// Complete IDL Grammar for Pest Parser
// Full ROS2 IDL specification implementation matching grammar.lark
WHITESPACE = _{ " " | "\t" | "\r" | "\n" }
COMMENT = _{ line_comment | block_comment }
line_comment = { "//" ~ (!"\n" ~ ANY)* }
block_comment = { "/*" ~ (!"*/" ~ ANY)* ~ "*/" }
// Keywords
KW_MODULE = @{ "module" ~ !(ASCII_ALPHANUMERIC | "_") }
KW_CONST = @{ "const" ~ !(ASCII_ALPHANUMERIC | "_") }
KW_STRUCT = @{ "struct" ~ !(ASCII_ALPHANUMERIC | "_") }
KW_ENUM = @{ "enum" ~ !(ASCII_ALPHANUMERIC | "_") }
KW_UNION = @{ "union" ~ !(ASCII_ALPHANUMERIC | "_") }
KW_SWITCH = @{ "switch" ~ !(ASCII_ALPHANUMERIC | "_") }
KW_CASE = @{ "case" ~ !(ASCII_ALPHANUMERIC | "_") }
KW_DEFAULT = @{ "default" ~ !(ASCII_ALPHANUMERIC | "_") }
KW_SEQUENCE = @{ "sequence" ~ !(ASCII_ALPHANUMERIC | "_") }
KW_STRING = @{ "string" ~ !(ASCII_ALPHANUMERIC | "_") }
KW_WSTRING = @{ "wstring" ~ !(ASCII_ALPHANUMERIC | "_") }
KW_FIXED = @{ "fixed" ~ !(ASCII_ALPHANUMERIC | "_") }
KW_ANY = @{ "any" ~ !(ASCII_ALPHANUMERIC | "_") }
KW_NATIVE = @{ "native" ~ !(ASCII_ALPHANUMERIC | "_") }
KW_VALUEBASE = @{ "ValueBase" ~ !(ASCII_ALPHANUMERIC | "_") }
KW_OBJECT = @{ "Object" ~ !(ASCII_ALPHANUMERIC | "_") }
KW_BOOLEAN = @{ "boolean" ~ !(ASCII_ALPHANUMERIC | "_") }
KW_OCTET = @{ "octet" ~ !(ASCII_ALPHANUMERIC | "_") }
KW_CHAR = @{ "char" ~ !(ASCII_ALPHANUMERIC | "_") }
KW_WCHAR = @{ "wchar" ~ !(ASCII_ALPHANUMERIC | "_") }
KW_FLOAT = @{ "float" ~ !(ASCII_ALPHANUMERIC | "_") }
KW_DOUBLE = @{ "double" ~ !(ASCII_ALPHANUMERIC | "_") }
KW_LONG = @{ "long" ~ !(ASCII_ALPHANUMERIC | "_") }
KW_SHORT = @{ "short" ~ !(ASCII_ALPHANUMERIC | "_") }
KW_UNSIGNED = @{ "unsigned" ~ !(ASCII_ALPHANUMERIC | "_") }
KW_SIGNED = @{ "signed" ~ !(ASCII_ALPHANUMERIC | "_") }
KW_INT8 = @{ "int8" ~ !(ASCII_ALPHANUMERIC | "_") }
KW_INT16 = @{ "int16" ~ !(ASCII_ALPHANUMERIC | "_") }
KW_INT32 = @{ "int32" ~ !(ASCII_ALPHANUMERIC | "_") }
KW_INT64 = @{ "int64" ~ !(ASCII_ALPHANUMERIC | "_") }
KW_UINT8 = @{ "uint8" ~ !(ASCII_ALPHANUMERIC | "_") }
KW_UINT16 = @{ "uint16" ~ !(ASCII_ALPHANUMERIC | "_") }
KW_UINT32 = @{ "uint32" ~ !(ASCII_ALPHANUMERIC | "_") }
KW_UINT64 = @{ "uint64" ~ !(ASCII_ALPHANUMERIC | "_") }
KW_FLOAT32 = @{ "float32" ~ !(ASCII_ALPHANUMERIC | "_") }
KW_FLOAT64 = @{ "float64" ~ !(ASCII_ALPHANUMERIC | "_") }
KW_TRUE = @{ ("true" | "TRUE") ~ !(ASCII_ALPHANUMERIC | "_") }
KW_FALSE = @{ ("false" | "FALSE") ~ !(ASCII_ALPHANUMERIC | "_") }
KW_TYPEDEF = @{ "typedef" ~ !(ASCII_ALPHANUMERIC | "_") }
KW_INTERFACE = @{ "interface" ~ !(ASCII_ALPHANUMERIC | "_") }
KW_EXCEPTION = @{ "exception" ~ !(ASCII_ALPHANUMERIC | "_") }
KW_RAISES = @{ "raises" ~ !(ASCII_ALPHANUMERIC | "_") }
KW_READONLY = @{ "readonly" ~ !(ASCII_ALPHANUMERIC | "_") }
KW_ATTRIBUTE = @{ "attribute" ~ !(ASCII_ALPHANUMERIC | "_") }
// Literals
decimal_literal = @{ (!"0" ~ ASCII_DIGIT ~ ASCII_DIGIT*) | "0" }
hex_literal = @{ "0" ~ ("x" | "X") ~ ASCII_HEX_DIGIT+ }
octal_literal = @{ "0" ~ ASCII_OCT_DIGIT+ }
float_literal = @{
ASCII_DIGIT+ ~ "." ~ ASCII_DIGIT* ~ (("e" | "E") ~ ("+" | "-")? ~ ASCII_DIGIT+)? |
"." ~ ASCII_DIGIT+ ~ (("e" | "E") ~ ("+" | "-")? ~ ASCII_DIGIT+)? |
ASCII_DIGIT+ ~ ("e" | "E") ~ ("+" | "-")? ~ ASCII_DIGIT+
}
string_literal = @{ "\"" ~ (!"\"" ~ (escape_seq | ANY))* ~ "\"" }
string_concat = { string_literal ~ (string_literal)* }
wide_string_literal = @{ "L\"" ~ (!"\"" ~ (escape_seq | ANY))* ~ "\"" }
char_literal = @{ "'" ~ (escape_seq | (!"'" ~ ANY)) ~ "'" }
wide_char_literal = @{ "L'" ~ (escape_seq | (!"'" ~ ANY)) ~ "'" }
escape_seq = @{ "\\" ~ (("x" ~ ASCII_HEX_DIGIT ~ ASCII_HEX_DIGIT) | ("u" ~ ASCII_HEX_DIGIT{4}) | ("U" ~ ASCII_HEX_DIGIT{8}) | ANY) }
boolean_literal = { KW_TRUE | KW_FALSE }
fixed_pt_literal = @{ ASCII_DIGIT+ ~ "." ~ ASCII_DIGIT+ ~ ("d" | "D") }
identifier = @{ ASCII_ALPHA ~ (ASCII_ALPHANUMERIC | "_")* }
// Main entry point
specification = { SOI ~ (include_directive | definition)* ~ EOI }
// Include directive
include_directive = { "#include" ~ (string_literal | angle_bracket_include) }
angle_bracket_include = { "<" ~ (!(">" | "\n") ~ ANY)+ ~ ">" }
// Definitions
definition = {
module |
const_dcl |
type_dcl |
except_dcl |
interface |
value |
type_id_dcl |
type_prefix_dcl |
event |
component |
home
}
// Type declarations
type_dcl = {
typedef_dcl |
struct_type |
union_type |
enum_type |
bitmask_type
}
typedef_dcl = { annotation* ~ KW_TYPEDEF ~ type_declarator ~ ";" }
type_declarator = { type_spec ~ declarators }
declarators = { declarator ~ ("," ~ declarator)* }
// Exception declaration
except_dcl = { annotation* ~ KW_EXCEPTION ~ identifier ~ "{" ~ member* ~ "}" }
// Interface declaration (stub for now)
interface = { annotation* ~ KW_INTERFACE ~ identifier ~ "{" ~ "}" }
value = { annotation* ~ "value" ~ identifier ~ "{" ~ "}" }
type_id_dcl = { "typeid" ~ scoped_name ~ string_literal }
type_prefix_dcl = { "typeprefix" ~ scoped_name ~ string_literal }
event = { annotation* ~ "event" ~ identifier ~ "{" ~ "}" }
component = { annotation* ~ "component" ~ identifier ~ "{" ~ "}" }
home = { annotation* ~ "home" ~ identifier ~ "{" ~ "}" }
// Module declaration
module = { annotation* ~ KW_MODULE ~ identifier ~ "{" ~ definition* ~ "}" ~ ";"? }
// Annotations
annotation = { "@" ~ scoped_name ~ ("(" ~ annotation_appl_params ~ ")")? }
annotation_appl_params = { (annotation_appl_param ~ ("," ~ annotation_appl_param)*) | const_expr }
annotation_appl_param = { identifier ~ "=" ~ const_expr }
// Constant declaration
const_dcl = { annotation* ~ KW_CONST ~ const_type ~ identifier ~ "=" ~ const_expr ~ ";" }
// Type declarations
struct_type = { annotation* ~ KW_STRUCT ~ identifier ~ "{" ~ member* ~ "}" ~ ";" }
member = { annotation* ~ type_spec ~ declarator ~ ";" }
enum_type = { annotation* ~ KW_ENUM ~ identifier ~ "{" ~ enumerator ~ ("," ~ enumerator)* ~ "}" ~ ";" }
enumerator = { annotation* ~ identifier }
bitmask_type = { annotation* ~ "bitmask" ~ identifier ~ "{" ~ bit_value ~ ("," ~ bit_value)* ~ ","? ~ "}" ~ ";" }
bit_value = { annotation* ~ identifier }
union_type = { annotation* ~ KW_UNION ~ identifier ~ KW_SWITCH ~ "(" ~ switch_type_spec ~ ")" ~ "{" ~ case* ~ "}" ~ ";"? }
switch_type_spec = { integer_type | char_type | boolean_type | scoped_name }
case = { case_label+ ~ element_spec }
case_label = { (KW_CASE ~ const_expr ~ ":") | (KW_DEFAULT ~ ":") }
element_spec = { annotation* ~ type_spec ~ declarator ~ ";" }
// Type specifications
type_spec = {
simple_type_spec |
constr_type_spec
}
simple_type_spec = {
base_type_spec |
template_type_spec |
scoped_name
}
base_type_spec = {
floating_pt_type |
integer_type |
char_type |
wide_char_type |
boolean_type |
octet_type |
any_type |
object_type |
value_base_type
}
template_type_spec = {
sequence_type |
string_type |
wide_string_type |
fixed_pt_type
}
constr_type_spec = {
struct_type |
union_type |
enum_type
}
// Basic types
floating_pt_type = {
KW_FLOAT | KW_DOUBLE |
("long" ~ KW_DOUBLE) |
KW_FLOAT32 | KW_FLOAT64
}
integer_type = { signed_int | unsigned_int }
signed_int = {
signed_longlong_int |
signed_long_int |
signed_short_int |
KW_INT8 | KW_INT16 | KW_INT32 | KW_INT64
}
unsigned_int = {
unsigned_longlong_int |
unsigned_long_int |
unsigned_short_int |
KW_UINT8 | KW_UINT16 | KW_UINT32 | KW_UINT64
}
signed_short_int = { KW_SHORT }
signed_long_int = { KW_LONG }
signed_longlong_int = { KW_LONG ~ KW_LONG }
unsigned_short_int = { KW_UNSIGNED ~ KW_SHORT }
unsigned_long_int = { KW_UNSIGNED ~ KW_LONG }
unsigned_longlong_int = { KW_UNSIGNED ~ KW_LONG ~ KW_LONG }
char_type = { KW_CHAR }
wide_char_type = { KW_WCHAR }
boolean_type = { KW_BOOLEAN }
octet_type = { KW_OCTET }
any_type = { KW_ANY }
object_type = { KW_OBJECT }
value_base_type = { KW_VALUEBASE }
// Template types
sequence_type = { KW_SEQUENCE ~ "<" ~ type_spec ~ ("," ~ positive_int_const)? ~ ">" }
string_type = { KW_STRING ~ ("<" ~ positive_int_const ~ ">")? }
wide_string_type = { KW_WSTRING ~ ("<" ~ positive_int_const ~ ">")? }
fixed_pt_type = { KW_FIXED ~ "<" ~ positive_int_const ~ "," ~ positive_int_const ~ ">" }
// Constant types (specific types must come before scoped_name to avoid matching keywords as identifiers)
const_type = {
integer_type |
char_type |
wide_char_type |
boolean_type |
floating_pt_type |
string_type |
wide_string_type |
fixed_pt_type |
octet_type |
scoped_name
}
// Expressions
const_expr = { or_expr }
or_expr = { xor_expr ~ ("|" ~ xor_expr)* }
xor_expr = { and_expr ~ ("^" ~ and_expr)* }
and_expr = { shift_expr ~ ("&" ~ shift_expr)* }
shift_expr = { add_expr ~ (("<<" | ">>") ~ add_expr)* }
add_expr = { mult_expr ~ (("+" | "-") ~ mult_expr)* }
mult_expr = { unary_expr ~ (("*" | "/" | "%") ~ unary_expr)* }
unary_expr = { (unary_operator ~ unary_expr) | primary_expr }
unary_operator = { "+" | "-" | "~" }
// Note: literal must come before scoped_name to handle boolean literals (TRUE/FALSE) correctly
primary_expr = { literal | scoped_name | ("(" ~ const_expr ~ ")") }
// Literals
literal = {
floating_pt_literal |
integer_literal |
character_literal |
wide_character_literal |
string_concat |
wide_string_literal |
boolean_literal |
fixed_pt_literal
}
integer_literal = { hex_literal | octal_literal | decimal_literal }
floating_pt_literal = { float_literal }
character_literal = { char_literal }
wide_character_literal = { wide_char_literal }
// Names and declarators
scoped_name = { identifier ~ ("::" ~ identifier)* }
declarator = { pointer_declarator | direct_declarator }
pointer_declarator = { "*" ~ direct_declarator }
direct_declarator = { identifier ~ ("[" ~ positive_int_const ~ "]")* }
positive_int_const = { const_expr }