fix(evidence): index only curated legacy documents
This commit is contained in:
@@ -117,6 +117,25 @@ def test_source_factory_preserves_legacy_first_order_and_filesystem_configuratio
|
|||||||
assert current[1].max_bytes == 1024
|
assert current[1].max_bytes == 1024
|
||||||
|
|
||||||
|
|
||||||
|
def test_legacy_source_discovers_only_curated_evidence_units(tmp_path):
|
||||||
|
legacy_root = tmp_path / "legacy"
|
||||||
|
(legacy_root / "evidence" / "source").mkdir(parents=True)
|
||||||
|
(legacy_root / "evidence" / "source" / "raw.md").write_text("raw source")
|
||||||
|
(legacy_root / "evidence" / "curated" / "domain").mkdir(parents=True)
|
||||||
|
(legacy_root / "evidence" / "curated" / "domain" / "patient.md").write_text("curated")
|
||||||
|
cfg = SimpleNamespace(evidence=SimpleNamespace(
|
||||||
|
source_root=legacy_root,
|
||||||
|
evidence_dir="evidence",
|
||||||
|
sources=[],
|
||||||
|
))
|
||||||
|
|
||||||
|
source = build_sources(cfg.evidence)[0]
|
||||||
|
|
||||||
|
assert [item.metadata["relative_path"] for item in source.discover()] == [
|
||||||
|
"curated/domain/patient.md"
|
||||||
|
]
|
||||||
|
|
||||||
|
|
||||||
def test_preprocessing_factory_forwards_only_evidence_pipeline_dependencies(monkeypatch):
|
def test_preprocessing_factory_forwards_only_evidence_pipeline_dependencies(monkeypatch):
|
||||||
captured = {}
|
captured = {}
|
||||||
|
|
||||||
|
|||||||
@@ -18,7 +18,11 @@ def build_sources(evidence: EvidenceSourcesConfig | None) -> list[EvidenceSource
|
|||||||
return []
|
return []
|
||||||
sources: list[EvidenceSource] = []
|
sources: list[EvidenceSource] = []
|
||||||
if evidence.source_root is not None:
|
if evidence.source_root is not None:
|
||||||
sources.append(FilesystemEvidenceSource(evidence.source_root / evidence.evidence_dir))
|
legacy_root = evidence.source_root / evidence.evidence_dir
|
||||||
|
legacy_patterns = (
|
||||||
|
("**/*.md",) if legacy_root.name == "curated" else ("curated/**/*.md",)
|
||||||
|
)
|
||||||
|
sources.append(FilesystemEvidenceSource(legacy_root, patterns=legacy_patterns))
|
||||||
for resource in evidence.sources:
|
for resource in evidence.sources:
|
||||||
match resource.type:
|
match resource.type:
|
||||||
case "filesystem":
|
case "filesystem":
|
||||||
|
|||||||
Reference in New Issue
Block a user