chore!: nvidia - drop Python 3.9 and use X|Y typing; fix default reranking model; improve tests (#2736)

anakin87 · web-flow · commit 2eb0ca8effc7 · 2026-01-13T12:42:58.000+01:00
* make test more robust

* chore!: nvidia - drop Python 3.9 and use X|Y typing

* try changing default model

* improve test

* fix fma
diff --git a/integrations/nvidia/pyproject.toml b/integrations/nvidia/pyproject.toml
@@ -7,23 +7,22 @@ name = "nvidia-haystack"
 dynamic = ["version"]
 description = ''
 readme = "README.md"
-requires-python = ">=3.9"
+requires-python = ">=3.10"
 license = "Apache-2.0"
 keywords = []
 authors = [{ name = "deepset GmbH", email = "info@deepset.ai" }]
 classifiers = [
   "License :: OSI Approved :: Apache Software License",
   "Development Status :: 4 - Beta",
   "Programming Language :: Python",
-  "Programming Language :: Python :: 3.9",
   "Programming Language :: Python :: 3.10",
   "Programming Language :: Python :: 3.11",
   "Programming Language :: Python :: 3.12",
   "Programming Language :: Python :: 3.13",
   "Programming Language :: Python :: Implementation :: CPython",
   "Programming Language :: Python :: Implementation :: PyPy",
 ]
-dependencies = ["haystack-ai>=2.19.0", "requests>=2.25.0", "tqdm>=4.21.0"]
+dependencies = ["haystack-ai>=2.22.0", "requests>=2.25.0", "tqdm>=4.21.0"]
 
 [project.urls]
 Documentation = "https://github.com/deepset-ai/haystack-core-integrations/tree/main/integrations/nvidia#readme"
@@ -82,7 +81,6 @@ disallow_incomplete_defs = true
 
 
 [tool.ruff]
-target-version = "py39"
 line-length = 120
 
 [tool.ruff.lint]
@@ -129,10 +127,6 @@ ignore = [
   "B008",
   "S101",
 ]
-unfixable = [
-  # Don't touch unused imports
-  "F401",
-]
 
 [tool.ruff.lint.isort]
 known-first-party = ["haystack_integrations"]
diff --git a/integrations/nvidia/src/haystack_integrations/components/embedders/nvidia/document_embedder.py b/integrations/nvidia/src/haystack_integrations/components/embedders/nvidia/document_embedder.py
@@ -5,7 +5,7 @@
 import os
 import warnings
 from dataclasses import replace
-from typing import Any, Optional, Union
+from typing import Any
 
 from haystack import Document, component, default_from_dict, default_to_dict, logging
 from haystack.utils import Secret, deserialize_secrets_inplace
@@ -39,17 +39,17 @@ class NvidiaDocumentEmbedder:
 
     def __init__(
         self,
-        model: Optional[str] = None,
-        api_key: Optional[Secret] = Secret.from_env_var("NVIDIA_API_KEY"),
+        model: str | None = None,
+        api_key: Secret | None = Secret.from_env_var("NVIDIA_API_KEY"),
         api_url: str = os.getenv("NVIDIA_API_URL", DEFAULT_API_URL),
         prefix: str = "",
         suffix: str = "",
         batch_size: int = 32,
         progress_bar: bool = True,
-        meta_fields_to_embed: Optional[list[str]] = None,
+        meta_fields_to_embed: list[str] | None = None,
         embedding_separator: str = "\n",
-        truncate: Optional[Union[EmbeddingTruncateMode, str]] = None,
-        timeout: Optional[float] = None,
+        truncate: EmbeddingTruncateMode | str | None = None,
+        timeout: float | None = None,
     ) -> None:
         """
         Create a NvidiaTextEmbedder component.
@@ -98,7 +98,7 @@ def __init__(
             truncate = EmbeddingTruncateMode.from_str(truncate)
         self.truncate = truncate
 
-        self.backend: Optional[Any] = None
+        self.backend: Any | None = None
         self._initialized = False
 
         if timeout is None:
@@ -235,7 +235,7 @@ def _embed_batch(self, texts_to_embed: list[str], batch_size: int) -> tuple[list
         return all_embeddings, {"usage": {"prompt_tokens": usage_prompt_tokens, "total_tokens": usage_total_tokens}}
 
     @component.output_types(documents=list[Document], meta=dict[str, Any])
-    def run(self, documents: list[Document]) -> dict[str, Union[list[Document], dict[str, Any]]]:
+    def run(self, documents: list[Document]) -> dict[str, list[Document] | dict[str, Any]]:
         """
         Embed a list of Documents.
 
@@ -268,7 +268,7 @@ def run(self, documents: list[Document]) -> dict[str, Union[list[Document], dict
         embeddings, metadata = self._embed_batch(texts_to_embed, self.batch_size)
 
         new_documents = []
-        for doc, emb in zip(documents, embeddings):
+        for doc, emb in zip(documents, embeddings, strict=True):
             new_documents.append(replace(doc, embedding=emb))
 
         return {"documents": new_documents, "meta": metadata}
diff --git a/integrations/nvidia/src/haystack_integrations/components/embedders/nvidia/text_embedder.py b/integrations/nvidia/src/haystack_integrations/components/embedders/nvidia/text_embedder.py
@@ -4,7 +4,7 @@
 
 import os
 import warnings
-from typing import Any, Optional, Union
+from typing import Any
 
 from haystack import component, default_from_dict, default_to_dict, logging
 from haystack.utils import Secret, deserialize_secrets_inplace
@@ -39,13 +39,13 @@ class NvidiaTextEmbedder:
 
     def __init__(
         self,
-        model: Optional[str] = None,
-        api_key: Optional[Secret] = Secret.from_env_var("NVIDIA_API_KEY"),
+        model: str | None = None,
+        api_key: Secret | None = Secret.from_env_var("NVIDIA_API_KEY"),
         api_url: str = os.getenv("NVIDIA_API_URL", DEFAULT_API_URL),
         prefix: str = "",
         suffix: str = "",
-        truncate: Optional[Union[EmbeddingTruncateMode, str]] = None,
-        timeout: Optional[float] = None,
+        truncate: EmbeddingTruncateMode | str | None = None,
+        timeout: float | None = None,
     ):
         """
         Create a NvidiaTextEmbedder component.
@@ -81,7 +81,7 @@ def __init__(
             truncate = EmbeddingTruncateMode.from_str(truncate)
         self.truncate = truncate
 
-        self.backend: Optional[Any] = None
+        self.backend: Any | None = None
         self._initialized = False
 
         if timeout is None:
@@ -187,7 +187,7 @@ def from_dict(cls, data: dict[str, Any]) -> "NvidiaTextEmbedder":
         return default_from_dict(cls, data)
 
     @component.output_types(embedding=list[float], meta=dict[str, Any])
-    def run(self, text: str) -> dict[str, Union[list[float], dict[str, Any]]]:
+    def run(self, text: str) -> dict[str, list[float] | dict[str, Any]]:
         """
         Embed a string.
 
diff --git a/integrations/nvidia/src/haystack_integrations/components/generators/nvidia/chat/chat_generator.py b/integrations/nvidia/src/haystack_integrations/components/generators/nvidia/chat/chat_generator.py
@@ -3,7 +3,7 @@
 # SPDX-License-Identifier: Apache-2.0
 
 import os
-from typing import Any, Optional
+from typing import Any
 
 from haystack import component, default_to_dict, logging
 from haystack.components.generators.chat import OpenAIChatGenerator
@@ -53,13 +53,13 @@ def __init__(
         *,
         api_key: Secret = Secret.from_env_var("NVIDIA_API_KEY"),
         model: str = "meta/llama-3.1-8b-instruct",
-        streaming_callback: Optional[StreamingCallbackT] = None,
-        api_base_url: Optional[str] = os.getenv("NVIDIA_API_URL", DEFAULT_API_URL),
-        generation_kwargs: Optional[dict[str, Any]] = None,
-        tools: Optional[ToolsType] = None,
-        timeout: Optional[float] = None,
-        max_retries: Optional[int] = None,
-        http_client_kwargs: Optional[dict[str, Any]] = None,
+        streaming_callback: StreamingCallbackT | None = None,
+        api_base_url: str | None = os.getenv("NVIDIA_API_URL", DEFAULT_API_URL),
+        generation_kwargs: dict[str, Any] | None = None,
+        tools: ToolsType | None = None,
+        timeout: float | None = None,
+        max_retries: int | None = None,
+        http_client_kwargs: dict[str, Any] | None = None,
     ) -> None:
         """
         Creates an instance of NvidiaChatGenerator.
diff --git a/integrations/nvidia/src/haystack_integrations/components/generators/nvidia/generator.py b/integrations/nvidia/src/haystack_integrations/components/generators/nvidia/generator.py
@@ -4,7 +4,7 @@
 
 import os
 import warnings
-from typing import Any, Optional, Union
+from typing import Any
 
 from haystack import component, default_from_dict, default_to_dict
 from haystack.utils.auth import Secret, deserialize_secrets_inplace
@@ -44,11 +44,11 @@ class NvidiaGenerator:
 
     def __init__(
         self,
-        model: Optional[str] = None,
+        model: str | None = None,
         api_url: str = os.getenv("NVIDIA_API_URL", DEFAULT_API_URL),
-        api_key: Optional[Secret] = Secret.from_env_var("NVIDIA_API_KEY"),
-        model_arguments: Optional[dict[str, Any]] = None,
-        timeout: Optional[float] = None,
+        api_key: Secret | None = Secret.from_env_var("NVIDIA_API_KEY"),
+        model_arguments: dict[str, Any] | None = None,
+        timeout: float | None = None,
     ) -> None:
         """
         Create a NvidiaGenerator component.
@@ -79,7 +79,7 @@ def __init__(
         self._api_key = api_key
         self._model_arguments = model_arguments or {}
 
-        self.backend: Optional[Any] = None
+        self.backend: Any | None = None
 
         self.is_hosted = is_hosted(api_url)
         if timeout is None:
@@ -171,7 +171,7 @@ def from_dict(cls, data: dict[str, Any]) -> "NvidiaGenerator":
         return default_from_dict(cls, data)
 
     @component.output_types(replies=list[str], meta=list[dict[str, Any]])
-    def run(self, prompt: str) -> dict[str, Union[list[str], list[dict[str, Any]]]]:
+    def run(self, prompt: str) -> dict[str, list[str] | list[dict[str, Any]]]:
         """
         Queries the model with the provided prompt.
 
diff --git a/integrations/nvidia/src/haystack_integrations/components/rankers/nvidia/ranker.py b/integrations/nvidia/src/haystack_integrations/components/rankers/nvidia/ranker.py
@@ -4,7 +4,7 @@
 
 import os
 import warnings
-from typing import Any, Optional, Union
+from typing import Any
 
 from haystack import Document, component, default_from_dict, default_to_dict, logging
 from haystack.utils import Secret, deserialize_secrets_inplace
@@ -47,16 +47,16 @@ class NvidiaRanker:
 
     def __init__(
         self,
-        model: Optional[str] = None,
-        truncate: Optional[Union[RankerTruncateMode, str]] = None,
+        model: str | None = None,
+        truncate: RankerTruncateMode | str | None = None,
         api_url: str = os.getenv("NVIDIA_API_URL", DEFAULT_API_URL),
-        api_key: Optional[Secret] = Secret.from_env_var("NVIDIA_API_KEY"),
+        api_key: Secret | None = Secret.from_env_var("NVIDIA_API_KEY"),
         top_k: int = 5,
         query_prefix: str = "",
         document_prefix: str = "",
-        meta_fields_to_embed: Optional[list[str]] = None,
+        meta_fields_to_embed: list[str] | None = None,
         embedding_separator: str = "\n",
-        timeout: Optional[float] = None,
+        timeout: float | None = None,
     ) -> None:
         """
         Create a NvidiaRanker component.
@@ -107,7 +107,7 @@ def __init__(
         self.api_url = url_validation(api_url)
         self.top_k = top_k
         self._initialized = False
-        self.backend: Optional[Any] = None
+        self.backend: Any | None = None
         self.is_hosted = is_hosted(api_url)
 
         self.query_prefix = query_prefix
@@ -192,7 +192,7 @@ def _prepare_documents_to_embed(self, documents: list[Document]) -> list[str]:
         return document_texts
 
     @component.output_types(documents=list[Document])
-    def run(self, query: str, documents: list[Document], top_k: Optional[int] = None) -> dict[str, list[Document]]:
+    def run(self, query: str, documents: list[Document], top_k: int | None = None) -> dict[str, list[Document]]:
         """
         Rank a list of documents based on a given query.
 
diff --git a/integrations/nvidia/src/haystack_integrations/utils/nvidia/models.py b/integrations/nvidia/src/haystack_integrations/utils/nvidia/models.py
@@ -3,7 +3,7 @@
 # SPDX-License-Identifier: Apache-2.0
 
 from dataclasses import dataclass
-from typing import Literal, Optional, Union
+from typing import Literal
 
 from .client import Client
 
@@ -24,13 +24,13 @@ class Model:
     """
 
     id: str
-    model_type: Optional[Literal["chat", "embedding", "ranking"]] = None
-    client: Optional[Union[Client, str]] = None
-    endpoint: Optional[str] = None
-    aliases: Optional[list] = None
-    base_model: Optional[str] = None
-    supports_tools: Optional[bool] = False
-    supports_structured_output: Optional[bool] = False
+    model_type: Literal["chat", "embedding", "ranking"] | None = None
+    client: Client | str | None = None
+    endpoint: str | None = None
+    aliases: list | None = None
+    base_model: str | None = None
+    supports_tools: bool | None = False
+    supports_structured_output: bool | None = False
 
     def __hash__(self) -> int:
         return hash(self.id)
@@ -480,7 +480,7 @@ def validate(self):
 
 DEFAULT_MODELS = {
     "embedding": "nvidia/nv-embedqa-e5-v5",
-    "ranking": "nvidia/nv-rerankqa-mistral-4b-v3",
+    "ranking": "nv-rerank-qa-mistral-4b:1",
     "chat": "meta/llama3-8b-instruct",
 }
 
diff --git a/integrations/nvidia/src/haystack_integrations/utils/nvidia/nim_backend.py b/integrations/nvidia/src/haystack_integrations/utils/nvidia/nim_backend.py
@@ -4,7 +4,7 @@
 
 import os
 import warnings
-from typing import Any, Literal, Optional, Union
+from typing import Any, Literal
 
 import requests
 from haystack import logging
@@ -23,12 +23,12 @@ class NimBackend:
     def __init__(
         self,
         api_url: str,
-        model_type: Optional[Literal["chat", "embedding", "ranking"]] = None,
-        model: Optional[str] = None,
-        api_key: Optional[Secret] = Secret.from_env_var("NVIDIA_API_KEY"),
-        model_kwargs: Optional[dict[str, Any]] = None,
-        client: Optional[Union[str, Client]] = None,
-        timeout: Optional[float] = None,
+        model_type: Literal["chat", "embedding", "ranking"] | None = None,
+        model: str | None = None,
+        api_key: Secret | None = Secret.from_env_var("NVIDIA_API_KEY"),
+        model_kwargs: dict[str, Any] | None = None,
+        client: str | Client | None = None,
+        timeout: float | None = None,
     ):
         headers = {
             "Content-Type": "application/json",
@@ -44,7 +44,7 @@ def __init__(
         self.api_url = api_url
         if isinstance(client, str):
             client = Client.from_str(client)
-        validated_model: Optional[Model] = None
+        validated_model: Model | None = None
         if is_hosted(self.api_url):
             if not api_key:
                 warnings.warn(
diff --git a/integrations/nvidia/src/haystack_integrations/utils/nvidia/utils.py b/integrations/nvidia/src/haystack_integrations/utils/nvidia/utils.py
@@ -3,7 +3,6 @@
 # SPDX-License-Identifier: Apache-2.0
 
 import warnings
-from typing import Optional
 from urllib.parse import urlparse, urlunparse
 
 from .client import Client
@@ -47,7 +46,7 @@ def is_hosted(api_url: str) -> bool:
     ]
 
 
-def lookup_model(name: str) -> Optional[Model]:
+def lookup_model(name: str) -> Model | None:
     """
     Lookup a model by name, using only the table of known models.
     The name is either:
@@ -65,7 +64,7 @@ def lookup_model(name: str) -> Optional[Model]:
     return model
 
 
-def determine_model(name: str) -> Optional[Model]:
+def determine_model(name: str) -> Model | None:
     """
     Determine the model to use based on a name, using
     only the table of known models.
@@ -85,7 +84,7 @@ def determine_model(name: str) -> Optional[Model]:
 
 def validate_hosted_model(
     model_name: str,
-    client: Optional[Client] = None,
+    client: Client | None = None,
 ) -> Model:
     """
     Checks if a given model is compatible with given client.
diff --git a/integrations/nvidia/tests/conftest.py b/integrations/nvidia/tests/conftest.py
@@ -2,7 +2,7 @@
 #
 # SPDX-License-Identifier: Apache-2.0
 
-from typing import Any, Optional
+from typing import Any
 
 import pytest
 from haystack.utils import Secret
@@ -12,7 +12,7 @@
 
 
 class MockBackend(NimBackend):
-    def __init__(self, model: str, api_key: Optional[Secret] = None, model_kwargs: Optional[dict[str, Any]] = None):
+    def __init__(self, model: str, api_key: Secret | None = None, model_kwargs: dict[str, Any] | None = None):
         api_key = api_key or Secret.from_env_var("NVIDIA_API_KEY")
         super().__init__(api_url="", model=model, api_key=api_key, model_kwargs=model_kwargs or {})
 
diff --git a/integrations/nvidia/tests/test_nvidia_chat_generator.py b/integrations/nvidia/tests/test_nvidia_chat_generator.py
diff --git a/integrations/nvidia/tests/test_ranker.py b/integrations/nvidia/tests/test_ranker.py
