test: Pinecone - relax flaky test (#2360)

* test: Pinecone - relax flaky test

* Trigger Build

Author: anakin87

integrations/pinecone/tests/test_document_store.py

@@ -276,7 +276,7 @@ def test_embedding_retrieval(self, document_store: PineconeDocumentStore):
         query_embedding = [0.1] * 768
         most_similar_embedding = [0.8] * 768
         second_best_embedding = [0.8] * 700 + [0.1] * 3 + [0.2] * 65
-        another_embedding = np.random.rand(768).tolist()
+        another_embedding = [0.1] * 384 + [-0.1] * 384
 
         docs = [
             Document(content="Most similar document", embedding=most_similar_embedding),
@@ -287,9 +287,11 @@ def test_embedding_retrieval(self, document_store: PineconeDocumentStore):
         document_store.write_documents(docs)
 
         results = document_store._embedding_retrieval(query_embedding=query_embedding, top_k=2, filters={})
+
         assert len(results) == 2
-        assert results[0].content == "Most similar document"
-        assert results[1].content == "2nd best document"
+        # Pinecone does not seem to guarantee the order of the results
+        assert "Most similar document" in [result.content for result in results]
+        assert "2nd best document" in [result.content for result in results]
 
     def test_close(self, document_store: PineconeDocumentStore):
         document_store._initialize_index()

integrations/pinecone/tests/test_document_store_async.py

@@ -70,7 +70,7 @@ async def test_embedding_retrieval(self, document_store_async: PineconeDocumentS
         query_embedding = [0.1] * 768
         most_similar_embedding = [0.8] * 768
         second_best_embedding = [0.8] * 700 + [0.1] * 3 + [0.2] * 65
-        another_embedding = np.random.rand(768).tolist()
+        another_embedding = [-0.1] * 384 + [0.1] * 384
 
         docs = [
             Document(content="Most similar document", embedding=most_similar_embedding),
@@ -83,9 +83,12 @@ async def test_embedding_retrieval(self, document_store_async: PineconeDocumentS
         results = await document_store_async._embedding_retrieval_async(
             query_embedding=query_embedding, top_k=2, filters={}
         )
+
         assert len(results) == 2
-        assert results[0].content == "Most similar document"
-        assert results[1].content == "2nd best document"
+
+        # Pinecone does not seem to guarantee the order of the results
+        assert "Most similar document" in [result.content for result in results]
+        assert "2nd best document" in [result.content for result in results]
 
     async def test_close(self, document_store_async: PineconeDocumentStore):
         await document_store_async._initialize_async_index()