OpenBB-finance/OpenBB · warning · KeyError

Could not find an indicator-like dimension for dataflow

Error message

Could not find an indicator-like dimension for dataflow '{dataflow_id}'.

What it means

Raised as a KeyError by get_indicators_in when the DSD parsed successfully but no dimension qualified as 'indicator-like' (e.g. no ACTIVITY-style dimension with codes), so full_indicator_list stayed empty. It indicates a structural mismatch between the dataflow's dimensions and the heuristics this utility uses to identify indicator dimensions.

Solutions

Catch KeyError for this dataflow and skip it when iterating the full catalog — not every dataflow has indicators.
Check the dataflow's dimension list (meta.datastructures[dsd_id]['dimensions']) to confirm no indicator dimension exists.
Report/patch the heuristic in metadata.py if a legitimate indicator dimension (different id than ACTIVITY) is being missed.

Example fix

# before
for df_id in meta.dataflows:
    indicators = meta.get_indicators_in(df_id)  # KeyError on layout-less dataflows

# after
for df_id in meta.dataflows:
    try:
        indicators = meta.get_indicators_in(df_id)
    except KeyError:
        continue  # dataflow has no indicator-like dimension

Defensive patterns

Strategy: fallback

Validate before calling

def likely_has_indicators(meta, df_id: str) -> bool:
    ref = meta.dataflows.get(df_id, {}).get('structureRef', {})
    dsd = meta.datastructures.get(ref.get('id', ''), {})
    return len(dsd.get('dimensions', [])) > 1  # heuristic: flat flows rarely expose indicators

Try / catch

try:
    indicators = meta.get_indicators_in(df_id)
except KeyError:
    indicators = []  # dataflow layout unsupported: degrade gracefully in catalog crawls

Prevention

When iterating all dataflows, treat KeyError as 'skip' by design.
Pin provider versions whose indicator heuristics match the IMF layout you depend on.

When it happens

Trigger: Dataflows whose dimensions are only frequency/region/time-style codes with no indicator codelist; codelist resolution failing for every candidate dimension (see _resolve_codelist_id); new dataflows published with an unfamiliar dimension layout.

Common situations: Exploratory scripts that iterate over every dataflow in the catalog and hit one with an unsupported layout; upstream IMF schema changes introducing new dataflow shapes; partial parameter-API responses swallowed by the 'except Exception' around get_dataflow_parameters.

Related errors

AI-assisted analysis of OpenBB-finance/OpenBB@3e071fcc2c (2026-08-14). Data as JSON: /api/errors/6f2028fb7f72dd08. Report an issue: GitHub.

Appendix: source

Thrown at openbb_platform/providers/imf/openbb_imf/utils/metadata.py:918

                    activity_codelist_id, {}
                )
                for code_id, code_name in codes_map.items():
                    series_id = f"{dataflow_id}::{code_id}"
                    entry = {
                        "dataflow_id": dataflow_id,
                        "dataflow_name": dataflow_name,
                        "structure_id": structure_id,
                        "agency_id": agency_id,
                        "dimension_id": "ACTIVITY",
                        "indicator": code_id,
                        "label": code_name,
                        "description": descriptions_map.get(code_id, ""),
                        "series_id": series_id,
                    }
                    full_indicator_list.append(entry)

        if not full_indicator_list:
            raise KeyError(
                f"Could not find an indicator-like dimension for dataflow '{dataflow_id}'."
            )

        return full_indicator_list

    def _resolve_codelist_id(
        self, dataflow_id: str, dsd_id: str | None, dim_id: str, dim_meta: dict
    ) -> str | None:
        if not dim_id:
            return None

        # Check for explicit codelist reference first
        representation = dim_meta.get("representation", {})
        codelist_ref = representation.get("codelist")
        if isinstance(codelist_ref, dict):
            return codelist_ref.get("id")
        if isinstance(codelist_ref, str):
            return codelist_ref

View on GitHub (pinned to 3e071fcc2c)