feat(preprocess): publish incremental Evidence corpus

2026-07-12 04:26:24 +02:00
parent 16a8bd9df6
commit b96d4f13b9
10 changed files with 708 additions and 0 deletions
@@ -0,0 +1,31 @@
+"""Runtime Evidence lookup bound to the atomically active corpus generation."""
+
+from tht.corpus.store import CorpusStore
+
+
+def active_evidence_hits(store: CorpusStore, vector_store, embedding, *, limit: int):
+    manifest = store.active_manifest()
+    if manifest is None or manifest.vector_generation is None:
+        return []
+    document_generations = dict(manifest.metadata.get("document_generations", {}))
+    active_documents = {document.document_id for document in manifest.documents}
+    hits = vector_store.search(["evidence"], embedding, limit=limit, kinds=["evidence"])
+    return [
+        hit for hit in hits
+        if hit.metadata.get("document_id") in active_documents
+        and hit.metadata.get("vector_generation")
+        == document_generations.get(hit.metadata.get("document_id"), manifest.vector_generation)
+    ]
+
+
+def resolve_evidence_file(store: CorpusStore, evidence_id: str) -> str:
+    manifest = store.active_manifest()
+    if manifest is None:
+        return ""
+    for document in manifest.documents:
+        frontmatter = document.metadata.get("frontmatter", {})
+        identifiers = {document.document_id, document.source_id, str(frontmatter.get("id", ""))}
+        if evidence_id in identifiers:
+            path = store.resolve_document(document.document_id)
+            return str(path) if path else ""
+    return ""