What causes "Column of type {dtypes[0]} has no attribute {expression._name}." in pathwaycom/pathway?

Fix the dtype first: parse strings with dt.strptime() into DATE_TIME_UTC/NAIVE before using .dt methods, or cast the column to the type the namespace requires

pathwaycom/pathway · error · AttributeError

Column of type {dtypes[0]} has no attribute {expression._nam

Error message

Column of type {dtypes[0]} has no attribute {expression._name}.

What it means

For a method-call expression (attribute access like .dt, .str, Json accessors, custom method mappings), the evaluator resolves a handler from the actual argument dtypes via expression.get_function(dtypes). If no method is registered for that dtype under the requested attribute name, it raises AttributeError mirroring python semantics: the column's type has no such attribute.

Source

Thrown at python/pathway/internals/graph_runner/expression_evaluator.py:782

    def eval_method_call(
        self,
        expression: expr.MethodCallExpression,
        eval_state: RowwiseEvalState | None = None,
    ):
        dtypes = tuple([arg._dtype for arg in expression._args])
        if (dtypes_and_handler := expression.get_function(dtypes)) is not None:
            new_dtypes, _, handler = dtypes_and_handler

            expressions = [
                self.eval_expression(
                    expr.CastExpression(dtype, arg),
                    eval_state=eval_state,
                )
                for dtype, arg in zip(new_dtypes, expression._args)
            ]
            return handler(*expressions)
        raise AttributeError(
            f"Column of type {dtypes[0]} has no attribute {expression._name}."
        )

    def eval_unwrap(
        self,
        expression: expr.UnwrapExpression,
        eval_state: RowwiseEvalState | None = None,
    ):
        val = self.eval_expression(expression._expr, eval_state=eval_state)
        return api.Expression.unwrap(val)

    def eval_fill_error(
        self,
        expression: expr.FillErrorExpression,
        eval_state: RowwiseEvalState | None = None,
    ):
        dtype = expression._dtype
        ret = self.eval_expression(

View on GitHub (pinned to fa2f74a464)

Solutions

Fix the dtype first: parse strings with dt.strptime() into DATE_TIME_UTC/NAIVE before using .dt methods, or cast the column to the type the namespace requires
Unwrap Optional: apply .unwrap() (or fill errors) so a Optional[DateTimeUtc] becomes DateTimeUtc before attribute access
Check the message's dtype against the method's supported types in the docs; correct the method name if it is a typo

Example fix

// before
formatted = pw.this.raw.dt.strftime('%Y-%m')  # raw is str
// after
parsed = pw.this.raw.dt.strptime('%Y-%m-%d %H:%M:%S%z')
formatted = parsed.dt.strftime('%Y-%m')

Defensive patterns

Strategy: type-guard

Validate before calling

def dtype_supports_namespace(dtype, namespace: str) -> bool:
    ns_requires = {"dt": {"DATE_TIME_NAIVE", "DATE_TIME_UTC"}, "str": {"STRING"}}
    return str(dtype) in ns_requires.get(namespace, set())

Type guard

from pathway.internals import dtype as dt

def is_datetime_dtype(dtype) -> bool:
    return dtype in (dt.DATE_TIME_UTC, dt.DATE_TIME_NAIVE)

Try / catch

try:
    run()
except AttributeError as e:
    if "has no attribute" in str(e):
        # parse/cast the column to the dtype the method requires, then retry
        ...

Prevention

strptime into a datetime dtype before using .dt methods
Unwrap Optional columns before attribute access
Add a schema assertion right after ingestion that key columns have the expected dtypes

When it happens

Trigger: Calling .dt methods on a str or int column (pw.this.col.dt.strftime(...) where col is not a datetime); .str.contains on an int column; Json accessors on a non-Json column; .dt on DateTimeNaive where only DateTimeUtc supports the method; typo in the method name.

Common situations: Schema drift: a column that used to parse as DateTimeUtc now arrives as str, breaking chained .dt calls; Optional-wrapped datetimes losing the namespace; copying example code against a differently-typed table.

Related errors

AI-assisted analysis of pathwaycom/pathway@fa2f74a464 (2026-08-15). Data as JSON: /api/errors/2c4f7278fcda247e. Report an issue: GitHub.