"""Offline French wind registry staging with conservative installation scope.
EICs identify network installations, not automatically whole farms. The complete
national wind extract is required for name/commune screening. Current operating
status is never extrapolated beyond the dated registry snapshot.
"""
from __future__ import annotations
import json
import re
from collections import Counter
from datetime import datetime
from typing import Any
from energy_data_retrieval.pv_catalogue.catalogue import canonical_json
from energy_data_retrieval.pv_catalogue.odre import normal_name, read_odre_json
from .catalogue import stable_id
from .gppd import evidence_record
REFERENCE_IDS = ("odre_wind_20260831", "odre_glossary_20261102")
[docs]
def read_inventory(content: bytes, *, sha256: str, expected_count: int) -> list[dict[str, Any]]:
"""Verify original bytes, national count, schema and unique EIC identities."""
rows = read_odre_json(content, sha256=sha256)
if len(rows) != expected_count or any(r.get("codefiliere") != "EOLIE" for r in rows):
raise ValueError("Require complete national EOLIE extract")
return rows
def _date(value: str | None, snapshot: str) -> str | None:
if not value:
return None
result = datetime.strptime(value, "%d/%m/%Y").date().isoformat()
if result > snapshot:
raise ValueError("Source date exceeds registry snapshot")
return result
[docs]
def stage_inventory(
rows: list[dict[str, Any]],
existing: list[dict[str, Any]],
*,
source_id: str,
now: str,
snapshot: str = "2026-08-31",
service_start: str = "2022-01-01",
service_end: str | None = None,
residual: bool = False,
) -> tuple[list[dict[str, Any]], list[dict[str, Any]]]:
"""Screen national identities and stage a bounded industrial-service cohort.
This bounded cohort avoids automatically treating historical WRI municipality
aggregates as modern EICs. Shared communes, repeated names, phase markers and
prior name/commune candidates are held. No fuzzy merge, turbine count from
nbGroupes, surveyed coordinates from administrative codes, or capacity factor.
All accepted capacities remain excluded until farm/phase scope is reconciled.
The residual partition selects missing service dates or missing/sub-1-MW
installed ratings; missing/nonpositive ratings remain held. Other cohorts
still participate in the national identity screen.
Bounds are inclusive ISO dates; the default preserves the original 2022+
cohort. Filtering never removes other years from the national identity screen.
"""
start = datetime.strptime(service_start, "%Y-%m-%d").date().isoformat()
end = datetime.strptime(service_end or snapshot, "%Y-%m-%d").date().isoformat()
if start > end or end > snapshot:
raise ValueError("Service cohort must be ordered and within the snapshot")
names = Counter(normal_name(r["nominstallation"] or "") for r in rows)
communes = Counter(r["codeinseecommuneimplantation"] for r in rows)
old = [r for r in existing if r["entity_type"] == "projects" and r.get("country_code") == "FRA"]
old_names = {
normal_name(n): r["record_id"] for r in old for n in [r["name"], *r.get("aliases", [])]
}
old_eics = {json.loads(r.get("external_ids_json") or "{}").get("odre_eic") for r in old}
accepted, ledger = [], []
for raw in sorted(rows, key=lambda r: r.get("codeeicresourceobject") or ""):
eic = raw["codeeicresourceobject"]
name = raw["nominstallation"] or ""
normalized = normal_name(name)
commune = raw["codeinseecommuneimplantation"]
service = _date(raw["datemiseenservice"], snapshot)
installed = raw["puismaxinstallee"]
reasons = []
if residual:
if service and installed is not None and installed >= 1000:
reasons.append("outside_residual_partition")
if installed is None or installed <= 0:
reasons.append("missing_or_nonpositive_installed_capacity")
elif not service or not start <= service <= end or installed is None or installed < 1000:
reasons.append(
"outside_2022_plus_ge1MW_service_cohort"
if service_start == "2022-01-01" and service_end is None
else "outside_selected_ge1MW_service_cohort"
)
if raw["codetechnologie"] != "TERRE":
reasons.append("not_explicit_onshore")
if raw["nbinstallations"] != 1:
reasons.append("aggregate_not_individual_installation")
if raw["regime"] != "En service" or raw.get("datederaccordement"):
reasons.append("service_or_disconnection_requires_review")
if not normalized or normalized in {"confidentiel", "prive"}:
reasons.append("anonymous_or_missing_name")
if normalized in {"hangar", "eolienne", "eoliennes", "parc eolien"}:
reasons.append("generic_name_requires_identity_review")
if not commune or commune.startswith(("97", "98")):
reasons.append("missing_or_overseas_location")
if communes[commune] > 1:
reasons.append("shared_commune_scope")
if names[normalized] > 1:
reasons.append("repeated_name")
if re.search(
r"\b(?:ADP|PDL|TRANCHE|EXTENSION|PHASE|II|III|IV|V|VI|IX)\b|\d\s*$", name, re.I
):
reasons.append("unit_phase_or_numbered_name")
candidates = sorted(
{
rid
for n, rid in old_names.items()
if n == normalized
or n == normal_name(raw["commune"] or "")
or (len(n) >= 6 and n in normalized)
}
)
if candidates or eic in old_eics:
reasons.append("existing_identity_candidate")
decision = {
"registry_id": eic,
"name": name,
"commune": raw["commune"],
"service_date": service,
"candidate_record_ids": candidates,
"decision": "deferred" if reasons else "added",
"reasons": reasons,
}
if reasons:
ledger.append(decision)
continue
values = {
"record_id": stable_id("projects", "ODRE-EIC", eic),
"entity_type": "projects",
"record_scope": "farm",
"name": name,
"country_code": "FRA",
"region": raw["region"],
"locality": raw["commune"],
"technology": "onshore",
"status": "operational",
"status_as_of": snapshot,
"status_date_basis": (
"ODRE En service at registry snapshot; not retrieval-day confirmation"
),
"aggregation_role": "unresolved",
"quality_status": "needs_review",
"capacity_scope": (
"Single EIC network installation; whole-farm/phase/repowering scope unverified"
),
"external_ids_json": canonical_json({"odre_eic": eic}),
"installed_capacity_mw": installed / 1000,
"export_capacity_mw": raw["puismaxrac"] / 1000
if raw.get("puismaxrac") is not None
else None,
"grid_utility": raw["gestionnaire"],
"substation": raw["postesource"],
"commissioning_date": service,
"location_accuracy": (
"Administrative delivery/installation commune; no surveyed coordinates"
),
"extensions_json": canonical_json(
{
"odre": {
"source_fields": raw,
"snapshot_date": snapshot,
"connection_date": _date(raw.get("dateraccordement"), snapshot),
"source_version_date": _date(raw.get("datedebutversion"), snapshot),
"commissioning_basis": (
"Industrial service as viewed by network manager; not whole-farm COD"
),
"generation_basis": (
"Native rolling 12-month kWh; exact per-row period endpoints "
"unspecified; production excludes self-consumption, injection is "
"network boundary. No CF calculated."
),
"group_count_basis": (
"nbGroupes is installed generating groups; "
"not promoted to turbine count"
),
"voltage_basis": "HTA/HTB are categories; no numeric voltage inferred",
}
}
),
"notes": (
"Dated registry observation, not independently verified performance. EIC scope "
"reconciliation required before capacity aggregation; no owner/model inferred "
"from installation name."
),
}
record = evidence_record(
values,
source_id=source_id,
now=now,
locator=f"codeeicresourceobject={eic}; national EOLIE JSON export",
confidence="medium",
)
evidence = json.loads(record["evidence_json"])
for target, field in [
("installed_capacity_mw", "puismaxinstallee"),
("export_capacity_mw", "puismaxrac"),
]:
if target in evidence:
evidence[target].update(
original_value=raw[field],
original_unit="kW",
normalization="original kW / 1000 = MW",
)
record["evidence_json"] = canonical_json(evidence)
accepted.append(record)
ledger.append({**decision, "record_id": record["record_id"]})
return accepted, ledger