Fix relative-json-pointer rejecting multi-digit integers containing a zero

[nyxst4ck]

The relative-json-pointer format checker rejects valid pointers whose
non-negative-integer prefix contains a 0 after the first digit.

The leading-zero guard checked the previous character rather than the first
digit:

# jsonschema/_format.py, is_relative_json_pointer
if i > 0 and int(instance[i - 1]) == 0:
    return False

So any integer prefix with a 0 followed by another digit was wrongly rejected,
while the bundled conformance suite only exercises "120" (where the 0 is
last) and so never caught it:

>>> from jsonschema import FormatChecker
>>> FormatChecker().conforms("100", "relative-json-pointer")
False        # should be True
>>> FormatChecker().conforms("100/foo", "relative-json-pointer")
False        # should be True

Per draft-handrews-relative-json-pointer-01 §3, only a leading zero is
forbidden: "0" by itself is valid, "01" is not, and "100" / "205" /
"1000" are valid.

Fix

Reject only when the accumulated integer is exactly "0" and another digit
follows (i.e. a genuine leading zero), instead of rejecting any digit that
follows a 0.

Tests

Added test_relative_json_pointer_allows_multidigit_integers to
jsonschema/tests/test_format.py: 0/1/120/100/205/1000/100/foo/100# are valid
and 01/00/007/01/a/01# are not. It fails on main and passes with the fix.
Full test_format.py + the JSON Schema test suite stay green (7628 passed);
ruff clean.

(Happy to add a CHANGELOG.rst entry if you'd like.)

[read-the-docs-community]

Documentation build overview

📚 python-jsonschema | 🛠️ Build #33172853 | 📁 Comparing 7987ad1 against latest (1a77549)

  🔍 Preview build  

1 file changed

± _modules/jsonschema/_format/index.html