cocoindex-io/cocoindex · error · ValueError

Unsupported field type: {field_def.type!r}

Error message

Unsupported field type: {field_def.type!r}

What it means

When creating the Valkey search index, `_create_index` maps each declared field definition's type to a Valkey FT schema field class: 'text' → TextField, 'numeric' → NumericField, etc. A `field_def.type` outside the supported set (e.g. 'vector' handled elsewhere, or an unrecognized string like 'string' or 'keyword') falls into the else branch and raises this ValueError.

Source

Thrown at python/cocoindex/connectors/valkey/_target.py:534

            attributes=attributes,
        )

        all_fields: list[Field] = [vector_field]
        for field_def in schema.fields:
            if field_def.type == "text":
                all_fields.append(
                    TextField(name=field_def.name, sortable=field_def.sortable)
                )
            elif field_def.type == "tag":
                all_fields.append(
                    TagField(name=field_def.name, sortable=field_def.sortable)
                )
            elif field_def.type == "numeric":
                all_fields.append(
                    NumericField(name=field_def.name, sortable=field_def.sortable)
                )
            else:
                raise ValueError(f"Unsupported field type: {field_def.type!r}")

        prefix = _make_prefix(index_name)
        options = FtCreateOptions(data_type=DataType.HASH, prefixes=[prefix])

        await ft.create(client, index_name, schema=all_fields, options=options)

    def reconcile(
        self,
        key: coco.StableKey,
        desired_state: _IndexSpec | coco.NonExistenceType,
        prev_possible_records: Collection[_IndexTrackingRecord],
        prev_may_be_missing: bool,
        /,
    ) -> (
        coco.TargetReconcileOutput[_IndexAction, _IndexTrackingRecord, _DocumentHandler]
        | None
    ):
        if not isinstance(key, tuple) or len(key) != 2:

View on GitHub (pinned to e84aa99b32)

Solutions

1. Change field_def.type to a supported value ('text' or 'numeric', per the mapping in _create_index; vector fields go through the vector-def path, not this list).
2. Fix typos in type strings by matching the exact literals used in the connector's _create_index implementation.
3. If you need a type that isn't supported (e.g. tag/boolean), store it as 'text' or 'numeric', or file/patch support for that field type in the connector.

Example fix

// before
FieldDef(name="category", type="keyword", sortable=False)

// after
FieldDef(name="category", type="text", sortable=False)

SUPPORTED = {"text", "numeric"}
bad = [f.name for f in field_defs if f.type not in SUPPORTED]
assert not bad, f"Unsupported Valkey field types: {bad}"

try:
    await component.reconcile(...)
except ValueError as e:
    if "Unsupported field type" in str(e):
        raise ConfigError("Use only 'text'/'numeric' field types (vectors go via VectorDef)") from e
    raise

When it happens

Trigger: Building an index whose field defs include a type string not among the handled cases in `_create_index` — e.g. a field_def with type="keyword" or a typo like "nmeric" — during `_apply_actions` when the index is first created.

Common situations: Hand-writing FieldDef lists with Redis-style type names ('keyword', 'boolean') that Valkey's FT.CREATE mapping in this connector doesn't support; typos in type strings; schema drift after the connector added new supported types.

Related errors

• Unsupported record type: {self.record_type}
• Primary key column '{pk}' not found in columns: {list(self.c
• build_node_index_create requires at least one field
• VectorSchemaProvider is required for NumPy ndarray type.
• primary_key {primary_key!r} not found in columns ({sorted(co

AI-assisted analysis of cocoindex-io/cocoindex@e84aa99b32 (2026-09-08). Data as JSON: /api/errors/2e2660f69c269e85. Report an issue: GitHub.