DAG-JSON `decode()` doesn't validate `{" /": ...}` namespace strictly

[sumanjeet0012]

The DAG-JSON spec says any {" /": ...} that isn't a CID link or bytes is an error in strict mode. _restore_from_json() silently passes through unrecognized patterns.

Problem

In dag/codecs/dag_json.py:

def _restore_from_json(node: Any) -> Any:
    if isinstance(node, dict):
        if len(node) == 1 and _LINK_KEY in node:
            link_value = node[_LINK_KEY]

            if isinstance(link_value, str):
                return _parse_cid_string(link_value)

            if isinstance(link_value, dict) and len(link_value) == 1 and "bytes" in link_value:
                return _base64_decode_no_pad(link_value["bytes"])

            # ← Unrecognized {"/": ...} patterns are silently ignored!

        return {k: _restore_from_json(v) for k, v in node.items()}
    # ...

This means:

# These should raise errors in strict mode, but don't:
decode(b'{" /": 123}')                    # Integer value
decode(b'{" /": [1, 2, 3]}')              # Array value
decode(b'{" /": {"unknown": "value"}}')   # Unrecognized dict

Proposed Solution

Add strict validation:

def _restore_from_json(node: Any, strict: bool = True) -> Any:
    if isinstance(node, dict):
        if len(node) == 1 and _LINK_KEY in node:
            link_value = node[_LINK_KEY]

            if isinstance(link_value, str):
                return _parse_cid_string(link_value)

            if isinstance(link_value, dict) and len(link_value) == 1 and "bytes" in link_value:
                return _base64_decode_no_pad(link_value["bytes"])

            if strict:
                raise ValueError(
                    f'Invalid DAG-JSON: {{"/": ...}} must be a CID string or '
                    f'{{"bytes": "..."}}, got {type(link_value).__name__}'
                )

        return {k: _restore_from_json(v, strict) for k, v in node.items()}
    # ...

Add a strict parameter to DagJsonCodec.decode():

def decode(self, data: bytes, strict: bool = True) -> IPLDNode:
    raw = json.loads(data)
    return _restore_from_json(raw, strict=strict)

Related

• DAG-JSON spec: https://ipld.io/specs/codecs/dag-json/spec/
• File: dag/codecs/dag_json.py