Lesson 22 · Rejecting trailing content

Rejecting trailing content

A JSON document is exactly one value, so anything after that value (other than whitespace) is an error. Today you make Parse insist the top-level value consumes the whole input.

The goal

Report content after a complete top-level value as an error at the extra token.

Start here - the target

TO DO

Scenario: Extra tokens after the value

Givena complete value followed by more non-whitespace

WhenParse is called

ThenParse("null null") gives "unexpected trailing content at line 1, column 6"

AndParse("[1] [2]") gives "unexpected trailing content at line 1, column 5", and Parse("1 2") gives "unexpected trailing content at line 1, column 3"

Background

A JSON text is defined as a single value with optional surrounding whitespace - not a sequence of values. So once Parse has read its one value, the only thing allowed to follow is the end of input. If another token appears - a second value, a stray bracket, leftover characters - the document is malformed, and the error points at where the extra content begins.

This matters more than it first looks. Without the check, Parse("[1] [2]") would happily return [1] and silently ignore the rest, hiding a real mistake. Requiring the top value to consume everything but trailing whitespace is what makes the parser strict about document boundaries. The scanner already skips whitespace, so a clean document ends in EOF right after its value; anything else is unexpected trailing content at the offending token. Note that a bare word like truex is caught even earlier, by the scanner, as a single Illegal token.

Make it work

// Parse reads ONE value, then must see EOF:
//   v := parseValue()
//   if next token is not EOF:
//     return ParseError{Msg: "unexpected trailing content", ...pos}
// trailing whitespace is already skipped by the scanner, so only a
// real extra token triggers this.

CheckpointDONE

Parse rejects any content after the top-level value. Commit and stop here.