
- strings with standard escapes
- symbols (permissive — any non-delimiter)
- characters #\c, #\space, #\newline, #\tab, etc.
- vectors #(...)
- proper lists (dotted-pair deferred to Phase 3 with lambda rest-args)
- reader macros: 'X `X ,X ,@X → (quote X) (quasiquote X) etc.
  (Scheme conventions — lowercase, no $ prefix)
- line comments ;
- nestable block comments #| ... |#
- datum comments #;<datum>

AST shape mirrors Kernel: numbers/booleans/lists pass through;
strings wrapped as {:scm-string ...} to distinguish from symbols
(bare SX strings); chars as {:scm-char ...}; vectors as
{:scm-vector (list ...)}.

62 tests in lib/scheme/tests/parse.sx cover atom kinds, escape
sequences, quote/quasiquote/unquote/unquote-splicing, all three
comment flavours, and classic Scheme idioms (lambda, define, let,
if-cond).

Note: SX cond branches evaluate only the LAST expression, so
multi-mutation branches need explicit (do ...) or (begin ...)
wrappers — caught during block-comment debugging.

chisel: consumes-lex (lex-digit?, lex-whitespace? from
lib/guest/lex.sx); pratt not consumed (no operator precedence
in Scheme).

2026-05-13 19:58:30 +00:00

7.3 KiB

Raw Blame History

View Raw