mist/crates/parser/src/grammar.pest

// ======================================================
// LEXICAL
// ======================================================

WHITESPACE = _{ " " | "\t" | "\r" | "\n" }
COMMENT    = _{ "//" ~ (!"\n" ~ ANY)* }

// ------------------------------------------------------
// PRIMITIVES
// ------------------------------------------------------

keyword = {
    "if"
  | "else"
  | "for"
  | "while"
  | "match"
  | "return"
  | "break"
  | "continue"
  | "struct"
  | "enum"
  | "class"
  | "trait"
  | "impl"
  | "pub"
  | "mut"
  | "let"
  | "true"
  | "false"
  | "dyn"
  | "loop"
}

identifier = @{
    !(keyword ~ !(ASCII_ALPHANUMERIC | "_")) ~ (ASCII_ALPHA | "_") ~ (ASCII_ALPHANUMERIC | "_")*
}

integer = @{ ASCII_DIGIT+ }

float = @{
    ASCII_DIGIT+ ~ "." ~ ASCII_DIGIT+
}

boolean = {
    "true"
  | "false"
}

string_lit = {
    "\"" ~ inner_str ~ "\""
}

inner_str = @{
    (!"\"" ~ ("\\\"" | ANY))*
}

semicolon = _{ ";" }

static_path = {
    identifier ~ ("::" ~ identifier)*
}

// ======================================================
// PROGRAM
// ======================================================

program = {
    SOI ~ top_level* ~ EOI
}

// ======================================================
// MODIFIERS
// ======================================================

visibility = { "pub" ~ ("(" ~ static_path ~ ")")? }
mutable    = { "mut" }
let_kw     = { "let" }

// ======================================================
// GENERICS
// ======================================================
generic_decl  = { lifetime | (identifier ~ (":" ~ type_expr_param ~ ("+" ~ type_expr_param)*)?) }
generics_decl = { "<" ~ (generic_decl ~ ("," ~ generic_decl)*)? ~ ">" }

generic  = { lifetime | type_expr_param }
generics = { "<" ~ (generic ~ ("," ~ generic)*)? ~ ">" }

// ======================================================
// TYPES
// ======================================================

tuple_type = {
    "[" ~ (type_expr ~ ("," ~ type_expr)*)? ~ "]"
}

path_type = {
    static_path ~ ("<" ~ (type_expr_param ~ ("," ~ type_expr_param)*)? ~ ">")?
}

ref_type = {
    mutable? ~ lifetime? ~ "*"
}

dyn_type = {
    "dyn"
}

type_expr = {
    (path_type | tuple_type) ~ (ref_type | dyn_type)*
}

lifetime = {
    "'" ~ identifier
}

type_expr_param = {
    type_expr
  | lifetime
}

// ======================================================
// LITERALS
// ======================================================

literal = {
    float
  | integer
  | boolean
  | string_lit
}

// ======================================================
// PATTERNS
// ======================================================

multi_pat_template = _{
    (pattern ~ ("," ~ pattern)*)?
}

tuple_pattern = {
    "[" ~ pattern ~ "," ~ multi_pat_template ~ "]"
}

named_tuple_pattern = {
    static_path ~ "[" ~ multi_pat_template ~ "]"
}

struct_pattern = {
    static_path ~ "{" ~ multi_pat_template ~ "}"
}

path_pattern = {
    mutable? ~ static_path
}

pattern = _{
    named_tuple_pattern
  | tuple_pattern
  | struct_pattern
  | literal
  | path_pattern
}

// ======================================================
// ATTRIBUTES / META
// ======================================================

attribute = {
    "#" ~ "[" ~ meta ~ "]"
}

mod_attribute = {
    "#" ~ "!" ~ "[" ~ meta ~ "]"
}

meta = {
    static_path ~ ("=" ~ literal | "(" ~ meta_list? ~ ")")?
}

meta_list = {
    (meta ~ ("," ~ meta)* ~ ","?)?
}

attributes = {
    attribute*
}

// ======================================================
// DECLARATION HELPERS
// ======================================================
var_decl = {
    (let_kw | type_expr) ~ pattern
}

field = {
    visibility? ~ type_expr ~ identifier
}

field_list = {
    (field ~ ("," ~ field)* ~ ","?)?
}

param_list = {
    (var_decl ~ ("," ~ var_decl)* ~ ","?)?
}

// ======================================================
// IMPORTS / MODULES
// ======================================================

import = {
    visibility? ~ "use" ~ "<" ~ static_path ~ ">" ~ semicolon
}

mod_package = {
    visibility? ~ "mod" ~ identifier ~ semicolon
}

// ======================================================
// STRUCTS
// ======================================================

struct_decl = {
    visibility? ~ "struct" ~ identifier ~ generics_decl? ~ "{" ~ field_list? ~ "}"
}

// ======================================================
// ENUMS
// ======================================================

enum_named = {
    identifier
}

enum_tuple = {
    identifier ~ tuple_type
}

enum_struct = {
    identifier ~ "{" ~ field_list ~ "}"
}

enum_field = _{
    enum_tuple
  | enum_struct
  | enum_named
}

enum_fields = _{
    enum_field ~ ("," ~ enum_field)* ~ ","?
}

enum_decl = {
    visibility? ~ "enum" ~ identifier ~ generics_decl? ~ "{" ~ enum_fields ~ "}"
}

// ======================================================
// CLASSES
// ======================================================

class_constructor = {
    visibility? ~ "constructor" ~ generics_decl? ~ "(" ~ param_list ~ ")" ~ block
}

class_fields = {
    class_field*
}

class_field = {
    field ~ ("=" ~ expr)? ~ semicolon
}

self_param = {
    "self" ~ mutable? ~ deref_px?
}

method = {
    visibility? ~ type_expr ~ identifier ~ generics_decl? ~ "(" ~ self_param? ~ ("," ~ param_list)? ~ ")" ~ block
}

class_item = _{
    impl_decl
  | method
}

class_decl = {
    visibility? ~ "class" ~ identifier ~ generics_decl? ~ (":" ~ type_expr)? ~ "{" ~ class_fields ~ class_constructor ~ class_item* ~ "}"
}

// ======================================================
// IMPL
// ======================================================
impl_for_decl = {
    "impl" ~ generics_decl? ~ type_expr ~ "for" ~ type_expr ~ "{" ~ method* ~ "}"
}

impl_decl = {
    "impl" ~ generics_decl? ~ type_expr ~ "{" ~ method* ~ "}"
}

// ======================================================
// Traits
// ======================================================
trait_requirements = { ":" ~ type_expr_param ~ ("+" ~ type_expr_param)* }

method_no_body = {
    visibility? ~ type_expr ~ identifier ~ generics_decl? ~ "(" ~ self_param? ~ ("," ~ param_list)? ~ ")" ~ semicolon
}

trait_item = _{
    method
  | method_no_body
}

trait_decl = {
    visibility? ~ "trait" ~ identifier ~ generics_decl? ~ trait_requirements? ~ "{" ~ trait_item* ~ "}"
}

// ======================================================
// FUNCTIONS
// ======================================================

function_decl = {
    visibility? ~ type_expr ~ identifier ~ "(" ~ param_list ~ ")" ~ block
}

// ======================================================
// TOP LEVEL
// ======================================================

top_level = {
    (attributes ~ (import | impl_for_decl | impl_decl | trait_decl | struct_decl | enum_decl | class_decl | mod_package | function_decl))
  | mod_attribute
}

// ======================================================
// STATEMENTS
// ======================================================

block = {
    "{" ~ statement_list ~ expr? ~ "}"
}

statement_list = {
    statement*
}

statement = _{
    control_flow
  | (expr ~ semicolon)
}

statement_wrapper = { statement }

statement_body = { statement_wrapper | expr }

// ------------------------------------------------------
// BASIC STATEMENTS
// ------------------------------------------------------

basic_stmt = {
    return_stmt
  | break_stmt
  | continue_stmt
  | var_decl_statement
}

return_stmt = {
    "return" ~ expr?
}

break_stmt = {
    "break"
}

continue_stmt = {
    "continue"
}

var_decl_statement = {
    var_decl ~ ("=" ~ expr)?
}

// ------------------------------------------------------
// CONDITIONALS
// ------------------------------------------------------

control_flow = {
    if_stmt
  | while_stmt
  | c_for_stmt
  | for_stmt
  | match_stmt
  | loop_stmt
  | block
}

statement_branch = { "(" ~ expr ~ ")" ~ statement_body }

else_if = {
    "else" ~ "if" ~ statement_branch
}

else_if_list = {
    else_if*
}

if_stmt = {
    "if" ~ statement_branch ~ else_if_list ~ ("else" ~ statement_body)?
}

// ------------------------------------------------------
// LOOPS
// ------------------------------------------------------

while_stmt = {
    "while" ~ statement_branch
}

c_for_stmt = {
    "for" ~ "(" ~ statement ~ statement ~ expr ~ ")" ~ statement_body
}

for_stmt = {
    "for" ~ "(" ~ pattern ~ ":" ~ expr ~ ")" ~ statement_body
}

loop_stmt = {
    "loop" ~ statement_body
}

// ------------------------------------------------------
// MATCH
// ------------------------------------------------------

pattern_list = { pattern ~ ("|" ~ pattern)* }

match_item = {
    pattern_list ~ "=>" ~ statement
}

match_stmt = {
    "match" ~ "(" ~ expr ~ ")" ~ "{" ~ match_item* ~ "}"
}

// ======================================================
// EXPRESSIONS
// ======================================================

// ------------------------------------------------------
// CLOSURES
// ------------------------------------------------------

closure = { type_expr? ~ "|" ~ param_list ~ "|" }

// ------------------------------------------------------
// TUPLES & ARRAYS
// ------------------------------------------------------

tuple = {
    "(" ~ (expr ~ ("," ~ expr)*)? ~ ")"
}

array = {
    "[" ~ expr ~ ";" ~ expr ~ "]" // Repeat notation: [0; 10]

  | "[" ~ (expr ~ ("," ~ expr)* ~ ","?)? ~ "]" // Standard elements initialization
}

// ------------------------------------------------------
// MACROS
// ------------------------------------------------------

macro_inner_v = @{
    string_lit
  | ("(" ~ macro_inner_v* ~ ")")
  | ("[" ~ macro_inner_v* ~ "]")
  | ("{" ~ macro_inner_v* ~ "}")
  | (!("(" | ")" | "[" | "]" | "{" | "}") ~ ANY)
}

// ------------------------------------------------------
// PRIMARY EXPRESSIONS
// ------------------------------------------------------

expr_path_segment = {
    identifier ~ ("::" ~ generics)?
}

expr_path = {
    expr_path_segment ~ ("::" ~ expr_path_segment)*
}

primary = {
    control_flow
  | basic_stmt
  | literal
  | expr_path
  | tuple
  | array
}

// ------------------------------------------------------
// POSTFIX EXPRESSIONS
// ------------------------------------------------------

field_px = {
    "." ~ identifier ~ ("::" ~ generics)?
}

call_px = {
    "(" ~ (expr ~ ("," ~ expr)*)? ~ ")"
}

macro_call_paren   = { "!" ~ "(" ~ macro_inner_v* ~ ")" }
macro_call_bracket = { "!" ~ "[" ~ macro_inner_v* ~ "]" }
macro_call_brace   = { "!" ~ "{" ~ macro_inner_v* ~ "}" }

struct_field_decl = {
    identifier ~ ":" ~ expr
}

struct_px = {
    "{" ~ (struct_field_decl ~ ("," ~ struct_field_decl)*)? ~ ","? ~ "}"
}

index_px = {
    "[" ~ expr ~ "]"
}

as_px = {
    "as" ~ type_expr
}

try_px = {
    "?"
}

increment = {
    "++"
}

decrement = {
    "--"
}

postfix = {
    field_px
  | call_px
  | macro_call_paren
  | macro_call_bracket
  | macro_call_brace
  | struct_px
  | index_px
  | as_px
  | try_px
  | increment
  | decrement
}

// ------------------------------------------------------
// PREFIX EXPRESSIONS
// ------------------------------------------------------

deref_px   = { "*" }
mut_ref_px = { "&" ~ mutable }
ref_px     = { "&" }
not_px     = { "!" }
neg_px     = { "-" }

prefix = {
    closure
  | deref_px
  | mut_ref_px
  | ref_px
  | not_px
  | neg_px
}

// ------------------------------------------------------
// OPERATORS
// ------------------------------------------------------

bin_op = {
    "+="
  | "-="
  | "/="
  | "%="
  | "&="
  | "|="
  | "^="
  | "<<="
  | ">>="
  | "<<"
  | ">>"
  | "..="
  | ".."
  | "<="
  | ">="
  | "=="
  | "!="
  | "&&"
  | "||"
  | "+"
  | "-"
  | "*"
  | "/"
  | "%"
  | "<"
  | ">"
  | "&"
  | "|"
  | "^"
  | "="
}

// ------------------------------------------------------
// EXPRESSIONS
// ------------------------------------------------------

expr = { term ~ (bin_op ~ term)* }

term = { prefix* ~ primary ~ postfix* }