energy_data_retrieval.data_center_catalogue.inventories#
Offline staging for bounded official operator and equipment inventories.
Only named inventory links are retained. Detail-page specifications, lifecycle, capacity boundaries and reuse conditions still require review before import.
- energy_data_retrieval.data_center_catalogue.inventories.parse_named_inventory(content, *, source_url, expected_count)[source]#
Stage exact atNorth site codes or CoolIT CDU product slugs, fail on drift.
Duplicate navigation/card links collapse only when their identity and URL agree. A second URL for one immutable identity requires manual review. Never infer MW, lifecycle, cooling, or physical scope from inventory links.
- Parameters:
content (bytes)
source_url (str)
expected_count (int)
- Return type:
dict[str, Any]
- energy_data_retrieval.data_center_catalogue.inventories.parse_energy_star_certified_data_centers(content, *, source_url, certification_year, expected_total_count, expected_selected_count)[source]#
Stage one exact certification-year partition of EPA’s data-center registry.
EPA deliberately reports only the state for these facilities. Facility names are therefore retained verbatim, but address-like text is never promoted to an address, locality or coordinate. Floor-space values retain their source text because the public table does not label a unit.
- Parameters:
content (bytes)
source_url (str)
certification_year (int)
expected_total_count (int)
expected_selected_count (int)
- Return type:
dict[str, Any]
- energy_data_retrieval.data_center_catalogue.inventories.parse_project_permit_matrix(content, *, source_url, expected_count)[source]#
Stage exact public project-permit rows without promoting scope or capacity.
Each row is one dated administrative observation. The adapter retains the authority’s identifier, project boundary and optional reported building area; it deliberately does not infer construction start, IT load or grid capacity.
- Parameters:
content (bytes)
source_url (str)
expected_count (int)
- Return type:
dict[str, Any]
- energy_data_retrieval.data_center_catalogue.inventories.parse_energy_star_ups_inventory(content, *, source_url, partner, expected_count)[source]#
Stage one complete ENERGY STAR UPS partner inventory without merging models.
The public EPA dataset is certification-entry oriented. Each ENERGY STAR unique ID is therefore retained as an immutable configuration identity; similarly named models are not collapsed and percentage-load runtime or efficiency values are not converted into unrelated electrical boundaries.
- Parameters:
content (bytes)
source_url (str)
partner (str)
expected_count (int)
- Return type:
dict[str, Any]
- energy_data_retrieval.data_center_catalogue.inventories.parse_energy_star_enterprise_servers_inventory(content, *, source_url, partner, expected_count)[source]#
Stage one complete ENERGY STAR enterprise-server partner inventory.
EPA publishes one immutable certification ID per server-family entry. The low-end, typical and high-end SERT configurations remain attached to that entry; they are not combined into a synthetic server input-power rating or split into configurations when the registry does not print a configuration identifier.
- Parameters:
content (bytes)
source_url (str)
partner (str)
expected_count (int)
- Return type:
dict[str, Any]
- energy_data_retrieval.data_center_catalogue.inventories.parse_energy_star_large_network_equipment_inventory(content, *, source_url, expected_count)[source]#
Stage the complete ENERGY STAR large-network-equipment inventory.
Each issuer ID remains one certified product entry. Minimum, typical and maximum test configurations stay attached to that identity and are never promoted into a generic device-input or rack-input power rating.
- Parameters:
content (bytes)
source_url (str)
expected_count (int)
- Return type:
dict[str, Any]
- energy_data_retrieval.data_center_catalogue.inventories.parse_energy_star_data_center_storage_inventory(content, *, source_url, dataset_kind, expected_count)[source]#
Stage one complete ENERGY STAR data-center-storage dataset.
Block-I/O and File-I/O certifications use different immutable identifiers, tested-configuration prefixes and workload metrics. This reader preserves the complete issuer row and never merges a storage family across datasets, converts PSU nameplates into equipment input, or treats tested cache/device quantities as deployed capacity.
- Parameters:
content (bytes)
source_url (str)
dataset_kind (str)
expected_count (int)
- Return type:
dict[str, Any]
- energy_data_retrieval.data_center_catalogue.inventories.parse_cdu_spec_matrix(content, *, source_url, expected_count)[source]#
Stage an extracted OEM CDU matrix without inventing optional variants.
The offline CSV keeps the model and its stated rating conditions together. It is a review aid, not an importer: source PDF locators and field evidence must still be attached before canonical publication.
- Parameters:
content (bytes)
source_url (str)
expected_count (int)
- Return type:
dict[str, Any]
- energy_data_retrieval.data_center_catalogue.inventories.parse_precision_cooling_matrix(content, *, source_url, expected_count)[source]#
Stage exact OEM cooling rows while retaining each published rating point.
This adapter deliberately accepts one documented model/configuration per row. Voltage choices, optional heaters/humidifiers and alternate rating points stay in reviewed observations; the adapter does not expand them into combinations.
- Parameters:
content (bytes)
source_url (str)
expected_count (int)
- Return type:
dict[str, Any]
- energy_data_retrieval.data_center_catalogue.inventories.parse_power_equipment_matrix(content, *, source_url, expected_count)[source]#
Stage exact transformer/switchgear table rows for manual review.
Maximum voltage/current values remain limits rather than nominal inputs, and transformer kVA is never converted to active kW.
- Parameters:
content (bytes)
source_url (str)
expected_count (int)
- Return type:
dict[str, Any]
- energy_data_retrieval.data_center_catalogue.inventories.parse_doe_transformer_certification_matrix(content, *, source_url, expected_count)[source]#
Stage exact public DOE transformer basic-model certification rows.
The represented efficiency remains a source-qualified observation because the public results table does not print the load/test condition. Apparent kVA is never converted to active kW, and the public listing is not treated as a DOE determination that a model complies with a conservation standard.
- Parameters:
content (bytes)
source_url (str)
expected_count (int)
- Return type:
dict[str, Any]
- energy_data_retrieval.data_center_catalogue.inventories.parse_doe_crac_certification_matrix(content, *, source_url, expected_count)[source]#
Stage exact public DOE computer-room air-conditioner basic-model rows.
The adapter retains the issuer’s printed Btu/h and SCFM values without repairing cross-field inconsistencies. A public CCMS listing is a manufacturer representation, not a DOE compliance determination.
- Parameters:
content (bytes)
source_url (str)
expected_count (int)
- Return type:
dict[str, Any]
- energy_data_retrieval.data_center_catalogue.inventories.parse_doe_ups_certification_matrix(content, *, source_url, expected_count)[source]#
Stage exact public DOE UPS basic/individual-model rows.
Watts and volt-amperes remain distinct. Average load-adjusted efficiency is retained with its supported input-dependency-mode boundary, and a public CCMS listing is not treated as a DOE compliance determination.
- Parameters:
content (bytes)
source_url (str)
expected_count (int)
- Return type:
dict[str, Any]
- energy_data_retrieval.data_center_catalogue.inventories.parse_ups_rating_matrix(content, *, source_url, expected_count)[source]#
Stage exact UPS family/rating rows without deriving active power.
Each CSV row must be an explicitly published OEM configuration. Voltage and frequency are retained as text so optional or multi-market choices are not silently converted into one selected nominal value. Active kW is accepted only when the source explicitly prints it for that row.
- Parameters:
content (bytes)
source_url (str)
expected_count (int)
- Return type:
dict[str, Any]
- energy_data_retrieval.data_center_catalogue.inventories.parse_server_accelerator_matrix(content, *, source_url, expected_count)[source]#
Stage explicit OEM server/accelerator compatibility rows for review.
The adapter accepts only rows which the OEM marks supported. Accelerator power remains a per-device limit and is never promoted to server or rack input. A range can be retained in
max_power_textwhile the normalized scalar stays null.- Parameters:
content (bytes)
source_url (str)
expected_count (int)
- Return type:
dict[str, Any]
- energy_data_retrieval.data_center_catalogue.inventories.parse_server_accelerator_shipping_matrix(content, *, source_url, expected_count)[source]#
Stage OEM shipping rows without treating maxima as installed counts.
Blank quantities remain null. Parenthesized quantities are retained only as configuration-dependent OEM maxima; callers must not promote them to an installed accelerator count or multiply them into server/rack power.
- Parameters:
content (bytes)
source_url (str)
expected_count (int)
- Return type:
dict[str, Any]
- energy_data_retrieval.data_center_catalogue.inventories.parse_generator_rating_matrix(content, *, source_url, expected_count)[source]#
Stage exact OEM generator rating rows without converting kVA to kW.
The extracted matrix must retain one source row per model/frequency pair. Rows without an explicit Data Center Continuous rating are not candidates; omission decisions remain in the research ledger. Portfolio voltage lists, duty claims and other shared conditions remain observations during review.
- Parameters:
content (bytes)
source_url (str)
expected_count (int)
- Return type:
dict[str, Any]
- energy_data_retrieval.data_center_catalogue.inventories.parse_generator_dcp_matrix(content, *, source_url, expected_count)[source]#
Stage exact Baudouin Data Centre Power rows without boundary conversion.
Engine gross mechanical power, generator active output and generator apparent output remain separate. Frequency is part of the immutable configuration identity; shared portfolio claims and possible alternator choices are not expanded into fabricated variants.
- Parameters:
content (bytes)
source_url (str)
expected_count (int)
- Return type:
dict[str, Any]
- energy_data_retrieval.data_center_catalogue.inventories.parse_volvo_penta_powergen_matrix(content, *, source_url, expected_count)[source]#
Stage exact Volvo Penta engine/frequency power-generation rows.
- Parameters:
content (bytes)
source_url (str)
expected_count (int)
- Return type:
dict[str, Any]
- energy_data_retrieval.data_center_catalogue.inventories.parse_supermicro_nvidia_systems(content, *, source_url, expected_count)[source]#
Stage exact current Supermicro HGX B200/B300 system rows.
The official NVIDIA-certified systems table may list one system more than once under different certification states. Those rows collapse only when every hardware field agrees, while all distinct certification states are retained. Per-GPU power and memory never become server or rack totals.
- Parameters:
content (bytes)
source_url (str)
expected_count (int)
- Return type:
dict[str, Any]
- energy_data_retrieval.data_center_catalogue.inventories.parse_lenovo_gpu_comparison(content, *, source_url, expected_count)[source]#
Stage Lenovo’s current GPU-option table from its embedded JSON payload.
The comparison page contains withdrawn and available products together. Only rows explicitly marked
Availableare staged. Board GPU count and per-GPU TDP remain separate; this adapter never multiplies them into server or rack input power. A staged row still requires field-level evidence review before canonical publication.- Parameters:
content (bytes)
source_url (str)
expected_count (int)
- Return type:
dict[str, Any]