InventoryResult

Section 8 · From docs/INVENTORY_REFERENCE.md on GitHub

cloudg.inventory.InventoryResult (defined in cloudg/inventory/mapper_result.py, re-exported from cloudg.inventory.mapper and the package root). A dataclass.

8.1 Fields

Field	Type	Meaning
assets	list[CloudAsset]	Every asset after deduplication and linking, including organization / hierarchy assets, external account placeholders and the account nodes added by add_account_hierarchy.
edges	list[NetworkEdge]	Collector edges, then linker edges, then hierarchy edges. The linker adds no edge whose source, target and edge_type an earlier edge already has. Parallel collector rules between the same two endpoints (one security group rule per port, say) are kept as separate edges.
coverage	list[CollectionCoverage]	One record per discovery step and per provider × account × region collection run (section 10). Exported under coverage in inventory-map.json.
providers	list[str]	The configured providers, lower case (["aws", "gcp"]).
regions	dict[str, list[str]]	Regions actually collected per provider after all expansion, e.g. {"aws": ["us-east-1", "eu-west-1"]}.
duration_ms	int	Wall-clock time of map_inventory(). 0 after load().
organization	dict \| None	OrganizationTopology.to_dict() when an AWS Organization was mapped, else None (section 17).
unresolved_references	list[dict]	Declared relations whose target resolved to nothing, at most 2000 (section 11).
throttling	dict \| None	The run's throttling summary (totals, messages, skipped, scopes, breakers, slowed_buckets), or None when no cloud API pushed back. The keys are described in the rate limits guide.

8.2 Methods and properties

Member	Returns	Notes
summary (property)	dict[str, Any]	Recomputed on every access. Section 9.
dependency_graph(include_hierarchy=False)	DependencyGraph	A new graph each call. Section 12.
analysis(top=25)	dict[str, Any]	Section 13.
export(output_dir)	dict[str, Path]	8.3.
load(path) (classmethod)	InventoryResult	8.4.

8.3 export

export(output_dir) creates the directory and writes:

Return key	File	Always
map	inventory-map.json	yes
graphml	inventory-map.graphml	yes
graph	inventory-graph.json	yes
dependencies	inventory-dependencies.json	yes
organization	inventory-organization.json	only when organization is set

The values are pathlib.Path objects. JSON files are written with indent=2 and default=str (datetimes and other non-JSON values become strings).

8.4 load

InventoryResult.load(path) accepts inventory-map.json or the directory holding it.

• assets and edges are re-validated into models (display_id is dropped before validation).
• providers comes from the file's providers, or summary.providers for older files.
• regions, unresolved_references and throttling are read back.
• organization is read from inventory-organization.json next to the map, when present.
• coverage is read back from the file's coverage list; a map written before that key existed loads with an empty list. duration_ms is 0: it is not stored in the files.
• raw_data is empty on every asset (it is never serialised). Linking rules that use raw payloads (instance profiles, Lambda roles from Role) therefore only work on fresh collections, not on loaded maps; the declared relations do work.

---