Protocol Buffers Edition 2023 Language Specification
The syntax is specified using Extended Backus-Naur Form (EBNF):
| alternation
() grouping
[] option (zero or one time)
{} repetition (any number of times)
Lexical Elements
Letters and Digits
letter = "A" ... "Z" | "a" ... "z"
capitalLetter = "A" ... "Z"
decimalDigit = "0" ... "9"
octalDigit = "0" ... "7"
hexDigit = "0" ... "9" | "A" ... "F" | "a" ... "f"
Identifiers
ident = letter { letter | decimalDigit | "_" }
fullIdent = ident { "." ident }
messageName = ident
enumName = ident
fieldName = ident
oneofName = ident
mapName = ident
serviceName = ident
rpcName = ident
streamName = ident
messageType = [ "." ] { ident "." } messageName
enumType = [ "." ] { ident "." } enumName
groupName = capitalLetter { letter | decimalDigit | "_" }
Integer Literals
intLit = decimalLit | octalLit | hexLit
decimalLit = [-] ( "1" ... "9" ) { decimalDigit }
octalLit = [-] "0" { octalDigit }
hexLit = [-] "0" ( "x" | "X" ) hexDigit { hexDigit }
Floating-point Literals
floatLit = [-] ( decimals "." [ decimals ] [ exponent ] | decimals exponent | "."decimals [ exponent ] ) | "inf" | "nan"
decimals = [-] decimalDigit { decimalDigit }
exponent = ( "e" | "E" ) [ "+" | "-" ] decimals
Boolean
boolLit = "true" | "false"
String Literals
strLit = strLitSingle { strLitSingle }
strLitSingle = ( "'" { charValue } "'" ) | ( '"' { charValue } '"' )
charValue = hexEscape | octEscape | charEscape | unicodeEscape | unicodeLongEscape | /[^\0\n\\]/
hexEscape = '\' ( "x" | "X" ) hexDigit [ hexDigit ]
octEscape = '\' octalDigit [ octalDigit [ octalDigit ] ]
charEscape = '\' ( "a" | "b" | "f" | "n" | "r" | "t" | "v" | '\' | "'" | '"' )
unicodeEscape = '\' "u" hexDigit hexDigit hexDigit hexDigit
unicodeLongEscape = '\' "U" ( "000" hexDigit hexDigit hexDigit hexDigit hexDigit |
"0010" hexDigit hexDigit hexDigit hexDigit
EmptyStatement
emptyStatement = ";"
Constant
constant = fullIdent | ( [ "-" | "+" ] intLit ) | ( [ "-" | "+" ] floatLit ) |
strLit | boolLit | MessageValue
MessageValue is defined in the
Text Format Language Specification.
Edition
The edition statement replaces the legacy syntax keyword, and is used to
define the edition that this file is using.
edition = "edition" "=" [ ( "'" decimalLit "'" ) | ( '"' decimalLit '"' ) ] ";"
Import Statement
The import statement is used to import another .proto’s definitions.
import = "import" [ "weak" | "public" ] strLit ";"
Example:
import public "other.proto";
Package
The package specifier can be used to prevent name clashes between protocol message types.
package = "package" fullIdent ";"
Example:
package foo.bar;
Option
Options can be used in proto files, messages, enums and services. An option can be a protobuf defined option or a custom option. For more information, see Options in the language guide. Options are also be used to control Feature Settings.
option = "option" optionName "=" constant ";"
optionName = ( ident | "(" ["."] fullIdent ")" )
For examples:
option java_package = "com.example.foo";
option features.enum_type = CLOSED;
Fields
Fields are the basic elements of a protocol buffer message. Fields can be normal fields, group fields, oneof fields, or map fields. A field has a label, type and field number.
label = [ "repeated" ]
type = "double" | "float" | "int32" | "int64" | "uint32" | "uint64"
| "sint32" | "sint64" | "fixed32" | "fixed64" | "sfixed32" | "sfixed64"
| "bool" | "string" | "bytes" | messageType | enumType
fieldNumber = intLit;
Normal field
Each field has a label, type, name, and field number. It may have field options.
field = [label] type fieldName "=" fieldNumber [ "[" fieldOptions "]" ] ";"
fieldOptions = fieldOption { "," fieldOption }
fieldOption = optionName "=" constant
Examples:
foo.bar nested_message = 2;
repeated int32 samples = 4 [packed=true];
Oneof and oneof field
A oneof consists of oneof fields and a oneof name. Oneof fields do not have labels.
oneof = "oneof" oneofName "{" { option | oneofField } "}"
oneofField = type fieldName "=" fieldNumber [ "[" fieldOptions "]" ] ";"
Example:
oneof foo {
string name = 4;
SubMessage sub_message = 9;
}
Map field
A map field has a key type, value type, name, and field number. The key type can be any integral or string type. Note, the key type may not be an enum.
mapField = "map" "<" keyType "," type ">" mapName "=" fieldNumber [ "[" fieldOptions "]" ] ";"
keyType = "int32" | "int64" | "uint32" | "uint64" | "sint32" | "sint64" |
"fixed32" | "fixed64" | "sfixed32" | "sfixed64" | "bool" | "string"
Example:
map<string, Project> projects = 3;
Extensions and Reserved
Extensions and reserved are message elements that declare a range of field numbers or field names.
Extensions
Extensions declare that a range of field numbers in a message are available for third-party extensions. Other people can declare new fields for your message type with those numeric tags in their own .proto files without having to edit the original file.
extensions = "extensions" ranges ";"
ranges = range { "," range }
range = intLit [ "to" ( intLit | "max" ) ]
Examples:
extensions 100 to 199;
extensions 4, 20 to max;
Reserved
Reserved declares a range of field numbers or names in a message or enum that can’t be used.
reserved = "reserved" ( ranges | reservedIdent ) ";"
fieldNames = fieldName { "," fieldName }
Examples:
reserved 2, 15, 9 to 11;
reserved foo, bar;
Top Level definitions
Enum definition
The enum definition consists of a name and an enum body. The enum body can have options, enum fields, and reserved statements.
enum = "enum" enumName enumBody
enumBody = "{" { option | enumField | emptyStatement | reserved } "}"
enumField = fieldName "=" [ "-" ] intLit [ "[" enumValueOption { "," enumValueOption } "]" ]";"
enumValueOption = optionName "=" constant
Example:
enum EnumAllowingAlias {
option allow_alias = true;
EAA_UNSPECIFIED = 0;
EAA_STARTED = 1;
EAA_RUNNING = 2 [(custom_option) = "hello world"];
}
Message definition
A message consists of a message name and a message body. The message body can have fields, nested enum definitions, nested message definitions, extend statements, extensions, groups, options, oneofs, map fields, and reserved statements. A message cannot contain two fields with the same name in the same message schema.
message = "message" messageName messageBody
messageBody = "{" { field | enum | message | extend | extensions | group |
option | oneof | mapField | reserved | emptyStatement } "}"
Example:
message Outer {
option (my_option).a = true;
message Inner { // Level 2
required int64 ival = 1;
}
map<int32, string> my_map = 2;
extensions 20 to 30;
}
None of the entities declared inside a message may have conflicting names. All of the following are prohibited:
message MyMessage {
string foo = 1;
message foo {}
}
message MyMessage {
string foo = 1;
oneof foo {
string bar = 2;
}
}
message MyMessage {
string foo = 1;
extend Extendable {
string foo = 2;
}
}
message MyMessage {
string foo = 1;
enum E {
foo = 0;
}
}
Extend
If a message in the same or imported .proto file has reserved a range for extensions, the message can be extended.
extend = "extend" messageType "{" {field | group} "}"
Example:
extend Foo {
int32 bar = 126;
}
Service definition
service = "service" serviceName "{" { option | rpc | emptyStatement } "}"
rpc = "rpc" rpcName "(" [ "stream" ] messageType ")" "returns" "(" [ "stream" ]
messageType ")" (( "{" { option | emptyStatement } "}" ) | ";" )
Example:
service SearchService {
rpc Search (SearchRequest) returns (SearchResponse);
}
Proto file
proto = [syntax] { import | package | option | topLevelDef | emptyStatement }
topLevelDef = message | enum | extend | service
An example .proto file:
edition = "2023";
import public "other.proto";
option java_package = "com.example.foo";
enum EnumAllowingAlias {
option allow_alias = true;
EAA_UNSPECIFIED = 0;
EAA_STARTED = 1;
EAA_RUNNING = 1;
EAA_FINISHED = 2 [(custom_option) = "hello world"];
}
message Outer {
option (my_option).a = true;
message Inner { // Level 2
int64 ival = 1 [features.field_presence = LEGACY_REQUIRED];
}
repeated Inner inner_message = 2;
EnumAllowingAlias enum_field = 3;
map<int32, string> my_map = 4;
extensions 20 to 30;
reserved reserved_field;
}
message Foo {
message GroupMessage {
bool a = 1;
}
GroupMessage groupmessage = [features.message_encoding = DELIMITED];
}