grammar-design.pest

// =====================================================================
// 1. Whitespace & Source
// =====================================================================
COMMENT    = _{ "#" ~ (!NEWLINE ~ ANY)* }
WHITESPACE = _{ " " | "\t" | NEWLINE }
source = { SOI ~ WHITESPACE* ~ nix_expression ~ WHITESPACE* ~ EOI }

// =====================================================================
// 2. Keywords
// =====================================================================
// A single source of truth for all reserved words.
// We can add `if`, `then`, `else`, etc. here as needed.
keywords = _{ "rec" | "let" | "in" | "with" | "inherit" | "true" | "false" | "null" }

// =====================================================================
// 3. Expressions (Hierarchical)
// =====================================================================

// The top-level rule for any expression. Order is important for precedence.
nix_expression = { let_in_expr | with_expr | function_expr | primary_expr } // if_expr would go here too

// -- Complex Expressions --

// let ... in:
let_in_expr = { "let" ~ attr_bindings_block ~ "in" ~ nix_expression }

// with ...; ...
with_expr = { "with" ~ nix_expression ~ ";" ~ nix_expression }

// NOTE: functions are not in scope yet, but show how it would fit
function_expr = { identifier_simple ~ ":" ~ nix_expression }

// -- Primary Expressions --

// Represents a single, non-keyword-led value or a grouped expression.
primary_expr = {
    literal |
    attrset |
    list |
    path_types |
    identifier |
    "(" ~ nix_expression ~ ")" // Grouping expressions
}

// =====================================================================
// 4. Building Blocks & Literals
// =====================================================================
// -- Identifiers --
identifier_simple = @{ !keywords ~ (ASCII_ALPHA | "_") ~ (ASCII_ALPHANUMERIC | "_" | "-")* }
identifier_part   = { identifier_simple | string }
identifier        = { identifier_part ~ ("." ~ identifier_part)* }

// -- Literals --
literal = { string | float | integer | boolean | null }
integer = @{ ("-")? ~ ASCII_DIGIT+ }
float   = @{ ("-")? ~ ASCII_DIGIT+ ~ "." ~ ASCII_DIGIT+ }
boolean = @{ "true" | "false" }
null    = @{ "null" }

// -- Paths --
path_types  = { search_path | path }
search_path = { "<" ~ identifier_simple ~ ">" }
path = @{
    ( "~/" | "../" | "./" | "/" ) ~ ( ( !( WHITESPACE | ";" ) ~ ANY )* ) |
    identifier_simple ~ ( "/" ~ ( !( WHITESPACE | ";" ) ~ ANY )* )+
}

// -- Collections and Bindings --
list    = { "[" ~ nix_expression* ~ "]" }
attrset = { rec? ~ "{" ~ attr_bindings_block? ~ "}" }

// A block of bindings for `attrset` and `let`
attr_bindings_block = { attr_binding+ }
attr_binding      = { binding | inherit_binding }
binding           = { identifier ~ "=" ~ nix_expression ~ ";" }
inherit_binding   = { "inherit" ~ WHITESPACE* ~ ("(" ~ nix_expression ~ ")")? ~ WHITESPACE* ~ identifier_simple+ ~ ";" }

// -- Strings and Interpolations --
string                = { "\"" ~ string_content* ~ "\"" }
interpolation         = { "${" ~ nix_expression ~ "}" }
escaped_quote         = @{ "\\" ~ "\"" }
escaped_interpolation = @{ "''${" }
string_literal_part   = @{ ( !("\"" | "$" | "\\" | "'") ~ ANY )+ }
dollar_literal        = @{ "$" ~ !("{") }
single_quote_literal  = @{ "'" ~ !"'" }
string_content = {
    escaped_quote |
    escaped_interpolation |
    interpolation |
    string_literal_part |
    dollar_literal |
    single_quote_literal
}


// =====================================================================
// 4. Building Blocks & Literals
// =====================================================================

// -- Identifiers --
// Now checks against the `keywords` rule.
identifier_simple = @{ !keywords ~ (ASCII_ALPHA | "_") ~ (ASCII_ALPHANUMERIC | "_" | "-")* }
identifier_part   = { identifier_simple | string }
identifier        = { identifier_part ~ ("." ~ identifier_part)* }

// -- Literals --
literal = { string | float | integer | boolean | null }
integer = @{ ("-")? ~ ASCII_DIGIT+ }
float   = @{ ("-")? ~ ASCII_DIGIT+ ~ "." ~ ASCII_DIGIT+ }
boolean = @{ "true" | "false" }
null    = @{ "null" }

// -- Paths --
path_types  = { search_path | path }
search_path = { "<" ~ identifier_simple ~ ">" }
path = @{
    ( "~/" | "../" | "./" | "/" ) ~ ( ( !( WHITESPACE | ";" ) ~ ANY )* ) |
    identifier_simple ~ ( "/" ~ ( !( WHITESPACE | ";" ) ~ ANY )* )+
}

// -- Collections and Bindings --
list    = { "[" ~ nix_expression* ~ "]" }
attrset = { rec? ~ "{" ~ attr_bindings_block? ~ "}" }

// A block of bindings for `attrset` and `let`
attr_bindings_block = { attr_binding+ }
attr_binding      = { binding | inherit_binding }
binding           = { identifier ~ "=" ~ nix_expression ~ ";" }
inherit_binding   = { "inherit" ~ WHITESPACE* ~ ("(" ~ nix_expression ~ ")")? ~ WHITESPACE* ~ identifier_simple+ ~ ";" }