apache/superset · error · Exception

Unknown type: {native_type}

Error message

Unknown type: {native_type}

What it means

get_sqla_type maps a native type string from an imported dataset's column metadata to a SQLAlchemy type using a fixed type_map plus a VARCHAR(n) regex. When the string matches neither, it raises a bare Exception('Unknown type: ...'). This runs during import when the dataset declares a column type the importer cannot translate (used e.g. by get_dtype for dataframe loading).

Source

Thrown at superset/commands/dataset/importers/v1/utils.py:109

    "FLOAT": Float(),
    "FLOAT64": Float(),
    "DOUBLE PRECISION": Float(),
    "DATE": Date(),
    "DATETIME": DateTime(),
    "TIMESTAMP WITHOUT TIME ZONE": DateTime(timezone=False),
    "TIMESTAMP WITH TIME ZONE": DateTime(timezone=True),
}


def get_sqla_type(native_type: str) -> TypeEngine:
    if native_type.upper() in type_map:
        return type_map[native_type.upper()]

    if match := VARCHAR.match(native_type):
        size = int(match.group(1))
        return String(size)

    raise Exception(  # pylint: disable=broad-exception-raised
        f"Unknown type: {native_type}"
    )


def get_dtype(df: pd.DataFrame, dataset: SqlaTable) -> dict[str, TypeEngine]:
    return {
        column.column_name: get_sqla_type(column.type)
        for column in dataset.columns
        if column.column_name in df.keys()
    }


def validate_data_uri(data_uri: str) -> None:
    """
    Validate that the data URI is permitted for dataset import.

    Local ``file://`` URIs are allowed only when the path is confined to the
    bundled examples folder.  All other URIs must match a pattern in

View on GitHub (pinned to f4587218dd)

Solutions

1. Edit the dataset YAML and change the offending column's 'type' to a known generic type (STRING, TEXT, INTEGER, TIMESTAMP WITH TIME ZONE, or VARCHAR(n)) before importing
2. Upgrade Superset to a version whose type_map includes the native type
3. If the type is legitimately needed, add it to type_map via a fork/patch and contribute it upstream

Example fix

# before (YAML column)
- column_name: payload
  type: JSONB
# after
- column_name: payload
  type: TEXT

import re
from superset.commands.dataset.importers.v1.utils import get_sqla_type

def check_types(config):
    bad = []
    for col in config.get('columns', []):
        t = col.get('type', '')
        try:
            get_sqla_type(t)
        except Exception:
            if not re.match(r'^VARCHAR\((\d+)\)$', t.upper()):
                bad.append((col.get('column_name'), t))
    return bad

bad = check_types(config)
assert not bad, f'untranslatable column types: {bad}'

def has_known_types(config: dict) -> bool:
    return not check_types(config)

try:
    import_dataset(config)
except Exception as ex:
    if str(ex).startswith('Unknown type:'):
        native = str(ex).split(':', 1)[1].strip()
        # map native -> generic (e.g. JSONB -> TEXT) in the config and retry once
        ...

When it happens

Trigger: Importing a dataset YAML whose column 'type' field is a native type not present in type_map (e.g. engine-specific types like 'JSONB', 'MONEY', 'ENUM(...)') and not matching the VARCHAR(n) pattern; or a typo'd/empty type string.

Common situations: Bundles exported from an engine whose type names are not in the map, hand-edited YAML with a wrong type string, or version drift where a newly supported source type is not yet in this Superset version's type_map.

Related errors

• {file_name} has no valid keys
• {file_name} is not a valid file
• ; ".join(str(message) for message in ex.messages)
• Data URI is not allowed.
• Only the default catalog is supported for this connection

AI-assisted analysis of apache/superset@f4587218dd (2026-08-14). Data as JSON: /api/errors/2363b7d8ce42901b. Report an issue: GitHub.