Agent Zero AI framework
Fix document query index reuse
Alessandro94065dbe8602
diff --git a/plugins/_document_query/helpers/document_query.py b/plugins/_document_query/helpers/document_query.pyindex 4a7fa897a..05dd112e4 100644--- a/plugins/_document_query/helpers/document_query.py+++ b/plugins/_document_query/helpers/document_query.py@@ -72,15 +72,30 @@ def _load_config(agent: Agent) -> dict:class DocumentQueryStore:"""FAISS Store for document query results."""+ CONTEXT_DATA_KEY = "_document_query_store"DEFAULT_CHUNK_SIZE = 1000DEFAULT_CHUNK_OVERLAP = 100DEFAULT_MAX_INDEX_CHUNKS = 1200+ _GET_LOCK = threading.RLock()- @staticmethod- def get(agent: Agent):+ @classmethod+ def get(cls, agent: Agent):if not agent or not agent.config:raise ValueError("Agent and agent config must be provided")- return DocumentQueryStore(agent)++ context = getattr(agent, "context", None)+ if context is None:+ return cls(agent)++ with cls._GET_LOCK:+ store = context.get_data(cls.CONTEXT_DATA_KEY, recursive=False)+ if not isinstance(store, cls):+ store = cls(agent)+ context.set_data(cls.CONTEXT_DATA_KEY, store, recursive=False)+ else:+ store.agent = agent+ store.config = _load_config(agent)+ return storedef __init__(self, agent: Agent):self.agent = agentdiff --git a/tests/test_document_query_plugin.py b/tests/test_document_query_plugin.pyindex e3ece0ebd..1009edc45 100644--- a/tests/test_document_query_plugin.py+++ b/tests/test_document_query_plugin.py@@ -1,13 +1,19 @@from __future__ import annotationsimport asyncio+import sysfrom pathlib import Pathimport pytestfrom PIL import Image+ROOT = Path(__file__).resolve().parents[1]+if str(ROOT) not in sys.path:+ sys.path.insert(0, str(ROOT))+from plugins._document_query import hooks as document_query_hooksfrom plugins._document_query.helpers.fetch import FetchedDocument, fetch_public_resource+import plugins._document_query.helpers.document_query as document_query_modulefrom plugins._document_query.helpers.document_query import (DocumentQueryHelper,DocumentQueryStore,@@ -19,9 +25,6 @@ from plugins._document_query.helpers.parsers.liteparse import LiteParseParserfrom plugins._document_query.helpers.parsers.text import TextParser-ROOT = Path(__file__).resolve().parents[1]--def run_async(coro):with asyncio.Runner() as runner:return runner.run(coro)@@ -52,6 +55,52 @@ class CountingAsyncParser(BaseParser):return document.uri+class _StoreContext:+ def __init__(self, context_id: str):+ self.id = context_id+ self.data = {}++ def get_data(self, key: str, recursive: bool = True):+ return self.data.get(key)++ def set_data(self, key: str, value, recursive: bool = True):+ self.data[key] = value+++class _StoreAgent:+ def __init__(self, context_id: str):+ self.config = object()+ self.context = _StoreContext(context_id)+++class _FakeVectorDB:+ def __init__(self):+ self.docs = []++ async def insert_documents(self, docs):+ ids = []+ for doc in docs:+ doc_id = f"doc-{len(self.docs)}"+ doc.metadata["id"] = doc_id+ ids.append(doc_id)+ self.docs.append(doc)+ return ids++ async def search_by_metadata(self, filter: str, limit: int = 0):+ document_uri = filter.split("'", 2)[1]+ docs = [+ doc+ for doc in self.docs+ if doc.metadata.get("document_uri") == document_uri+ ]+ return docs[:limit] if limit > 0 else docs++ async def delete_documents_by_ids(self, ids: list[str]):+ removed = [doc for doc in self.docs if doc.metadata.get("id") in ids]+ self.docs = [doc for doc in self.docs if doc.metadata.get("id") not in ids]+ return removed++def test_fetch_file_detects_mimetype_and_reads_once(tmp_path):document = tmp_path / "notes.txt"document.write_text("hello\nworld\n", encoding="utf-8")@@ -196,6 +245,41 @@ def test_document_query_allows_uncapped_index_chunks():assert len(chunks) > 10+def test_document_query_store_reuses_vector_db_per_context(monkeypatch):+ monkeypatch.setattr(+ document_query_module,+ "_load_config",+ lambda _agent: {+ "chunk_size": 100,+ "chunk_overlap": 10,+ "max_index_chunks": 20,+ },+ )+ monkeypatch.setattr(+ DocumentQueryStore,+ "init_vector_db",+ lambda _self: _FakeVectorDB(),+ )++ agent = _StoreAgent("ctx-one")+ store = DocumentQueryStore.get(agent)++ success, ids = run_async(+ store.add_document("alpha beta gamma " * 20, "/tmp/book.txt")+ )+ second_store = DocumentQueryStore.get(agent)++ assert success is True+ assert ids+ assert second_store is store+ assert second_store.vector_db is store.vector_db+ assert run_async(second_store.document_exists("/tmp/book.txt")) is True++ isolated_store = DocumentQueryStore.get(_StoreAgent("ctx-two"))+ assert isolated_store is not store+ assert run_async(isolated_store.document_exists("/tmp/book.txt")) is False++def test_document_query_thumbnail_matches_plugin_hub_limits():thumbnail = ROOT / "plugins" / "_document_query" / "webui" / "thumbnail.jpg"