Source code for energy_data_retrieval.wind_catalogue.portfolio

"""Offline staging of complete, reviewed OEM turbine specification columns.

Input is an attributed transcription checked against the rendered original,
not a text-extraction heuristic. Column identity never supplies missing values.
"""

from __future__ import annotations

import hashlib
import json
from typing import Any

from energy_data_retrieval.pv_catalogue.catalogue import canonical_json

from .catalogue import SCHEMA, stable_id
from .gppd import evidence_record

REFERENCE_IDS = ("goldwind_portfolio_2025q3",)
SPECIFICATIONS = {
    "platform",
    "rated_power_kw",
    "rated_power_conditions",
    "rotor_diameter_m",
    "swept_area_m2",
    "blade_count",
    "iec_class",
    "cut_in_m_s",
    "cut_out_m_s",
    "rated_wind_speed_m_s",
    "design_life_years",
    "design_life_conditions",
    "temperature_min_c",
    "temperature_max_c",
    "generator_type",
    "gearbox_type",
    "converter_type",
    "pitch_yaw_control",
    "tower_options",
    "hub_height_options_json",
    "extensions_json",
    "certifications_json",
}


[docs] def stage_portfolio( review: dict[str, Any], original: bytes, existing: list[dict[str, Any]], *, now: str ) -> tuple[list[dict[str, Any]], list[dict[str, Any]]]: """Stage every explicit column once; defer existing models for revision review. Unreported quantities remain absent (typed null in snapshots). Hub choices, electrical boundaries and certification observations stay nested. Neither model-code geometry nor generator voltage becomes a turbine output claim. """ if hashlib.sha256(original).hexdigest() != review["sha256"]: raise ValueError("Original portfolio checksum mismatch") expected, columns = review["expected_models"], review["columns"] names = [c["model"] for c in columns] if ( not expected or len(set(expected)) != len(expected) or len(set(names)) != len(names) or set(names) != set(expected) ): raise ValueError("Incomplete or duplicate portfolio columns") if not review["manufacturer"] or not review["revision"]: raise ValueError("Manufacturer and document revision required") prior = { (r.get("manufacturer"), r.get("model")) for r in existing if r["entity_type"] == "equipment" } records, ledger = [], [] for column in columns: model, values = column["model"], dict(column["values"]) if not model or not column["locator"] or set(values) - SPECIFICATIONS: raise ValueError("Explicit model, locator and supported specifications required") if set(values) != set(column["original_units"]): raise ValueError("Every transcribed field requires original units") for field, condition in [ ("rated_power_kw", "rated_power_conditions"), ("design_life_years", "design_life_conditions"), ]: if values.get(field) is not None and not values.get(condition): raise ValueError(f"{field} requires source conditions") key = (review["manufacturer"], model) if key in prior: ledger.append( { "model": model, "decision": "existing_identity_review", "reason": "Reconcile source revision; do not create a second model", } ) continue row = evidence_record( { "record_id": stable_id("equipment", "OEM portfolio", *key), "entity_type": "equipment", "record_scope": "model", "name": " ".join(key), "manufacturer": key[0], "model": model, "equipment_category": "wind_turbine", "source_revision": review["revision"], "variant": "Explicit published column; site-specific options nested", "capacity_scope": "Manufacturer specification, not an installed fleet", "notes": "Not independently validated performance; no numerical curves, " "component models or compatibility inferred from platform labels.", **values, }, source_id=review["source_id"], now=now, locator=column["locator"], ) evidence = json.loads(row["evidence_json"]) for field, unit in column["original_units"].items(): if field in evidence: evidence[field].update( unit=SCHEMA["fields"][field].get("unit", unit), original_unit=unit, revision=review["revision"], confidence="high", normalization="Reviewed table value; no model-code inference", ) row["evidence_json"] = canonical_json(evidence) records.append(row) ledger.append( { "model": model, "decision": "addition", "record_id": row["record_id"], "source_id": review["source_id"], "locator": column["locator"], } ) return sorted(records, key=lambda r: r["record_id"]), ledger