Rebuild precomputed retrieval without vector indexes

.gitignore

@@ -13,4 +13,9 @@ wheels/
 .env
 
 # DS_Store files
-.DS_Store
+.DS_Store
+
+# Precomputed example retrieval indexes are rebuilt from chunks.json
+assets/example_bills/**/vector_store/
+*.faiss
+*.pkl

scripts/generate_precomputed_examples.py

@@ -23,13 +23,7 @@ from services.example_bills import EXAMPLE_BILL_MANIFEST_PATH, ExampleBill, load
 from services.ingest import fetch_url_content
 from services.precomputed_assets import write_precomputed_asset
 from services.providers import get_default_api_key, instantiate_client
-from services.rag_pipeline import (
-    build_vector_store,
-    document_hash_for_text,
-    generate_analysis_once,
-    save_vector_store,
-    split_into_chunks,
-)
+from services.rag_pipeline import document_hash_for_text, generate_analysis_once, split_into_chunks
 
 
 def _parse_args() -> argparse.Namespace:
@@ -68,7 +62,6 @@ def _has_complete_assets(bill: ExampleBill) -> bool:
         and bill.document_path.exists()
         and bill.chunks_path.exists()
         and bill.metadata_path.exists()
-        and bill.vector_store_dir.exists()
     )
 
 
@@ -113,8 +106,6 @@ def main() -> int:
             document_text = fetch_url_content(bill.source_url, timeout_seconds=args.fetch_timeout)
             chunks = split_into_chunks(document_text)
             analysis = generate_analysis_once(provider_client, document_text)
-            vector_store = build_vector_store(chunks)
-            save_vector_store(vector_store, bill.vector_store_dir)
 
             document_hash = document_hash_for_text(document_text)
             write_precomputed_asset(

services/precomputed_assets.py

@@ -33,8 +33,6 @@ def load_precomputed_asset(bill: ExampleBill) -> PrecomputedExampleAsset | None:
         return None
     if not bill.chunks_path.exists():
         return None
-    if not bill.vector_store_dir.exists():
-        return None
 
     analysis_payload = json.loads(bill.analysis_path.read_text(encoding="utf-8"))
     document_text = bill.document_path.read_text(encoding="utf-8")

services/rag_pipeline.py

@@ -263,7 +263,7 @@ def load_vector_store(vector_store_dir: Path) -> FAISS:
 
 def hydrate_precomputed_example(asset: PrecomputedExampleAsset) -> CachedDocumentArtifacts:
     analysis = AnalysisResult.model_validate(asset.analysis_payload)
-    vector_store = load_vector_store(asset.bill.vector_store_dir)
+    vector_store = build_vector_store(asset.chunks)
     artifacts = CachedDocumentArtifacts(
         document_hash=asset.bill.document_hash or document_hash_for_text(asset.document_text),
         document_text=asset.document_text,

tests/test_precomputed_assets.py

@@ -0,0 +1,62 @@
+from __future__ import annotations
+
+import json
+
+from services.example_bills import ExampleBill
+from services.precomputed_assets import load_precomputed_asset
+from services.rag_pipeline import hydrate_precomputed_example
+
+
+def test_load_precomputed_asset_does_not_require_vector_store_dir(monkeypatch, tmp_path) -> None:
+    from services import example_bills as example_bills_module
+
+    monkeypatch.setattr(example_bills_module, "EXAMPLE_BILL_ASSETS_ROOT", tmp_path)
+    bill = ExampleBill(
+        id="example-bill",
+        title="Example Bill",
+        source_url="https://example.com/bill.pdf",
+        document_hash="hash",
+        asset_dir="example-bill",
+    )
+    root_dir = tmp_path / bill.asset_dir
+    root_dir.mkdir(parents=True)
+    (root_dir / "analysis.json").write_text(json.dumps({"executive_summary": "Summary"}), encoding="utf-8")
+    (root_dir / "document.txt").write_text("Document text", encoding="utf-8")
+    (root_dir / "chunks.json").write_text(json.dumps(["Chunk one", "Chunk two"]), encoding="utf-8")
+
+    asset = load_precomputed_asset(bill)
+
+    assert asset is not None
+    assert asset.document_text == "Document text"
+    assert asset.chunks == ["Chunk one", "Chunk two"]
+
+
+def test_hydrate_precomputed_example_rebuilds_vector_store_from_chunks(monkeypatch, tmp_path) -> None:
+    from services import example_bills as example_bills_module
+
+    monkeypatch.setattr(example_bills_module, "EXAMPLE_BILL_ASSETS_ROOT", tmp_path)
+    bill = ExampleBill(
+        id="example-bill",
+        title="Example Bill",
+        source_url="https://example.com/bill.pdf",
+        document_hash="hash",
+        asset_dir="example-bill",
+    )
+    root_dir = tmp_path / bill.asset_dir
+    root_dir.mkdir(parents=True)
+    (root_dir / "analysis.json").write_text(json.dumps({"executive_summary": "Summary"}), encoding="utf-8")
+    (root_dir / "document.txt").write_text("Document text", encoding="utf-8")
+    (root_dir / "chunks.json").write_text(json.dumps(["Chunk one", "Chunk two"]), encoding="utf-8")
+
+    asset = load_precomputed_asset(bill)
+    assert asset is not None
+
+    built_chunks: list[list[str]] = []
+
+    monkeypatch.setattr("services.rag_pipeline.build_vector_store", lambda chunks: built_chunks.append(list(chunks)) or object())
+
+    artifacts = hydrate_precomputed_example(asset)
+
+    assert built_chunks == [["Chunk one", "Chunk two"]]
+    assert artifacts.vector_store is not None
+    assert artifacts.document_text == "Document text"