Warning

The FMR maintenance client and the .Stat connectors are experimental and subject to change without prior notice.

Publishing structures from FMR to .Stat

This tutorial moves structural metadata from a Fusion Metadata Registry (FMR) to a .Stat Suite instance, end to end, using pysdmx.

The pipeline, under an agency you are authorised to maintain:

  1. author a minimal structure graph (ConceptScheme + DSD + Dataflow);

  2. upload it to the FMR;

  3. download it back from the FMR with all its references;

  4. submit the downloaded structures and a tiny dataset to .Stat.

Downloading with references=all is deliberate: it returns the whole dependency graph, which is exactly what is forwarded to .Stat.

Important

Requires a writable FMR and a writable .Stat Suite instance. Moving structures needs only the base pysdmx package (they are exchanged as SDMX-JSON); the data extra (pysdmx[data]) is needed only for the final dataset-upload step. The target agency must already exist in the SDMX:AGENCIES scheme – the FMR rejects structures whose maintenance agency is unknown.

Authentication

The FMR maintenance client takes either a bearer access_token or HTTP Basic user/password; .Stat takes a bearer token. Read them from the environment:

export FMR_ENDPOINT="https://my.fmr/sdmx/v2"  # SDMX-REST v2 base
export FMR_TOKEN="eyJ..."      # or FMR_USER + FMR_PASSWORD
export DOTSTAT_TOKEN="eyJ..."  # a fresh .Stat bearer token

Upload to the FMR

RegistryMaintenanceClient posts the artefacts to the FMR. Replace adds new artefacts and overwrites existing ones. Authenticate with a bearer token or HTTP Basic credentials:

from pysdmx.api.fmr.maintenance import (
    RegistryMaintenanceClient,
    StructureAction,
)

# with a bearer token ...
client = RegistryMaintenanceClient(
    api_endpoint=FMR_ENDPOINT, access_token=FMR_TOKEN
)
# ... or with HTTP Basic credentials:
# client = RegistryMaintenanceClient(
#     api_endpoint=FMR_ENDPOINT, user=FMR_USER, password=FMR_PASSWORD
# )

client.put_structures(
    structures=structures, action=StructureAction.Replace
)

Download from the FMR (references=all)

The same SDMX-REST base serves reads. A RestService structure query with references=all returns the full graph, which read_sdmx() parses into a message:

from io import BytesIO

from pysdmx.api.qb import (
    ApiVersion,
    RestService,
    StructureDetail,
    StructureFormat,
    StructureQuery,
    StructureReference,
    StructureType,
)
from pysdmx.io import read_sdmx

agency_id = "<your-agency>"

svc = RestService(
    api_endpoint=FMR_ENDPOINT,
    api_version=ApiVersion.V2_0_0,
    structure_format=StructureFormat.SDMX_JSON_2_0_0,
    token=FMR_TOKEN,
)
query = StructureQuery(
    artefact_type=StructureType.DATAFLOW,
    agency_id=agency_id,
    resource_id="DF_PYSDMX_FMR",
    version="1.0",
    detail=StructureDetail.FULL,
    references=StructureReference.ALL,
)
graph = read_sdmx(BytesIO(svc.structure(query)), validate=False)

Submit to .Stat

Finally, submit the downloaded structures, upload a dataset, and poll the asynchronous import to completion:

from pysdmx.api.stat import StatUploader

uploader = StatUploader(
    nsi_endpoint=NSI,
    transfer_endpoint=TRANSFER,
    dataspace=SPACE,
    token=DOTSTAT_TOKEN,
)
uploader.submit_structure(structures=graph.structures)

imported = uploader.submit_data(dataset=dataset)
final = uploader.submission_status(imported.request_id, wait=True)
print(final.execution_status, final.outcome)

Note

Some .Stat deployments drop a measure’s local representation when ingesting SDMX-JSON (submit_structure’s default), so the stored DSD can lose the measure’s data type / codelist. If that matters, submit the structures as SDMX-ML 2.1 (structure_format= with an SDMX-ML format) or read them back after upload to verify.

See the .Stat connector guide for the .Stat submission details, and reconciling .Stat against FMR for the reverse clean-up direction.