A modern configuration language with a focus on readability and performance.
Glyph's reference Zig implementation takes advantage of multiple things:
- Tokenizer - Streaming tokenizer with SIMD scanning for variable-length tokens.
- Parser - Recursive descent parser with one-token lookahead, and a zero-copy AST.
- Serializer - Deserializes into and serializes from structs using comptime reflection.
As of f3a80d0 the Zig library implementation is fully implemented and working.
Below is a simple PEG grammar for Glyph's syntax, for those implementing it themselves.
File ← Spacing (RuneDecl / Entry)* EOF
Entry ← Key ':' Spacing Value Spacing
Value ← Object / Array / String / Expr / FieldRef / Bool / Nil
Object ← '{' Spacing (RuneDecl / Entry)* '}' Spacing
Array ← '[' Spacing ArrayBody? ']' Spacing
ArrayBody ← Value (',' Spacing Value)* ','? Spacing
Expr ← ExprPrimary (ExprOp ExprPrimary)*
ExprPrimary ← '(' Spacing Expr ')' / ExprFunc / RuneRef / FieldRef / Float / Integer / '-' ExprPrimary
ExprOp ← Spacing ('+' / '-' / '*' / '/') Spacing
ExprFunc ← Ident '(' Spacing Expr (',' Spacing Expr)* ')' Spacing
RuneDecl ← RuneName ':' Spacing Value Spacing
RuneRef ← RuneName Accessor*
RuneName ← '$' [a-zA-Z_] [a-zA-Z0-9_]*
FieldRef ← FieldName Accessor*
FieldName ← '@' [a-zA-Z_] [a-zA-Z0-9_]*
Accessor ← '.' Key / '[' Integer ']'
String ← '"' StringChar* '"'
StringChar ← '\\' [\"nrtbf/] / !'"' .
Float ← '-'? [0-9]+ '.' [0-9]+
Integer ← '-'? [0-9]+
Bool ← 'true' / 'false'
Nil ← 'nil'
Key ← Spacing [a-zA-Z_] [a-zA-Z0-9_]*
Ident ← [a-zA-Z_] [a-zA-Z0-9_]*
Spacing ← (WS / Comment)*
WS ← [ \t\r\n]+
Comment ← '#' (!'\n' .)* '\n'?
EOF ← !.
