Report DDS

Six-section JSON report for the DDS record's current version. Re-shapes data already on fm-api (no fresh assessment calls). The output's wire shape is determined by the version's report_model field — eudr_default_v1 for new records.

Path params

ParamDescription
recordIdDDS record id

sections — partial report

Comma-separated list of the parts to return. Omit it and you get the whole statement, byte for byte as before.

SectionWhat it carries
headerOperator, record identity, processing date
data_lineageSources and generation provenance
data_integrityLocation-validation confidence rollup
articlesArticle 9, deforestation, legality, Article 10, audit trail
appendixNon-compliant, exempt and operator-flipped plots
signatureSigning block
plot_registerEvery plot with its geometry — nested inside articles.deforestation, and the bulk of a large statement

plot_register is named separately from its parent so the compliance picture can be fetched without the tonnage: ?sections=header,data_integrity,articles. Being nested, it cannot be asked for without articles — that combination is a 400, as is any name outside the table.

dds_hash is unaffected: it is computed over the whole report before anything is pruned, so a partial fetch and a full one agree on it (see Hash semantics below). What was left out is reported in response headers rather than in the body, which stays a strict subset of the hashed tree:

HeaderValue
x-dds-omitted-sectionsComma-separated names of every section dropped, spelled with underscores exactly as ?sections= spells them. A report of what was left out, not a retry token: if plot_register is in it, a follow-up request needs articles alongside it, since the register is nested inside that section
x-dds-omitted-<section>Row count that section would have carried, e.g. x-dds-omitted-plot-register: 4210. The header name is hyphenated, because a header name conventionally cannot carry underscores

So an absent section is never confused with an empty one. Both headers are CORS-exposed, so a browser client can read them too.

sections bounds the RESPONSE, not the work behind it: the report is assembled and hashed in full before anything is pruned, so a narrow selection cuts payload and client-side parsing but not server latency.

The row count is the number of rows in the section's plot / event lists — plot_register, the appendix's per-verdict plots and its operator_attachments.attachments, the article audit_trail, and data_lineage's dataset_versions. Sections that hold no such list (header, data_integrity, signature) report 0: they are objects, not rows, and their absence is announced by x-dds-omitted-sections regardless.

Status fields on conclusion_and_scope

The conclusion-and-scope block carries three status fields. They look similar but are deliberately distinct:

FieldValuesWhat it tells you
submission_statussubmitted_compliant / submitted_non_compliant / not_submitted_compliant / not_submitted_non_compliantThe composite "have we filed × is the data clean?" verdict the PDF header inlines. Driven by the same TRACES-completion signal as computed_status.status crossed with the per-plot deforestation buckets.
computed_status.statusdraft / submitting / submitted / manually_submitted / submission_failedRaw audit-log derivation — which submit-lifecycle event was the latest. Used by reconciliation (flat_matches / discrepancies) to detect drift between the audit log and the remaining TRACES-identifier columns on dds_record_version.
computed_status.lifecycle_statusdraft / awaiting_submission / submitting / submitted / accepted / rejected / stale§8.1 dashboard verdict — composes computed_status.status with TRACES round-trip status (GAP-13) and the staleness flag (GAP-10) to produce the enum a UI renders. A submitted status can resolve to accepted / rejected / stale here, and a draft status can resolve to awaiting_submission.

Rule of thumb: read computed_status.status if you're verifying the audit log lines up with the persisted state; read computed_status.lifecycle_status if you're showing the operator where the DDS sits; read submission_status if you want the combined filed × compliant summary the PDF prints.

methodology.datasets vs methodology.dataset_versions

datasets lists the reference datasets declared for the report model. dataset_versions identifies dataset versions used by the assessments on this DDS version, with sources[] identifying the assessment types that referenced them. It can be empty when provenance is not available.

Read dataset provenance alongside methodology_versions when comparing reports created at different times.

Per-plot compliance verdict & operator overrides

Every entry in deforestation_assessment.per_plot[] carries two fields that summarise the plot's compliance state after operator input is applied:

FieldValuesMeaning
false_positive_resolution.operator_action"included" / "excluded" / "none"What the operator did with the plot — wrote a justification, removed it from scope via an exclusion-axis finding, or neither.
false_positive_resolution.compliance_override"compliant" / "non_compliant" / nullThe operator's manual override of the computed verdict (set via the compliance_override field on Set Justification). null means no override is set — the verdict reverts to the alert-derived value.
false_positive_resolution.justification_text, …evidence_urlsstring / arrayThe narrative + supporting links that ride on the same row as the override.
effective_verdict"compliant" / "non_compliant" / nullThe verdict an inspector should read. Excluded plots and metric-less plots are null (out of scope / no signal yet). Otherwise: the override wins when set, else the alert-derived verdict (critical → non_compliant, everything else → compliant).

The §6 plot_register[].justification block mirrors the same compliance_override field for callers that read the operator decision off the plot register rather than the assessment block.

Raw alert labels (deforestation_assessment.per_plot[].deforestation.deforestation_alert, …deforestation_confidence, and plot_register[].deforestation_alert) are never rewritten by an override — they always reflect the upstream signal. A diverging effective_verdict is the signal that operator interpretation has shifted the verdict.

Aggregate counts on the version (read via the version endpoint or the report's scope block) follow the same posture: a byVerdict rollup reflects post-override state, while byDeforestationAlert stays raw.

Hash semantics

attributes.dds_hash is a sha256:<hex> fingerprint of the report's content. It lets you detect whether two reports describe the same data, and lets an auditor verify a report hasn't been altered. The same (record, version) data always produces the same hash — it does not change just because the report was regenerated at a different time.

How it is computed

The hash is taken over this JSON /report envelope (the GeoJSON variant only echoes the value under properties.dds_hash — it is not itself the hashed document). To compute it:

  1. Remove the four generation-time fields (they record when the report was rendered, not what it contains, so they must not affect the hash):
    • attributes.dds_hash (it can't contain its own hash)
    • attributes.generated_at
    • attributes.header.processing_date
    • attributes.data_lineage.generated_at
  2. Canonicalise the remaining object: recursively sort every object's keys alphabetically. Array order is preserved (it is part of the content).
  3. Serialise to JSON with no insignificant whitespace.
  4. SHA-256 the UTF-8 bytes and prefix the lowercase hex digest with sha256:.

Reproducing / verifying the hash

To independently verify a report:

  1. GET this JSON /report (not .geojson).
  2. Drop the four fields listed above.
  3. Canonicalise + serialise + SHA-256 as above.
  4. Compare your digest to the attributes.dds_hash you removed.

The canonicalisation follows JavaScript's JSON.stringify semantics. If you reproduce it in another language, match those exactly — the common pitfalls are number formatting (e.g. 2.0 serialises as 2, not 2.0), minimal string escaping (only JSON-required escapes, non-ASCII left as UTF-8), and key ordering by UTF-16 code unit. Differences in any of these change the bytes and therefore the digest.

Errors

StatusCodeWhen
400VALIDATION_ERRORsections names an unknown section, or plot_register without articles
401UNAUTHORIZEDMissing/invalid API key
404NOT_FOUNDRecord isn't visible to the caller

Required scope: ingestions:read