pandas-dev/pandas · error · TypeError

Resolver of type ' ' does not implement the __getitem__…

Error message

Resolver of type '{name}' does not implement the __getitem__ method

What it means

Raised by _check_resolvers() when an element of the resolvers list/tuple does not implement __getitem__. Resolvers are dict-like objects injected for variable name lookup (e.g., DataFrame.index and DataFrame.columns in df.query). Each resolver must support square-bracket access so the scope can resolve names like 'index' or column names to values.

Solutions

1. Ensure every element of resolvers is a mapping or supports __getitem__ — wrap custom objects in a dict or add a __getitem__ method.
2. Pass resolvers as a list/tuple of dicts: resolvers=[{'col': values}, {'index': idx}].
3. If you only need DataFrame column/index resolution, rely on df.query/df.eval defaults and do not pass resolvers at all.

Example fix

// before
pd.eval('a + 1', resolvers=[my_dataclass_instance])

// after
pd.eval('a + 1', resolvers=[{'a': my_series}])

def validate_resolvers(resolvers):
    if resolvers is None:
        return
    for r in resolvers:
        if not hasattr(r, '__getitem__'):
            raise TypeError(f'Resolver {type(r).__name__} must support __getitem__')

validate_resolvers(resolvers)
pd.eval(expr, resolvers=resolvers)

from collections.abc import Mapping
def is_valid_resolver(r) -> bool:
    return isinstance(r, Mapping) or hasattr(r, '__getitem__')

try:
    pd.eval(expr, resolvers=resolvers)
except TypeError as e:
    if '__getitem__' in str(e):
        # replace offending resolver with a dict
        ...
    raise

When it happens

Trigger: Passing resolvers=[some_object] where some_object is a plain int, a custom class without __getitem__, a frozenset, or any non-mapping type. Common when users pass a single dict without wrapping it in a list — though a bare dict does have __getitem__, the error appears with non-dict objects like a Series.index passed directly as a resolver element rather than wrapped.

Common situations: Building a custom resolver pipeline for query(); passing a named tuple or dataclass instance expecting it to behave like a mapping; migrating code that previously passed a list of dicts and accidentally included a non-dict element.

Related errors

• Cannot assign expression output to target
• expr must be a string to be evaluated
• Only named functions are supported
• unsupported operand type(s) for
• can only assign a single expression

AI-assisted analysis of pandas-dev/pandas@3b7651241d (2026-08-11). Data as JSON: /api/errors/df00a55251082c38. Report an issue: GitHub.

Appendix: source

Thrown at pandas/core/computation/eval.py:107

    parser : str

    Raises
    ------
    KeyError
      * If an invalid parser is passed
    """
    if parser not in PARSERS:
        raise KeyError(
            f"Invalid parser '{parser}' passed, valid parsers are {PARSERS.keys()}"
        )


def _check_resolvers(resolvers) -> None:
    if resolvers is not None:
        for resolver in resolvers:
            if not hasattr(resolver, "__getitem__"):
                name = type(resolver).__name__
                raise TypeError(
                    f"Resolver of type '{name}' does not "
                    "implement the __getitem__ method"
                )


def _check_expression(expr) -> None:
    """
    Make sure an expression is not an empty string

    Parameters
    ----------
    expr : object
        An object that can be converted to a string

    Raises
    ------
    ValueError
      * If expr is an empty string

View on GitHub (pinned to 3b7651241d)