JSON grammar rules are inconsistent.

[springcomp]

The JMESPath grammar mostly includes rules for parsing valid JSON texts. However, the JSON RFC includes errata that were not included in this spec.

In particular, the two unescaped-char and escaped-char rules are inconsistent:

quoted-string     = quote 1*(unescaped-char / escaped-char) quote
unescaped-char    = %x20-21 / %x23-5B / %x5D-10FFFF
escape            = "\"
quote             = %x22   ; Double quote: '"'
escaped-char      = escape (
                        %x22 /          ; "    quotation mark  U+0022
                        %x5C /          ; \    reverse solidus U+005C
                        %x2F /          ; /    solidus         U+002F
                        %x62 /          ; b    backspace       U+0008
                        %x66 /          ; f    form feed       U+000C
                        %x6E /          ; n    line feed       U+000A
                        %x72 /          ; r    carriage return U+000D
                        %x74 /          ; t    tab             U+0009
                        %x75 4HEXDIG )  ; uXXXX                U+XXXX

The / U+002F SOLIDUS character is present in those two rules. It should, however, be excluded from the unsecaped-char rule.

This PR fixes this issue.

[gibson042]

Sorry I missed this PR a couple weeks ago, but that JSON RFC erratum has not been verified and basically cannot be accepted because JSON is synchronized between the IETF and ECMA... U+002F solidus "/" is allowed inside JSON strings both unescaped and preceded by an escaping backslash.

But in JMESPath, the unescaped-char nonterminal is primarily for identifier via quoted-string and shouldn't even relate to JSON text at all.

[springcomp]

@gibson042 I trust you that the JSON errata has not been verified, and I agree that solidus should be valid escaped or non-escaped. I will revert this change.

However, the quoted-string rule from JMESPath is shared between identifier and json-value. JSON and JMESPath do have some relation at least.

But reverting the PR should be sufficient I reckon.

[gibson042]

However, the quoted-string rule from JMESPath is shared between identifier and json-value. JSON and JMESPath do have some relation at least.

Right, and that strikes me as a problem... the keys for JSON object members must be JSON strings, but right now JMESPath inconsistently has both json-value = false / null / true / json-object / json-array / json-number / json-quoted-string (where json-quoted-string introduces a \` escape and disallows unescaped `) and member = quoted-string name-separator json-value (where quoted-string allows unescaped ` and does not have a \` escape), which doesn't make any sense.

[springcomp]

I think you are right. Would that be a simple oversight that could be fixed with?

member = json-quoted-string name-separator json-value

[gibson042]

Yes, I think so.

[springcomp]

125.