refactor(storage): DRY DocumentRepository

Author: alpeerkaraca

src/everspring_mcp/storage/repository.py

@@ -180,6 +180,27 @@ def __init__(self, db: aiosqlite.Connection) -> None:
         """
         self.db = db
 
+    def _to_row_tuple(self, doc: DocumentRecord) -> tuple[Any, ...]:
+        """Convert a DocumentRecord to a tuple for sqlite insertion."""
+        return (
+            doc.id,
+            doc.url,
+            doc.title,
+            doc.module,
+            doc.submodule,
+            doc.major_version,
+            doc.minor_version,
+            doc.patch_version,
+            doc.content_hash,
+            doc.file_path,
+            doc.s3_key,
+            doc.size_bytes,
+            doc.scraped_at.isoformat(),
+            doc.synced_at.isoformat() if doc.synced_at else None,
+            doc.schema_version,
+            int(doc.is_indexed),
+        )
+
     async def insert(self, doc: DocumentRecord) -> None:
         """Insert a new document.
 
@@ -197,24 +218,7 @@ async def insert(self, doc: DocumentRecord) -> None:
                 schema_version, is_indexed
             ) VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?)
             """,
-            (
-                doc.id,
-                doc.url,
-                doc.title,
-                doc.module,
-                doc.submodule,
-                doc.major_version,
-                doc.minor_version,
-                doc.patch_version,
-                doc.content_hash,
-                doc.file_path,
-                doc.s3_key,
-                doc.size_bytes,
-                doc.scraped_at.isoformat(),
-                doc.synced_at.isoformat() if doc.synced_at else None,
-                doc.schema_version,
-                int(doc.is_indexed),
-            ),
+            self._to_row_tuple(doc),
         )
         await self.db.commit()
         logger.debug("Inserted document: %s", doc.id)
@@ -246,24 +250,7 @@ async def upsert(self, doc: DocumentRecord) -> None:
                     ELSE documents.is_indexed 
                 END
             """,
-            (
-                doc.id,
-                doc.url,
-                doc.title,
-                doc.module,
-                doc.submodule,
-                doc.major_version,
-                doc.minor_version,
-                doc.patch_version,
-                doc.content_hash,
-                doc.file_path,
-                doc.s3_key,
-                doc.size_bytes,
-                doc.scraped_at.isoformat(),
-                doc.synced_at.isoformat() if doc.synced_at else None,
-                doc.schema_version,
-                int(doc.is_indexed),
-            ),
+            self._to_row_tuple(doc),
         )
         await self.db.commit()
         logger.debug("Upserted document: %s", doc.id)

src/everspring_mcp/vector/embeddings.py

@@ -131,12 +131,8 @@ async def embed(self, texts: list[str], batch_size: int) -> list[dict[str, Any]]
         ]
 
 
-class BGESlimStrategy(EmbeddingStrategy):
-    """Concrete strategy for the BGE-Base (Slim) model."""
-
-    @property
-    def tier_name(self) -> str:
-        return SLIM_TIER
+class DenseEmbeddingStrategy(EmbeddingStrategy):
+    """Base strategy for models that only return dense embeddings."""
 
     async def embed(self, texts: list[str], batch_size: int) -> list[dict[str, Any]]:
         model = await asyncio.to_thread(self._ensure_loaded)
@@ -153,27 +149,21 @@ async def embed(self, texts: list[str], batch_size: int) -> list[dict[str, Any]]
         ]
 
 
-class BGEXSlimStrategy(EmbeddingStrategy):
+class BGESlimStrategy(DenseEmbeddingStrategy):
+    """Concrete strategy for the BGE-Base (Slim) model."""
+
+    @property
+    def tier_name(self) -> str:
+        return SLIM_TIER
+
+
+class BGEXSlimStrategy(DenseEmbeddingStrategy):
     """Concrete strategy for the BGE-Small (X-Slim) model."""
 
     @property
     def tier_name(self) -> str:
         return XSLIM_TIER
 
-    async def embed(self, texts: list[str], batch_size: int) -> list[dict[str, Any]]:
-        model = await asyncio.to_thread(self._ensure_loaded)
-        embeddings = await asyncio.to_thread(
-            model.encode,
-            texts,
-            batch_size=batch_size,
-            convert_to_numpy=True,
-            show_progress_bar=False,
-            normalize_embeddings=True,
-        )
-        return [
-            {"dense": emb.tolist(), "sparse": None} for emb in np.asarray(embeddings)
-        ]
-
 
 class StrategyFactory:
     """Factory for creating embedding strategies based on tier."""
@@ -240,6 +230,7 @@ async def prefetch_model(self) -> None:
 
 __all__ = [
     "EmbeddingStrategy",
+    "DenseEmbeddingStrategy",
     "BGEM3Strategy",
     "BGESlimStrategy",
     "BGEXSlimStrategy",

tests/test_vector_indexer.py

@@ -190,16 +190,17 @@ async def mark_indexed(self, doc_ids: list[str]) -> int:
             content="c1",
             metadata={"module": "spring-boot"},
             embedding=[0.1, 0.2],
+            sparse_weights=None,
         ),
         types.SimpleNamespace(
             chunk_id="doc1-1",
             document_id="doc1",
             content="c2",
             metadata={"module": "spring-boot"},
             embedding=[0.3, 0.4],
+            sparse_weights=None,
         ),
     ]
-
     flushed_chunks, flushed_docs = await VectorIndexer._flush_vector_payloads(
         indexer,
         payloads=payloads,