FastembedRanker: remove the backend as suggested

integrations/fastembed/src/haystack_integrations/components/rankers/fastembed/ranker.py

@@ -2,7 +2,7 @@
 
 from haystack import Document, component, default_from_dict, default_to_dict, logging
 
-from .ranker_backend.fastembed_backend import _FastembedRankerBackendFactory
+from fastembed.rerank.cross_encoder import TextCrossEncoder
 
 logger = logging.getLogger(__name__)
 
@@ -106,8 +106,8 @@ def warm_up(self):
         """
         Initializes the component.
         """
-        if not hasattr(self, "ranker_backend"):
-            self.ranker_backend = _FastembedRankerBackendFactory.get_ranker_backend(
+        if not hasattr(self, "ranker"):
+            self.ranker = TextCrossEncoder(
                 model_name=self.model_name,
                 cache_dir=self.cache_dir,
                 threads=self.threads,
@@ -164,14 +164,14 @@ def run(self, query: str, documents: List[Document], top_k: Optional[int] = None
             msg = f"top_k must be > 0, but got {top_k}"
             raise ValueError(msg)
 
-        if not hasattr(self, "ranker_backend"):
+        if not hasattr(self, "ranker"):
             msg = "The ranker model has not been loaded. Please call warm_up() before running."
             raise RuntimeError(msg)
 
         fastembed_input_docs = self._prepare_fastembed_input_docs(documents)
 
         scores = list(
-            self.ranker_backend.rerank(
+            self.ranker.rerank(
                 query=query,
                 documents=fastembed_input_docs,
                 batch_size=self.batch_size,

integrations/fastembed/tests/test_fastembed_ranker.py

@@ -159,29 +159,6 @@ def test_from_dict_with_custom_init_parameters(self):
         assert ranker.meta_fields_to_embed == ["test_field"]
         assert ranker.meta_data_separator == " | "
 
-    @patch("haystack_integrations.components.rankers.fastembed.ranker._FastembedRankerBackendFactory")
-    def test_warmup(self, mocked_factory):
-        """
-        Test for checking ranker instances after warm-up.
-        """
-        ranker = FastembedRanker(model_name="BAAI/bge-reranker-base")
-        mocked_factory.get_ranker_backend.assert_not_called()
-        ranker.warm_up()
-        mocked_factory.get_ranker_backend.assert_called_once_with(
-            model_name="BAAI/bge-reranker-base", cache_dir=None, threads=None, local_files_only=False
-        )
-
-    @patch("haystack_integrations.components.rankers.fastembed.ranker._FastembedRankerBackendFactory")
-    def test_warmup_does_not_reload(self, mocked_factory):
-        """
-        Test for checking backend instances after multiple warm-ups.
-        """
-        ranker = FastembedRanker(model_name="Xenova/ms-marco-MiniLM-L-12-v2")
-        mocked_factory.get_ranker_backend.assert_not_called()
-        ranker.warm_up()
-        ranker.warm_up()
-        mocked_factory.get_ranker_backend.assert_called_once()
-
     def test_embed_incorrect_input_format(self):
         """
         Test for checking incorrect input format when creating embedding.
@@ -214,13 +191,13 @@ def test_embed_metadata(self):
             model_name="model_name",
             meta_fields_to_embed=["meta_field"],
         )
-        ranker.ranker_backend = MagicMock()
+        ranker.ranker = MagicMock()
 
         documents = [Document(content=f"document-number {i}", meta={"meta_field": f"meta_value {i}"}) for i in range(5)]
         query = "test"
         ranker.run(query=query, documents=documents)
 
-        ranker.ranker_backend.rerank.assert_called_once_with(
+        ranker.ranker.rerank.assert_called_once_with(
             query=query,
             documents=[
                 "meta_value 0\ndocument-number 0",