Formal Grammar

The complete formal grammar specification for JSON5 implementers and language enthusiasts.

TL;DR: This page contains the formal grammar specification for JSON5 in ABNF notation. It's primarily intended for implementers building JSON5 parsers or tools.

Notation Values Objects Arrays Strings Numbers Lexical

Grammar Notation

This specification uses a modified ABNF (Augmented Backus-Naur Form) notation similar to that used in ECMAScript specifications.

Notation Conventions

Notation	Meaning	Example
`A B`	A followed by B (concatenation)	`{ }` means `{` then `}`
`A \| B`	A or B (alternation)	`true \| false`
`A?`	Zero or one A (optional)	`Sign?` optional sign
`A*`	Zero or more A	`Digit*` any digits
`A+`	One or more A	`Digit+` at least one digit
`"text"`	Literal text	`"null"` the word null
`[a-z]`	Character range	`[0-9]` digits 0-9
`<NAME>`	Unicode category or named character	`<TAB>` tab character

JSON5 Values

A JSON5 document consists of a single JSON5 value optionally surrounded by whitespace and comments.

Value Productions

JSON5Text : JSON5Value JSON5Value : JSON5Null JSON5Boolean JSON5String JSON5Number JSON5Object JSON5Array JSON5Null : "null" JSON5Boolean : "true" "false"

Note: Unlike JSON, which only allows objects or arrays at the top level, JSON5 allows any value type as the root element.

Object Grammar

Objects are unordered collections of key-value pairs enclosed in curly braces.

Object Productions

JSON5Object : "{" "}" "{" JSON5MemberList "}" "{" JSON5MemberList "," "}" JSON5MemberList : JSON5Member JSON5MemberList "," JSON5Member JSON5Member : JSON5MemberName ":" JSON5Value JSON5MemberName : JSON5Identifier JSON5String

Identifier Grammar

JSON5Identifier : IdentifierName IdentifierName : IdentifierStart IdentifierName IdentifierPart IdentifierStart : UnicodeLetter "$" "_" "\" UnicodeEscapeSequence IdentifierPart : IdentifierStart UnicodeCombiningMark UnicodeDigit UnicodeConnectorPunctuation <ZWNJ> <ZWJ> UnicodeLetter : any character in Unicode categories "Lu" "Ll" "Lt" "Lm" "Lo" "Nl" UnicodeDigit : any character in Unicode category "Nd" UnicodeCombiningMark : any character in Unicode categories "Mn" "Mc" UnicodeConnectorPunctuation : any character in Unicode category "Pc"

Array Grammar

Arrays are ordered sequences of values enclosed in square brackets.

Array Productions

JSON5Array : "[" "]" "[" JSON5ElementList "]" "[" JSON5ElementList "," "]" JSON5ElementList : JSON5Value JSON5ElementList "," JSON5Value

String Grammar

Strings can use single or double quotes and support various escape sequences.

String Productions

JSON5String : '"' JSON5DoubleStringCharacter? '"' "'" JSON5SingleStringCharacter? "'" JSON5DoubleStringCharacter : JSON5DoubleStringCharacter JSON5DoubleStringCharacter? JSON5SingleStringCharacter : JSON5SingleStringCharacter JSON5SingleStringCharacter? JSON5DoubleStringCharacter : SourceCharacter but not '"' or "\" or LineTerminator "\" EscapeSequence LineContinuation <LS> <PS> JSON5SingleStringCharacter : SourceCharacter but not "'" or "\" or LineTerminator "\" EscapeSequence LineContinuation <LS> <PS> LineContinuation : "\" LineTerminatorSequence

Escape Sequences

EscapeSequence : CharacterEscapeSequence "0" [lookahead not in DecimalDigit] HexEscapeSequence UnicodeEscapeSequence CharacterEscapeSequence : SingleEscapeCharacter NonEscapeCharacter SingleEscapeCharacter : "'" | '"' | "\" | "b" | "f" | "n" | "r" | "t" | "v" NonEscapeCharacter : SourceCharacter but not EscapeCharacter or LineTerminator EscapeCharacter : SingleEscapeCharacter DecimalDigit "x" "u" HexEscapeSequence : "x" HexDigit HexDigit UnicodeEscapeSequence : "u" HexDigit HexDigit HexDigit HexDigit HexDigit : [0-9a-fA-F]

Number Grammar

Numbers can be decimal or hexadecimal, and include special values like Infinity and NaN.

Number Productions

JSON5Number : JSON5NumericLiteral "Infinity" "NaN" JSON5NumericLiteral : NumericLiteral "+" NumericLiteral "-" NumericLiteral NumericLiteral : DecimalLiteral HexIntegerLiteral DecimalLiteral : DecimalIntegerLiteral "." DecimalDigits? ExponentPart? "." DecimalDigits ExponentPart? DecimalIntegerLiteral ExponentPart? DecimalIntegerLiteral : "0" NonZeroDigit DecimalDigits? DecimalDigits : DecimalDigit DecimalDigits DecimalDigit DecimalDigit : [0-9] NonZeroDigit : [1-9] ExponentPart : ExponentIndicator SignedInteger ExponentIndicator : "e" | "E" SignedInteger : DecimalDigits "+" DecimalDigits "-" DecimalDigits HexIntegerLiteral : "0x" HexDigit+ "0X" HexDigit+

Note: Unlike JSON, JSON5 allows leading decimal points (.5), trailing decimal points (5.), explicit positive signs (+5), and hexadecimal literals (0xFF).

Lexical Grammar

Whitespace, line terminators, and comments that separate tokens.

Whitespace

WhiteSpace : <TAB> U+0009 Tab <VT> U+000B Vertical Tab <FF> U+000C Form Feed <SP> U+0020 Space <NBSP> U+00A0 No-Break Space <BOM> U+FEFF Byte Order Mark <USP> Any Unicode "Zs" category character

Line Terminators

LineTerminator : <LF> U+000A Line Feed <CR> U+000D Carriage Return <LS> U+2028 Line Separator <PS> U+2029 Paragraph Separator LineTerminatorSequence : <LF> <CR> [lookahead not in <LF>] <LS> <PS> <CR> <LF>

Comments

Comment : MultiLineComment SingleLineComment MultiLineComment : "/*" MultiLineCommentChars? "*/" MultiLineCommentChars : MultiLineNotAsteriskChar MultiLineCommentChars? "*" PostAsteriskCommentChars? PostAsteriskCommentChars : MultiLineNotForwardSlashOrAsteriskChar MultiLineCommentChars? "*" PostAsteriskCommentChars? MultiLineNotAsteriskChar : SourceCharacter but not "*" MultiLineNotForwardSlashOrAsteriskChar : SourceCharacter but not "/" or "*" SingleLineComment : "//" SingleLineCommentChars? SingleLineCommentChars : SingleLineCommentChar SingleLineCommentChars? SingleLineCommentChar : SourceCharacter but not LineTerminator

Reserved Words as Keys

Reserved words are valid unquoted object keys in JSON5. The JSON5Identifier production above is IdentifierName with no exclusion, and ECMAScript 5.1 defines IdentifierName to include every reserved word. So { class: 'container' } and { null: 1 } parse, and both the JavaScript and the Python reference implementations accept them.

These are the words ECMAScript 5.1 reserves for its own grammar. JSON5 permits all of them as unquoted keys:

ReservedWord : Keyword FutureReservedWord NullLiteral BooleanLiteral Keyword : "break" "case" "catch" "continue" "debugger" "default" "delete" "do" "else" "finally" "for" "function" "if" "in" "instanceof" "new" "return" "switch" "this" "throw" "try" "typeof" "var" "void" "while" "with" FutureReservedWord : "class" "const" "enum" "export" "extends" "import" "super" FutureReservedWord (strict mode) : "implements" "interface" "let" "package" "private" "protected" "public" "static" "yield" NullLiteral : "null" BooleanLiteral : "true" "false"

Style note: Quoting a reserved word is optional, not required - { class: 'value' }, { "class": "value" } and { 'class': 'value' } all parse to the same object. Quote it when you want the key to read unambiguously to someone skimming the file. What genuinely cannot be unquoted is a key that is not an IdentifierName at all: { 'my-key': 1 }, { '123abc': 1 } and { 'has spaces': 1 } must be quoted.

Building a JSON5 Parser?

Check out the reference implementation or explore language-specific libraries.

Reference Implementation All Libraries