pandas-dev/pandas · error · NotImplementedError

arithmetic operations are not supported inside an HDFStore…

Error message

arithmetic operations are not supported inside an HDFStore 'where' filter; instead store a precomputed column as a data_column and query that, or read the data and apply the filter in pandas (e.g. df[df['A'] % 3 == 0]).

What it means

Raised in the PyTables BinOp.prune when self.op is in ARITH_OPS_SYMS ('+', '-', '*', '/', '**', '//', '%'). HDFStore 'where' filters are translated into PyTables/numexpr conditions on stored data_columns; arithmetic inside the condition would require computing new columns on disk, which PyTables does not support. Per GH#41100, rather than grow the query grammar the team chose to raise NotImplementedError with a pointer to alternatives, since PyTables support is in maintenance mode.

Solutions

1. Precompute the arithmetic result into a new column and store it as a data_column: df['A_plus_B'] = df['A'] + df['B']; store.put('df', df, format='table', data_columns=['A_plus_B']); store.select('df', where='A_plus_B > 1').
2. Read the data and filter in pandas: df = store.get('df'); df[df['A'] + df['B'] > 1].
3. Avoid HDFStore where filters for derived expressions; use Parquet + boolean indexing for new code.

Example fix

// before
store.select('df', where='A + B > 1')

// after
df['A_plus_B'] = df['A'] + df['B']
store.put('df', df, format='table', data_columns=['A_plus_B'])
store.select('df', where='A_plus_B > 1')
# or simply
df = store.get('df')
df[df['A'] + df['B'] > 1]

from pandas.core.computation.ops import ARITH_OPS_SYMS
import re

def where_has_arith(where: str) -> bool:
    # crude check: any arithmetic operator token outside string literals
    return bool(re.search(r'[+\-*/%]|\*\*|//', where))

assert not where_has_arith(where_clause), 'arithmetic is not allowed in HDFStore where clauses'

from pandas.core.computation.ops import ARITH_OPS_SYMS

def is_arith_op(op: str) -> bool:
    return op in ARITH_OPS_SYMS

try:
    store.select('df', where='A + B > 1')
except NotImplementedError as e:
    if 'arithmetic operations are not supported' in str(e):
        df = store.get('df')
        df[(df['A'] + df['B']) > 1]
    else:
        raise

When it happens

Trigger: store.select('df', where='A + B > 1'), store.select('df', where='A % 3 == 0'), or any arithmetic operator inside the where clause. Also df.query() evaluated against an HDFStore-backed TermVisitor.

Common situations: Porting a pandas boolean expression (df[df['A'] + df['B'] > 1]) to an HDFStore select without precomputing. Expecting numexpr/PyTables to evaluate derived expressions. Trying to filter on a ratio (A/B) without a stored column.

Related errors

• Cannot compare of type to column
• cannot subscript with
• cannot use an invert condition when passing to numexpr
• name is not defined
• passing a filterable condition to a non-table indexer

AI-assisted analysis of pandas-dev/pandas@3b7651241d (2026-08-11). Data as JSON: /api/errors/b4214346d172170f. Report an issue: GitHub.

Appendix: source

Thrown at pandas/core/computation/pytables.py:136

    op: str
    queryables: dict[str, Any]
    condition: str | None

    def __init__(self, op: str, lhs, rhs, queryables: dict[str, Any], encoding) -> None:
        super().__init__(op, lhs, rhs)
        self.queryables = queryables
        self.encoding = encoding
        self.condition = None

    def _disallow_scalar_only_bool_ops(self) -> None:
        pass

    def prune(self, klass):
        if self.op in ARITH_OPS_SYMS:
            # GH#41100: arithmetic in a where-clause is not supported. PyTables
            # support is in maintenance mode, so rather than grow the query
            # grammar we raise with a pointer to a working alternative.
            raise NotImplementedError(
                "arithmetic operations are not supported inside an HDFStore "
                "'where' filter; instead store a precomputed column as a "
                "data_column and query that, or read the data and apply the "
                "filter in pandas (e.g. df[df['A'] % 3 == 0])."
            )

        def pr(left, right):
            """create and return a new specialized BinOp from myself"""
            if left is None:
                return right
            elif right is None:
                return left

            k = klass
            if isinstance(left, ConditionBinOp):
                if isinstance(right, ConditionBinOp):
                    k = JointConditionBinOp
                elif isinstance(left, k):

View on GitHub (pinned to 3b7651241d)