feat(storage): index exact content references

This commit is contained in:
DCCONSTRUCTIONS
2026-07-29 15:21:03 +03:00
parent 611853d3af
commit 32c5d0dda5
13 changed files with 1081 additions and 32 deletions
+19 -2
View File
@@ -6,6 +6,9 @@ from pathlib import Path
from k1link.compute.e44_data_amplification_audit import (
build_e44_data_amplification_audit,
)
from k1link.compute.e50_content_reference_index import (
build_e50_content_reference_index,
)
from k1link.web.artifact_health_api import ArtifactHealthService
@@ -23,14 +26,21 @@ def test_artifact_health_reports_live_inventory_and_exact_audit(tmp_path: Path)
_write(second / "shared.bin", b"shared")
_write(second / "only-b.bin", b"bb")
e44_results = runtime / "compute-experiments" / "e44" / "results"
build_e44_data_amplification_audit(
e44 = build_e44_data_amplification_audit(
artifact_roots={"e30": first, "e40": second},
output_root=e44_results,
)
e50_results = runtime / "compute-experiments" / "e50" / "results"
e50 = build_e50_content_reference_index(
e44_result_root=e44.result_root,
artifact_roots={"e30": first, "e40": second},
output_root=e50_results,
)
service = ArtifactHealthService(
lambda: runtime,
lambda: e44_results,
e50_results_root_provider=lambda: e50_results,
cache_seconds=0,
)
document = service.snapshot()
@@ -40,7 +50,13 @@ def test_artifact_health_reports_live_inventory_and_exact_audit(tmp_path: Path)
assert document.inventory.logical_bytes > 0
assert document.exact_audit.state == "exact-current"
assert document.exact_audit.duplicate_bytes == len(b"shared")
assert document.content_references.state == "not-indexed"
assert document.content_references.state == "indexed"
assert document.content_references.result_id == e50.result_id
assert document.content_references.logical_reference_count == 4
assert document.content_references.canonical_content_count == 3
assert document.content_references.duplicate_reference_count == 1
assert document.content_references.exact_duplicate_bytes == len(b"shared")
assert document.content_references.physical_reclamation_applied is False
assert document.content_references.storage_migration_authorized is False
@@ -81,3 +97,4 @@ def test_artifact_health_reports_unavailable_runtime(tmp_path: Path) -> None:
assert document.overall_state == "unavailable"
assert document.inventory.state == "unavailable"
assert document.exact_audit.state == "unavailable"
assert document.content_references.state == "not-indexed"