| 336 | return documents |
| 337 | |
| 338 | def add_documents(self, documents: Sequence[Document]) -> None: |
| 339 | super().maybe_add_ids(documents) |
| 340 | for doc in documents: |
| 341 | doc.metadata.id = str(PostgresDB._id_to_uuid(doc.metadata.id, doc.metadata)) |
| 342 | |
| 343 | embeddings = self.embedding_fn([doc.content for doc in documents]) |
| 344 | |
| 345 | batch_size = self.config.batch_size |
| 346 | with self.SessionLocal() as session: |
| 347 | for i in range(0, len(documents), batch_size): |
| 348 | batch_docs = documents[i : i + batch_size] |
| 349 | batch_embeddings = embeddings[i : i + batch_size] |
| 350 | |
| 351 | new_records = [ |
| 352 | { |
| 353 | "id": doc.metadata.id, |
| 354 | "embedding": embedding, |
| 355 | "document": doc.content, |
| 356 | "cmetadata": doc.metadata.model_dump(), |
| 357 | } |
| 358 | for doc, embedding in zip(batch_docs, batch_embeddings) |
| 359 | ] |
| 360 | |
| 361 | if new_records: |
| 362 | stmt = insert(self.embeddings_table).values(new_records) |
| 363 | session.execute(stmt) |
| 364 | session.commit() |
| 365 | |
| 366 | @staticmethod |
| 367 | def _id_to_uuid(id: str, obj: object) -> str: |