French DPE Leads Properties and Details Scraper

This Actor is prepared to match ADEME energy-performance certificates with legal-person property owners in the DGFiP archive, then optionally enrich matched SIRENs from the public SIRENE API. Live source requests are currently disabled: Chrome could not open the configured ADEME records endpoint, and the current ADEME usage-terms page was access denied. The default run therefore finishes as SUCCEEDED / SKIPPED_POLICY with zero requests and zero dataset rows.

Source and data scope

The ADEME dataset page identifies the source as Open Licence 2.0, but its records endpoint could not be inspected in Chrome. The DGFiP join covers legal-person owner rows only; no match does not prove that a property has no owner. The Actor does not use alternate browsers, HTTP clients, proxies, indirect routes, or access-control workarounds to reach a blocked source.

The input schema describes bounded source, owner-join, SIRENE, and monitoring controls for a future reviewed run. While the policy gate in src/config.js is closed, these settings do not make source requests. The optional Slack digest runs only when monitoring is enabled and the caller supplies a webhook; the current policy skip makes no digest request.

Dataset and run status

The dataset contains only validated source-backed property DPE rows: lead for a matched legal-person owner and dpe-unmatched when a valid DPE has no owner match. Operational diagnostics, empty-result placeholders, and error records are not dataset rows. Stage errors and request totals are summarized under OUTPUT and RUN_METADATA.

• SUCCEEDED / COMPLETED: collection finished without failures, including a clean empty result or reaching a configured item bound.
• SUCCEEDED / PARTIAL: at least one real property row was stored before a later source, dataset, or summary failure; the Actor exits successfully.
• FAILED / FAILED: one or more failures occurred and no real property row was stored; the Actor exits with code 1.
• SUCCEEDED / SKIPPED_POLICY: source review is not verified in Chrome, so the Actor makes no source requests and exits successfully.

recordCount, successfulItems, failureCount, and request counts in the summary reconcile with the active dataset. Its diagnostic row count remains zero.

Local checks

npm test
npm run check
npm run validate
apify validate-schema

The offline tests use synthetic ADEME/DGFiP/SIRENE responses. The default local Actor run exercises only the policy gate and makes no source request. These checks do not prove live extraction or current source access.

Runtime storage and dependencies

The runtime uses Node.js HTTPS fetch and Apify’s documented default Dataset/KVS REST APIs. It no longer includes the unused Apify SDK/Crawlee HTTP caching stack. JSON input and records are supported; the platform provides the run token and default storage IDs through environment variables. Credentials are sent only in the Authorization header to api.apify.com. Storage redirects are refused. Dataset writes are not retried after uncertain failures, to prevent duplicates; read and idempotent summary writes may retry once. Local verification uses APIFY_LOCAL_STORAGE_DIR and preserves existing dataset rows. Source permission and access requirements still apply.