chore: import upstream snapshot with attribution

test/components/embedders/__init__.py

@@ -0,0 +1,3 @@
+# SPDX-FileCopyrightText: 2022-present deepset GmbH <info@deepset.ai>
+#
+# SPDX-License-Identifier: Apache-2.0

test/components/embedders/test_azure_document_embedder.py

@@ -0,0 +1,385 @@
+# SPDX-FileCopyrightText: 2022-present deepset GmbH <info@deepset.ai>
+#
+# SPDX-License-Identifier: Apache-2.0
+
+import os
+from unittest.mock import AsyncMock, MagicMock, Mock, patch
+
+import pytest
+from openai import APIError, OpenAIError
+
+import haystack.components.embedders.azure_document_embedder as azure_document_embedder_module
+from haystack import Document
+from haystack.components.embedders import AzureOpenAIDocumentEmbedder
+from haystack.utils.auth import Secret
+from haystack.utils.azure import default_azure_ad_token_provider
+
+
+class TestAzureOpenAIDocumentEmbedder:
+    def test_init_default(self, monkeypatch):
+        monkeypatch.setenv("AZURE_OPENAI_API_KEY", "fake-api-key")
+        embedder = AzureOpenAIDocumentEmbedder(azure_endpoint="https://example-resource.azure.openai.com/")
+        assert embedder.azure_deployment == "text-embedding-ada-002"
+        assert embedder.model == "text-embedding-ada-002"
+        assert embedder.dimensions is None
+        assert embedder.organization is None
+        assert embedder.prefix == ""
+        assert embedder.suffix == ""
+        assert embedder.batch_size == 32
+        assert embedder.progress_bar is True
+        assert embedder.meta_fields_to_embed == []
+        assert embedder.embedding_separator == "\n"
+        assert embedder.timeout is None
+        assert embedder.max_retries is None
+        assert embedder.default_headers == {}
+        assert embedder.azure_ad_token_provider is None
+        assert embedder.http_client_kwargs is None
+        assert embedder.client is None
+        assert embedder.async_client is None
+
+    def test_init_with_0_max_retries(self, monkeypatch):
+        """Tests that the max_retries init param is set correctly if equal 0"""
+        monkeypatch.setenv("AZURE_OPENAI_API_KEY", "fake-api-key")
+        embedder = AzureOpenAIDocumentEmbedder(
+            azure_endpoint="https://example-resource.azure.openai.com/", max_retries=0
+        )
+        assert embedder.azure_deployment == "text-embedding-ada-002"
+        assert embedder.model == "text-embedding-ada-002"
+        assert embedder.dimensions is None
+        assert embedder.organization is None
+        assert embedder.prefix == ""
+        assert embedder.suffix == ""
+        assert embedder.batch_size == 32
+        assert embedder.progress_bar is True
+        assert embedder.meta_fields_to_embed == []
+        assert embedder.embedding_separator == "\n"
+        assert embedder.default_headers == {}
+        assert embedder.azure_ad_token_provider is None
+        assert embedder.max_retries == 0
+        assert embedder.client is None
+        assert embedder.async_client is None
+
+    def test_to_dict(self, monkeypatch):
+        monkeypatch.setenv("AZURE_OPENAI_API_KEY", "fake-api-key")
+        component = AzureOpenAIDocumentEmbedder(azure_endpoint="https://example-resource.azure.openai.com/")
+        data = component.to_dict()
+        assert data == {
+            "type": "haystack.components.embedders.azure_document_embedder.AzureOpenAIDocumentEmbedder",
+            "init_parameters": {
+                "api_key": {"env_vars": ["AZURE_OPENAI_API_KEY"], "strict": False, "type": "env_var"},
+                "azure_ad_token": {"env_vars": ["AZURE_OPENAI_AD_TOKEN"], "strict": False, "type": "env_var"},
+                "api_version": "2023-05-15",
+                "azure_deployment": "text-embedding-ada-002",
+                "dimensions": None,
+                "azure_endpoint": "https://example-resource.azure.openai.com/",
+                "organization": None,
+                "prefix": "",
+                "suffix": "",
+                "batch_size": 32,
+                "progress_bar": True,
+                "meta_fields_to_embed": [],
+                "embedding_separator": "\n",
+                "max_retries": None,
+                "timeout": None,
+                "default_headers": {},
+                "azure_ad_token_provider": None,
+                "http_client_kwargs": None,
+                "raise_on_failure": False,
+            },
+        }
+
+    def test_to_dict_with_parameters(self, monkeypatch):
+        monkeypatch.setenv("AZURE_OPENAI_API_KEY", "fake-api-key")
+        component = AzureOpenAIDocumentEmbedder(
+            azure_endpoint="https://example-resource.azure.openai.com/",
+            azure_deployment="text-embedding-ada-002",
+            dimensions=768,
+            organization="HaystackCI",
+            timeout=60.0,
+            max_retries=10,
+            prefix="prefix ",
+            suffix=" suffix",
+            default_headers={"x-custom-header": "custom-value"},
+            azure_ad_token_provider=default_azure_ad_token_provider,
+            http_client_kwargs={"proxy": "http://example.com:3128", "verify": False},
+            raise_on_failure=True,
+        )
+        data = component.to_dict()
+        assert data == {
+            "type": "haystack.components.embedders.azure_document_embedder.AzureOpenAIDocumentEmbedder",
+            "init_parameters": {
+                "api_key": {"env_vars": ["AZURE_OPENAI_API_KEY"], "strict": False, "type": "env_var"},
+                "azure_ad_token": {"env_vars": ["AZURE_OPENAI_AD_TOKEN"], "strict": False, "type": "env_var"},
+                "api_version": "2023-05-15",
+                "azure_deployment": "text-embedding-ada-002",
+                "dimensions": 768,
+                "azure_endpoint": "https://example-resource.azure.openai.com/",
+                "organization": "HaystackCI",
+                "prefix": "prefix ",
+                "suffix": " suffix",
+                "batch_size": 32,
+                "progress_bar": True,
+                "meta_fields_to_embed": [],
+                "embedding_separator": "\n",
+                "max_retries": 10,
+                "timeout": 60.0,
+                "default_headers": {"x-custom-header": "custom-value"},
+                "azure_ad_token_provider": "haystack.utils.azure.default_azure_ad_token_provider",
+                "http_client_kwargs": {"proxy": "http://example.com:3128", "verify": False},
+                "raise_on_failure": True,
+            },
+        }
+
+    def test_from_dict(self, monkeypatch):
+        monkeypatch.setenv("AZURE_OPENAI_API_KEY", "fake-api-key")
+        data = {
+            "type": "haystack.components.embedders.azure_document_embedder.AzureOpenAIDocumentEmbedder",
+            "init_parameters": {
+                "api_key": {"env_vars": ["AZURE_OPENAI_API_KEY"], "strict": False, "type": "env_var"},
+                "azure_ad_token": {"env_vars": ["AZURE_OPENAI_AD_TOKEN"], "strict": False, "type": "env_var"},
+                "api_version": "2023-05-15",
+                "azure_deployment": "text-embedding-ada-002",
+                "dimensions": None,
+                "azure_endpoint": "https://example-resource.azure.openai.com/",
+                "organization": None,
+                "prefix": "",
+                "suffix": "",
+                "batch_size": 32,
+                "progress_bar": True,
+                "meta_fields_to_embed": [],
+                "embedding_separator": "\n",
+                "max_retries": 5,
+                "timeout": 30.0,
+                "default_headers": {},
+                "azure_ad_token_provider": None,
+                "http_client_kwargs": None,
+                "raise_on_failure": False,
+            },
+        }
+        component = AzureOpenAIDocumentEmbedder.from_dict(data)
+        assert component.azure_deployment == "text-embedding-ada-002"
+        assert component.azure_endpoint == "https://example-resource.azure.openai.com/"
+        assert component.api_version == "2023-05-15"
+        assert component.max_retries == 5
+        assert component.timeout == 30.0
+        assert component.prefix == ""
+        assert component.suffix == ""
+        assert component.default_headers == {}
+        assert component.azure_ad_token_provider is None
+        assert component.http_client_kwargs is None
+        assert component.raise_on_failure is False
+
+    def test_from_dict_with_parameters(self, monkeypatch):
+        monkeypatch.setenv("AZURE_OPENAI_API_KEY", "fake-api-key")
+        data = {
+            "type": "haystack.components.embedders.azure_document_embedder.AzureOpenAIDocumentEmbedder",
+            "init_parameters": {
+                "api_key": {"env_vars": ["AZURE_OPENAI_API_KEY"], "strict": False, "type": "env_var"},
+                "azure_ad_token": {"env_vars": ["AZURE_OPENAI_AD_TOKEN"], "strict": False, "type": "env_var"},
+                "api_version": "2023-05-15",
+                "azure_deployment": "text-embedding-ada-002",
+                "dimensions": 768,
+                "azure_endpoint": "https://example-resource.azure.openai.com/",
+                "organization": "HaystackCI",
+                "prefix": "prefix ",
+                "suffix": " suffix",
+                "batch_size": 32,
+                "progress_bar": True,
+                "meta_fields_to_embed": [],
+                "embedding_separator": "\n",
+                "max_retries": 10,
+                "timeout": 60.0,
+                "default_headers": {"x-custom-header": "custom-value"},
+                "azure_ad_token_provider": "haystack.utils.azure.default_azure_ad_token_provider",
+                "http_client_kwargs": {"proxy": "http://example.com:3128", "verify": False},
+                "raise_on_failure": True,
+            },
+        }
+        component = AzureOpenAIDocumentEmbedder.from_dict(data)
+        assert component.azure_deployment == "text-embedding-ada-002"
+        assert component.azure_endpoint == "https://example-resource.azure.openai.com/"
+        assert component.api_version == "2023-05-15"
+        assert component.max_retries == 10
+        assert component.timeout == 60.0
+        assert component.prefix == "prefix "
+        assert component.suffix == " suffix"
+        assert component.default_headers == {"x-custom-header": "custom-value"}
+        assert component.azure_ad_token_provider is not None
+        assert component.http_client_kwargs == {"proxy": "http://example.com:3128", "verify": False}
+        assert component.raise_on_failure is True
+
+    def test_embed_batch_handles_exceptions_gracefully(self, caplog):
+        embedder = AzureOpenAIDocumentEmbedder(
+            azure_endpoint="https://test.openai.azure.com",
+            api_key=Secret.from_token("fake-api-key"),
+            azure_deployment="text-embedding-ada-002",
+            embedding_separator=" | ",
+        )
+        embedder.warm_up()
+        assert embedder.client is not None
+
+        fake_texts_to_embed = {"1": "text1", "2": "text2"}
+
+        with patch.object(
+            embedder.client.embeddings,
+            "create",
+            side_effect=APIError(message="Mocked error", request=Mock(), body=None),
+        ):
+            embedder._embed_batch(texts_to_embed=fake_texts_to_embed, batch_size=32)
+
+        assert len(caplog.records) == 1
+        assert "Failed embedding of documents 1, 2 caused by Mocked error" in caplog.text
+
+    def test_embed_batch_raises_exception_on_failure(self):
+        embedder = AzureOpenAIDocumentEmbedder(
+            azure_endpoint="https://test.openai.azure.com",
+            api_key=Secret.from_token("fake-api-key"),
+            azure_deployment="text-embedding-ada-002",
+            raise_on_failure=True,
+        )
+        embedder.warm_up()
+        assert embedder.client is not None
+        fake_texts_to_embed = {"1": "text1", "2": "text2"}
+        with patch.object(
+            embedder.client.embeddings,
+            "create",
+            side_effect=APIError(message="Mocked error", request=Mock(), body=None),
+        ):
+            with pytest.raises(APIError, match="Mocked error"):
+                embedder._embed_batch(texts_to_embed=fake_texts_to_embed, batch_size=2)
+
+    @pytest.mark.integration
+    @pytest.mark.skipif(
+        not os.environ.get("AZURE_OPENAI_API_KEY", None) and not os.environ.get("AZURE_OPENAI_ENDPOINT", None),
+        reason=(
+            "Please export env variables called AZURE_OPENAI_API_KEY containing "
+            "the Azure OpenAI key, AZURE_OPENAI_ENDPOINT containing "
+            "the Azure OpenAI endpoint URL to run this test."
+        ),
+    )
+    def test_run(self):
+        docs = [
+            Document(content="I love cheese", meta={"topic": "Cuisine"}),
+            Document(content="A transformer is a deep learning architecture", meta={"topic": "ML"}),
+        ]
+        # the default model is text-embedding-ada-002 even if we don't specify it, but let's be explicit
+        embedder = AzureOpenAIDocumentEmbedder(
+            azure_deployment="text-embedding-ada-002",
+            meta_fields_to_embed=["topic"],
+            embedding_separator=" | ",
+            organization="HaystackCI",
+        )
+
+        result = embedder.run(documents=docs)
+        documents_with_embeddings = result["documents"]
+        metadata = result["meta"]
+
+        assert isinstance(documents_with_embeddings, list)
+        assert len(documents_with_embeddings) == len(docs)
+        for doc, new_doc in zip(docs, documents_with_embeddings, strict=True):
+            assert doc.embedding is None
+            assert new_doc is not doc
+            assert isinstance(new_doc, Document)
+            assert isinstance(new_doc.embedding, list)
+            assert len(new_doc.embedding) == 1536
+            assert all(isinstance(x, float) for x in new_doc.embedding)
+
+        assert metadata["usage"]["prompt_tokens"] == 15
+        assert metadata["usage"]["total_tokens"] == 15
+        assert "ada" in metadata["model"]
+
+
+@pytest.fixture
+def mock_azure_clients(monkeypatch):
+    monkeypatch.setenv("AZURE_OPENAI_API_KEY", "fake")
+    sync_cls = MagicMock(name="AzureOpenAI")
+    async_cls = MagicMock(name="AsyncAzureOpenAI")
+    async_cls.return_value.close = AsyncMock()
+    monkeypatch.setattr(azure_document_embedder_module, "AzureOpenAI", sync_cls)
+    monkeypatch.setattr(azure_document_embedder_module, "AsyncAzureOpenAI", async_cls)
+    return sync_cls, async_cls
+
+
+class TestComponentLifecycle:
+    def test_warm_up_uses_default_timeout_and_max_retries(self, monkeypatch):
+        monkeypatch.setenv("AZURE_OPENAI_API_KEY", "fake-api-key")
+        embedder = AzureOpenAIDocumentEmbedder(azure_endpoint="https://example-resource.azure.openai.com/")
+        embedder.warm_up()
+        assert embedder.client is not None
+        assert embedder.client.max_retries == 5
+        assert embedder.client.timeout == 30.0
+
+    def test_warm_up_uses_timeout_and_max_retries_from_parameters(self, monkeypatch):
+        monkeypatch.setenv("AZURE_OPENAI_API_KEY", "fake-api-key")
+        embedder = AzureOpenAIDocumentEmbedder(
+            azure_endpoint="https://example-resource.azure.openai.com/", timeout=40.0, max_retries=1
+        )
+        embedder.warm_up()
+        assert embedder.client is not None
+        assert embedder.client.max_retries == 1
+        assert embedder.client.timeout == 40.0
+
+    def test_warm_up_uses_timeout_and_max_retries_from_env_vars(self, monkeypatch):
+        monkeypatch.setenv("AZURE_OPENAI_API_KEY", "fake-api-key")
+        monkeypatch.setenv("OPENAI_TIMEOUT", "100")
+        monkeypatch.setenv("OPENAI_MAX_RETRIES", "10")
+        embedder = AzureOpenAIDocumentEmbedder(azure_endpoint="https://example-resource.azure.openai.com/")
+        embedder.warm_up()
+        assert embedder.client is not None
+        assert embedder.client.max_retries == 10
+        assert embedder.client.timeout == 100.0
+
+    def test_key_resolved_at_warm_up_not_init(self, monkeypatch):
+        monkeypatch.delenv("AZURE_OPENAI_API_KEY", raising=False)
+        monkeypatch.delenv("AZURE_OPENAI_AD_TOKEN", raising=False)
+        embedder = AzureOpenAIDocumentEmbedder(azure_endpoint="https://example-resource.azure.openai.com/")
+        assert embedder.client is None
+        with pytest.raises(OpenAIError):
+            embedder.warm_up()
+
+    def test_sync_lifecycle(self, mock_azure_clients):
+        sync_cls, _ = mock_azure_clients
+        sync_client = sync_cls.return_value
+        embedder = AzureOpenAIDocumentEmbedder(azure_endpoint="https://example-resource.azure.openai.com/")
+        assert embedder.client is None
+        assert embedder.async_client is None
+
+        embedder.warm_up()
+        assert embedder.client is sync_cls.return_value
+        assert embedder.async_client is None
+
+        embedder.close()
+        sync_client.close.assert_called_once()
+        assert embedder.client is None
+
+    async def test_async_lifecycle(self, mock_azure_clients):
+        _, async_cls = mock_azure_clients
+        async_client = async_cls.return_value
+        embedder = AzureOpenAIDocumentEmbedder(azure_endpoint="https://example-resource.azure.openai.com/")
+
+        await embedder.warm_up_async()
+        assert embedder.async_client is async_cls.return_value
+        assert embedder.client is None
+
+        await embedder.close_async()
+        async_client.close.assert_awaited_once()
+        assert embedder.async_client is None
+
+    async def test_close_is_safe_without_warm_up(self, mock_azure_clients):
+        embedder = AzureOpenAIDocumentEmbedder(azure_endpoint="https://example-resource.azure.openai.com/")
+        embedder.close()
+        await embedder.close_async()
+        assert embedder.client is None
+        assert embedder.async_client is None
+
+    async def test_close_and_close_async_are_independent(self, mock_azure_clients):
+        embedder = AzureOpenAIDocumentEmbedder(azure_endpoint="https://example-resource.azure.openai.com/")
+        embedder.warm_up()
+        await embedder.warm_up_async()
+
+        embedder.close()
+        assert embedder.client is None
+        assert embedder.async_client is not None
+
+        await embedder.close_async()
+        assert embedder.async_client is None

test/components/embedders/test_azure_text_embedder.py

@@ -0,0 +1,296 @@
+# SPDX-FileCopyrightText: 2022-present deepset GmbH <info@deepset.ai>
+#
+# SPDX-License-Identifier: Apache-2.0
+
+import os
+from unittest.mock import AsyncMock, MagicMock
+
+import pytest
+from openai import OpenAIError
+
+import haystack.components.embedders.azure_text_embedder as azure_text_embedder_module
+from haystack.components.embedders import AzureOpenAITextEmbedder
+from haystack.utils.azure import default_azure_ad_token_provider
+
+
+class TestAzureOpenAITextEmbedder:
+    def test_init_default(self, monkeypatch):
+        monkeypatch.setenv("AZURE_OPENAI_API_KEY", "fake-api-key")
+        embedder = AzureOpenAITextEmbedder(azure_endpoint="https://example-resource.azure.openai.com/")
+
+        assert embedder.api_key.resolve_value() == "fake-api-key"
+        assert embedder.azure_deployment == "text-embedding-ada-002"
+        assert embedder.model == "text-embedding-ada-002"
+        assert embedder.dimensions is None
+        assert embedder.organization is None
+        assert embedder.prefix == ""
+        assert embedder.suffix == ""
+        assert embedder.timeout is None
+        assert embedder.max_retries is None
+        assert embedder.default_headers == {}
+        assert embedder.azure_ad_token_provider is None
+        assert embedder.http_client_kwargs is None
+        assert embedder.client is None
+        assert embedder.async_client is None
+
+    def test_init_with_zero_max_retries(self, monkeypatch):
+        """Tests that the max_retries init param is set correctly if equal 0"""
+        monkeypatch.setenv("AZURE_OPENAI_API_KEY", "fake-api-key")
+        embedder = AzureOpenAITextEmbedder(azure_endpoint="https://example-resource.azure.openai.com/", max_retries=0)
+
+        assert embedder.api_key.resolve_value() == "fake-api-key"
+        assert embedder.azure_deployment == "text-embedding-ada-002"
+        assert embedder.model == "text-embedding-ada-002"
+        assert embedder.dimensions is None
+        assert embedder.organization is None
+        assert embedder.prefix == ""
+        assert embedder.suffix == ""
+        assert embedder.default_headers == {}
+        assert embedder.azure_ad_token_provider is None
+        assert embedder.max_retries == 0
+        assert embedder.client is None
+        assert embedder.async_client is None
+
+    def test_to_dict_default(self, monkeypatch):
+        monkeypatch.setenv("AZURE_OPENAI_API_KEY", "fake-api-key")
+        component = AzureOpenAITextEmbedder(azure_endpoint="https://example-resource.azure.openai.com/")
+        data = component.to_dict()
+        assert data == {
+            "type": "haystack.components.embedders.azure_text_embedder.AzureOpenAITextEmbedder",
+            "init_parameters": {
+                "api_key": {"env_vars": ["AZURE_OPENAI_API_KEY"], "strict": False, "type": "env_var"},
+                "azure_ad_token": {"env_vars": ["AZURE_OPENAI_AD_TOKEN"], "strict": False, "type": "env_var"},
+                "azure_deployment": "text-embedding-ada-002",
+                "dimensions": None,
+                "organization": None,
+                "azure_endpoint": "https://example-resource.azure.openai.com/",
+                "api_version": "2023-05-15",
+                "max_retries": None,
+                "timeout": None,
+                "prefix": "",
+                "suffix": "",
+                "default_headers": {},
+                "azure_ad_token_provider": None,
+                "http_client_kwargs": None,
+            },
+        }
+
+    def test_to_dict_with_params(self, monkeypatch):
+        monkeypatch.setenv("AZURE_OPENAI_API_KEY", "fake-api-key")
+        component = AzureOpenAITextEmbedder(
+            azure_endpoint="https://example-resource.azure.openai.com/",
+            azure_deployment="text-embedding-ada-002",
+            dimensions=768,
+            organization="HaystackCI",
+            timeout=60.0,
+            max_retries=10,
+            prefix="prefix ",
+            suffix=" suffix",
+            default_headers={"x-custom-header": "custom-value"},
+            azure_ad_token_provider=default_azure_ad_token_provider,
+            http_client_kwargs={"proxy": "http://example.com:3128", "verify": False},
+        )
+        data = component.to_dict()
+        assert data == {
+            "type": "haystack.components.embedders.azure_text_embedder.AzureOpenAITextEmbedder",
+            "init_parameters": {
+                "api_key": {"env_vars": ["AZURE_OPENAI_API_KEY"], "strict": False, "type": "env_var"},
+                "azure_ad_token": {"env_vars": ["AZURE_OPENAI_AD_TOKEN"], "strict": False, "type": "env_var"},
+                "azure_deployment": "text-embedding-ada-002",
+                "dimensions": 768,
+                "organization": "HaystackCI",
+                "azure_endpoint": "https://example-resource.azure.openai.com/",
+                "api_version": "2023-05-15",
+                "max_retries": 10,
+                "timeout": 60.0,
+                "prefix": "prefix ",
+                "suffix": " suffix",
+                "default_headers": {"x-custom-header": "custom-value"},
+                "azure_ad_token_provider": "haystack.utils.azure.default_azure_ad_token_provider",
+                "http_client_kwargs": {"proxy": "http://example.com:3128", "verify": False},
+            },
+        }
+
+    def test_from_dict(self, monkeypatch):
+        monkeypatch.setenv("AZURE_OPENAI_API_KEY", "fake-api-key")
+        data = {
+            "type": "haystack.components.embedders.azure_text_embedder.AzureOpenAITextEmbedder",
+            "init_parameters": {
+                "api_key": {"env_vars": ["AZURE_OPENAI_API_KEY"], "strict": False, "type": "env_var"},
+                "azure_ad_token": {"env_vars": ["AZURE_OPENAI_AD_TOKEN"], "strict": False, "type": "env_var"},
+                "azure_deployment": "text-embedding-ada-002",
+                "dimensions": None,
+                "organization": None,
+                "azure_endpoint": "https://example-resource.azure.openai.com/",
+                "api_version": "2023-05-15",
+                "max_retries": 5,
+                "timeout": 30.0,
+                "prefix": "",
+                "suffix": "",
+                "default_headers": {},
+                "http_client_kwargs": None,
+            },
+        }
+        component = AzureOpenAITextEmbedder.from_dict(data)
+        assert component.azure_deployment == "text-embedding-ada-002"
+        assert component.model == "text-embedding-ada-002"
+        assert component.azure_endpoint == "https://example-resource.azure.openai.com/"
+        assert component.api_version == "2023-05-15"
+        assert component.max_retries == 5
+        assert component.timeout == 30.0
+        assert component.prefix == ""
+        assert component.suffix == ""
+        assert component.default_headers == {}
+        assert component.azure_ad_token_provider is None
+        assert component.http_client_kwargs is None
+
+    def test_from_dict_with_parameters(self, monkeypatch):
+        monkeypatch.setenv("AZURE_OPENAI_API_KEY", "fake-api-key")
+        data = {
+            "type": "haystack.components.embedders.azure_text_embedder.AzureOpenAITextEmbedder",
+            "init_parameters": {
+                "api_key": {"env_vars": ["AZURE_OPENAI_API_KEY"], "strict": False, "type": "env_var"},
+                "azure_ad_token": {"env_vars": ["AZURE_OPENAI_AD_TOKEN"], "strict": False, "type": "env_var"},
+                "azure_deployment": "text-embedding-ada-002",
+                "dimensions": 768,
+                "organization": "HaystackCI",
+                "azure_endpoint": "https://example-resource.azure.openai.com/",
+                "api_version": "2023-05-15",
+                "max_retries": 10,
+                "timeout": 60.0,
+                "prefix": "prefix ",
+                "suffix": " suffix",
+                "default_headers": {"x-custom-header": "custom-value"},
+                "azure_ad_token_provider": "haystack.utils.azure.default_azure_ad_token_provider",
+                "http_client_kwargs": {"proxy": "http://example.com:3128", "verify": False},
+            },
+        }
+        component = AzureOpenAITextEmbedder.from_dict(data)
+        assert component.azure_deployment == "text-embedding-ada-002"
+        assert component.model == "text-embedding-ada-002"
+        assert component.azure_endpoint == "https://example-resource.azure.openai.com/"
+        assert component.api_version == "2023-05-15"
+        assert component.max_retries == 10
+        assert component.timeout == 60.0
+        assert component.prefix == "prefix "
+        assert component.suffix == " suffix"
+        assert component.default_headers == {"x-custom-header": "custom-value"}
+        assert component.azure_ad_token_provider is not None
+        assert component.http_client_kwargs == {"proxy": "http://example.com:3128", "verify": False}
+
+    @pytest.mark.integration
+    @pytest.mark.skipif(
+        not os.environ.get("AZURE_OPENAI_API_KEY", None) and not os.environ.get("AZURE_OPENAI_ENDPOINT", None),
+        reason=(
+            "Please export env variables called AZURE_OPENAI_API_KEY containing "
+            "the Azure OpenAI key, AZURE_OPENAI_ENDPOINT containing "
+            "the Azure OpenAI endpoint URL to run this test."
+        ),
+    )
+    def test_run(self):
+        # the default model is text-embedding-ada-002 even if we don't specify it, but let's be explicit
+        embedder = AzureOpenAITextEmbedder(
+            azure_deployment="text-embedding-ada-002", prefix="prefix ", suffix=" suffix", organization="HaystackCI"
+        )
+        result = embedder.run(text="The food was delicious")
+
+        assert len(result["embedding"]) == 1536
+        assert all(isinstance(x, float) for x in result["embedding"])
+        assert result["meta"]["usage"] == {"prompt_tokens": 6, "total_tokens": 6}
+        assert "ada" in result["meta"]["model"]
+
+
+@pytest.fixture
+def mock_azure_clients(monkeypatch):
+    monkeypatch.setenv("AZURE_OPENAI_API_KEY", "fake")
+    sync_cls = MagicMock(name="AzureOpenAI")
+    async_cls = MagicMock(name="AsyncAzureOpenAI")
+    async_cls.return_value.close = AsyncMock()
+    monkeypatch.setattr(azure_text_embedder_module, "AzureOpenAI", sync_cls)
+    monkeypatch.setattr(azure_text_embedder_module, "AsyncAzureOpenAI", async_cls)
+    return sync_cls, async_cls
+
+
+class TestComponentLifecycle:
+    def test_warm_up_uses_default_timeout_and_max_retries(self, monkeypatch):
+        monkeypatch.setenv("AZURE_OPENAI_API_KEY", "fake-api-key")
+        embedder = AzureOpenAITextEmbedder(azure_endpoint="https://example-resource.azure.openai.com/")
+        embedder.warm_up()
+        assert embedder.client is not None
+        assert embedder.client.max_retries == 5
+        assert embedder.client.timeout == 30.0
+
+    def test_warm_up_uses_timeout_and_max_retries_from_parameters(self, monkeypatch):
+        monkeypatch.setenv("AZURE_OPENAI_API_KEY", "fake-api-key")
+        embedder = AzureOpenAITextEmbedder(
+            azure_endpoint="https://example-resource.azure.openai.com/", timeout=40.0, max_retries=1
+        )
+        embedder.warm_up()
+        assert embedder.client is not None
+        assert embedder.client.max_retries == 1
+        assert embedder.client.timeout == 40.0
+
+    def test_warm_up_uses_timeout_and_max_retries_from_env_vars(self, monkeypatch):
+        monkeypatch.setenv("AZURE_OPENAI_API_KEY", "fake-api-key")
+        monkeypatch.setenv("OPENAI_TIMEOUT", "100")
+        monkeypatch.setenv("OPENAI_MAX_RETRIES", "10")
+        embedder = AzureOpenAITextEmbedder(azure_endpoint="https://example-resource.azure.openai.com/")
+        embedder.warm_up()
+        assert embedder.client is not None
+        assert embedder.client.max_retries == 10
+        assert embedder.client.timeout == 100.0
+
+    def test_key_resolved_at_warm_up_not_init(self, monkeypatch):
+        monkeypatch.delenv("AZURE_OPENAI_API_KEY", raising=False)
+        monkeypatch.delenv("AZURE_OPENAI_AD_TOKEN", raising=False)
+        embedder = AzureOpenAITextEmbedder(azure_endpoint="https://example-resource.azure.openai.com/")
+        assert embedder.client is None
+        with pytest.raises(OpenAIError):
+            embedder.warm_up()
+
+    def test_sync_lifecycle(self, mock_azure_clients):
+        sync_cls, _ = mock_azure_clients
+        sync_client = sync_cls.return_value
+        embedder = AzureOpenAITextEmbedder(azure_endpoint="https://example-resource.azure.openai.com/")
+        assert embedder.client is None
+        assert embedder.async_client is None
+
+        embedder.warm_up()
+        assert embedder.client is sync_cls.return_value
+        assert embedder.async_client is None
+
+        embedder.close()
+        sync_client.close.assert_called_once()
+        assert embedder.client is None
+
+    async def test_async_lifecycle(self, mock_azure_clients):
+        _, async_cls = mock_azure_clients
+        async_client = async_cls.return_value
+        embedder = AzureOpenAITextEmbedder(azure_endpoint="https://example-resource.azure.openai.com/")
+
+        await embedder.warm_up_async()
+        assert embedder.async_client is async_cls.return_value
+        assert embedder.client is None
+
+        await embedder.close_async()
+        async_client.close.assert_awaited_once()
+        assert embedder.async_client is None
+
+    async def test_close_is_safe_without_warm_up(self, mock_azure_clients):
+        embedder = AzureOpenAITextEmbedder(azure_endpoint="https://example-resource.azure.openai.com/")
+        embedder.close()
+        await embedder.close_async()
+        assert embedder.client is None
+        assert embedder.async_client is None
+
+    async def test_close_and_close_async_are_independent(self, mock_azure_clients):
+        embedder = AzureOpenAITextEmbedder(azure_endpoint="https://example-resource.azure.openai.com/")
+        embedder.warm_up()
+        await embedder.warm_up_async()
+
+        embedder.close()
+        assert embedder.client is None
+        assert embedder.async_client is not None
+
+        await embedder.close_async()
+        assert embedder.async_client is None

test/components/embedders/test_mock_document_embedder.py

@@ -0,0 +1,119 @@
+# SPDX-FileCopyrightText: 2022-present deepset GmbH <info@deepset.ai>
+#
+# SPDX-License-Identifier: Apache-2.0
+
+import pytest
+
+from haystack import Document, Pipeline
+from haystack.components.embedders import MockDocumentEmbedder, MockTextEmbedder
+
+
+def _ones(text: str) -> list[float]:
+    """Module-level embedding function used to test `embedding_fn` and its serialization."""
+    return [1.0, 1.0, 1.0]
+
+
+class TestMockDocumentEmbedder:
+    @pytest.mark.parametrize(
+        ("args", "kwargs", "match"),
+        [
+            (([0.1],), {"embedding_fn": _ones}, "either 'embedding' or 'embedding_fn'"),
+            ((), {"dimension": -1}, "must be a positive integer"),
+        ],
+    )
+    def test_init_rejects_invalid_config(self, args, kwargs, match):
+        with pytest.raises(ValueError, match=match):
+            MockDocumentEmbedder(*args, **kwargs)
+
+    def test_embeds_documents(self):
+        embedder = MockDocumentEmbedder(dimension=16)
+        result = embedder.run([Document(content="first"), Document(content="second")])
+        embeddings = [doc.embedding for doc in result["documents"]]
+        assert all(len(embedding) == 16 for embedding in embeddings)
+        assert embeddings[0] != embeddings[1]
+
+    def test_consistent_with_text_embedder(self):
+        # the same prepared text yields the same embedding from both mock embedders (shared deterministic algorithm)
+        text_embedding = MockTextEmbedder(dimension=8).run("pizza")["embedding"]
+        doc_embedding = MockDocumentEmbedder(dimension=8).run([Document(content="pizza")])["documents"][0].embedding
+        assert text_embedding == doc_embedding
+
+    def test_fixed_embedding(self):
+        result = MockDocumentEmbedder([0.5, 0.5]).run([Document(content="a"), Document(content="b")])
+        assert all(doc.embedding == [0.5, 0.5] for doc in result["documents"])
+
+    def test_embedding_fn(self):
+        result = MockDocumentEmbedder(embedding_fn=_ones).run([Document(content="a")])
+        assert result["documents"][0].embedding == [1.0, 1.0, 1.0]
+
+    def test_meta_fields_to_embed_affect_embedding(self):
+        document = Document(content="hello", meta={"title": "Greetings"})
+        without_meta = MockDocumentEmbedder(dimension=8).run([document])["documents"][0].embedding
+        with_meta = (
+            MockDocumentEmbedder(dimension=8, meta_fields_to_embed=["title"]).run([document])["documents"][0].embedding
+        )
+        assert without_meta != with_meta
+
+    def test_preserves_document_fields(self):
+        document = Document(content="hello", meta={"title": "Greetings"})
+        embedded = MockDocumentEmbedder(dimension=8).run([document])["documents"][0]
+        assert embedded.id == document.id
+        assert embedded.content == "hello"
+        assert embedded.meta == {"title": "Greetings"}
+        assert embedded.embedding is not None
+        # the original document is not mutated
+        assert document.embedding is None
+
+    def test_empty_documents(self):
+        result = MockDocumentEmbedder().run([])
+        assert result["documents"] == []
+        assert result["meta"]["usage"] == {"prompt_tokens": 0, "total_tokens": 0}
+
+    def test_meta(self):
+        meta = MockDocumentEmbedder(dimension=4).run([Document(content="a b"), Document(content="c")])["meta"]
+        assert meta["model"] == "mock-model"
+        assert meta["usage"] == {"prompt_tokens": 3, "total_tokens": 3}
+
+    @pytest.mark.parametrize("documents", ["not a list", [1, 2, 3]])
+    def test_run_rejects_non_documents(self, documents):
+        with pytest.raises(TypeError, match="expects a list of Documents"):
+            MockDocumentEmbedder().run(documents)
+
+    async def test_run_async(self):
+        embedder = MockDocumentEmbedder(dimension=8)
+        documents = [Document(content="hello")]
+        async_embedding = (await embedder.run_async(documents))["documents"][0].embedding
+        assert async_embedding == embedder.run(documents)["documents"][0].embedding
+
+    def test_warm_up_sets_flag_and_run_auto_warms(self):
+        embedder = MockDocumentEmbedder(dimension=8)
+        assert embedder._is_warmed_up is False
+        embedder.warm_up()
+        assert embedder._is_warmed_up is True
+        # run auto-warms a fresh instance
+        fresh = MockDocumentEmbedder(dimension=8)
+        assert len(fresh.run([Document(content="hello")])["documents"][0].embedding) == 8
+        assert fresh._is_warmed_up is True
+
+    @pytest.mark.parametrize(
+        "embedder",
+        [
+            MockDocumentEmbedder(
+                dimension=8, model="m", meta={"k": "v"}, meta_fields_to_embed=["title"], embedding_separator=" | "
+            ),
+            MockDocumentEmbedder(embedding_fn=_ones),
+        ],
+        ids=["deterministic", "embedding_fn"],
+    )
+    def test_serialization_roundtrip(self, embedder):
+        restored = MockDocumentEmbedder.from_dict(embedder.to_dict())
+        assert isinstance(restored, MockDocumentEmbedder)
+        document = Document(content="hello", meta={"title": "t"})
+        assert restored.run([document])["documents"][0].embedding == embedder.run([document])["documents"][0].embedding
+
+    def test_in_pipeline(self):
+        pipeline = Pipeline()
+        pipeline.add_component("embedder", MockDocumentEmbedder(dimension=8))
+        restored = Pipeline.from_dict(pipeline.to_dict())
+        result = restored.run({"embedder": {"documents": [Document(content="hello")]}})
+        assert len(result["embedder"]["documents"][0].embedding) == 8

test/components/embedders/test_mock_text_embedder.py

@@ -0,0 +1,120 @@
+# SPDX-FileCopyrightText: 2022-present deepset GmbH <info@deepset.ai>
+#
+# SPDX-License-Identifier: Apache-2.0
+
+import math
+
+import pytest
+
+from haystack import Pipeline
+from haystack.components.embedders import MockTextEmbedder
+from haystack.components.embedders.mock_utils import _l2_normalize
+
+
+def _ones(text: str) -> list[float]:
+    """Module-level embedding function used to test `embedding_fn` and its serialization."""
+    return [1.0, 1.0, 1.0]
+
+
+def test_l2_normalize_handles_zero_vector():
+    # defensive guard in the shared deterministic-embedding helper: a zero vector is returned unchanged
+    assert _l2_normalize([0.0, 0.0]) == [0.0, 0.0]
+
+
+class TestMockTextEmbedder:
+    @pytest.mark.parametrize(
+        ("args", "kwargs", "exception", "match"),
+        [
+            (([0.1, 0.2],), {"embedding_fn": _ones}, ValueError, "either 'embedding' or 'embedding_fn'"),
+            ((), {"dimension": 0}, ValueError, "must be a positive integer"),
+            (([],), {}, ValueError, "must not be empty"),
+            ((["not", "numbers"],), {}, TypeError, "must be a sequence of numbers"),
+        ],
+    )
+    def test_init_rejects_invalid_config(self, args, kwargs, exception, match):
+        with pytest.raises(exception, match=match):
+            MockTextEmbedder(*args, **kwargs)
+
+    def test_deterministic_embedding(self):
+        embedding = MockTextEmbedder(dimension=16).run("hello")["embedding"]
+        assert len(embedding) == 16
+        assert all(isinstance(value, float) for value in embedding)
+        # embeddings are L2-normalized, like real ones
+        assert math.isclose(math.sqrt(sum(value * value for value in embedding)), 1.0, abs_tol=1e-9)
+
+    def test_deterministic_distinguishes_texts(self):
+        embedder = MockTextEmbedder(dimension=8)
+        assert embedder.run("pizza")["embedding"] == embedder.run("pizza")["embedding"]
+        assert embedder.run("pizza")["embedding"] != embedder.run("pasta")["embedding"]
+        # determinism holds across instances and processes (stable hash, not the salted built-in hash)
+        assert (
+            MockTextEmbedder(dimension=8).run("x")["embedding"] == MockTextEmbedder(dimension=8).run("x")["embedding"]
+        )
+
+    def test_fixed_embedding(self):
+        embedder = MockTextEmbedder([0.1, 0.2, 0.3])
+        assert embedder.run("anything")["embedding"] == [0.1, 0.2, 0.3]
+        assert embedder.run("something else")["embedding"] == [0.1, 0.2, 0.3]
+
+    def test_embedding_fn(self):
+        assert MockTextEmbedder(embedding_fn=_ones).run("hello")["embedding"] == [1.0, 1.0, 1.0]
+
+    def test_embedding_fn_invalid_return_raises(self):
+        # embedding_fn deliberately returns a non-vector to exercise the runtime type check
+        embedder = MockTextEmbedder(embedding_fn=lambda text: "not a vector")  # type: ignore[arg-type, return-value]
+        with pytest.raises(TypeError, match="must be a sequence of numbers"):
+            embedder.run("hello")
+
+    def test_prefix_suffix_affect_embedding(self):
+        plain = MockTextEmbedder(dimension=8).run("hello")["embedding"]
+        prefixed = MockTextEmbedder(dimension=8, prefix="search: ").run("hello")["embedding"]
+        assert plain != prefixed
+
+    def test_meta(self):
+        meta = MockTextEmbedder(dimension=4).run("a b c")["meta"]
+        assert meta["model"] == "mock-model"
+        assert meta["usage"] == {"prompt_tokens": 3, "total_tokens": 3}
+        # init model and meta are reflected and merged
+        custom = MockTextEmbedder(dimension=4, model="custom", meta={"extra": "value"}).run("hi")["meta"]
+        assert custom["model"] == "custom"
+        assert custom["extra"] == "value"
+
+    def test_run_rejects_non_string(self):
+        # a non-string input is passed on purpose to exercise the runtime type check
+        with pytest.raises(TypeError, match="expects a string"):
+            MockTextEmbedder().run(["not", "a", "string"])  # type: ignore[arg-type]
+
+    async def test_run_async(self):
+        embedder = MockTextEmbedder(dimension=8)
+        assert (await embedder.run_async("hello"))["embedding"] == embedder.run("hello")["embedding"]
+
+    def test_warm_up_sets_flag_and_run_auto_warms(self):
+        embedder = MockTextEmbedder(dimension=8)
+        assert embedder._is_warmed_up is False
+        embedder.warm_up()
+        assert embedder._is_warmed_up is True
+        # run auto-warms a fresh instance
+        fresh = MockTextEmbedder(dimension=8)
+        assert len(fresh.run("hello")["embedding"]) == 8
+        assert fresh._is_warmed_up is True
+
+    @pytest.mark.parametrize(
+        "embedder",
+        [
+            MockTextEmbedder(dimension=8, model="m", meta={"k": "v"}, prefix="p", suffix="s"),
+            MockTextEmbedder(embedding_fn=_ones),
+            MockTextEmbedder([0.1, 0.2, 0.3]),
+        ],
+        ids=["deterministic", "embedding_fn", "fixed"],
+    )
+    def test_serialization_roundtrip(self, embedder):
+        restored = MockTextEmbedder.from_dict(embedder.to_dict())
+        assert isinstance(restored, MockTextEmbedder)
+        assert restored.run("hello")["embedding"] == embedder.run("hello")["embedding"]
+
+    def test_in_pipeline(self):
+        pipeline = Pipeline()
+        pipeline.add_component("embedder", MockTextEmbedder(dimension=8))
+        restored = Pipeline.from_dict(pipeline.to_dict())
+        result = restored.run({"embedder": {"text": "hello"}})
+        assert len(result["embedder"]["embedding"]) == 8

test/components/embedders/test_openai_document_embedder.py

@@ -0,0 +1,431 @@
+# SPDX-FileCopyrightText: 2022-present deepset GmbH <info@deepset.ai>
+#
+# SPDX-License-Identifier: Apache-2.0
+
+import contextlib
+import os
+from unittest.mock import AsyncMock, MagicMock, Mock, patch
+
+import pytest
+from openai import APIError
+
+import haystack.components.embedders.openai_document_embedder as openai_document_embedder_module
+from haystack import Document
+from haystack.components.embedders.openai_document_embedder import OpenAIDocumentEmbedder
+from haystack.utils.auth import Secret
+
+
+class TestOpenAIDocumentEmbedder:
+    def test_init_default(self, monkeypatch):
+        monkeypatch.setenv("OPENAI_API_KEY", "fake-api-key")
+        embedder = OpenAIDocumentEmbedder()
+        assert embedder.api_key.resolve_value() == "fake-api-key"
+        assert embedder.model == "text-embedding-ada-002"
+        assert embedder.organization is None
+        assert embedder.prefix == ""
+        assert embedder.suffix == ""
+        assert embedder.batch_size == 32
+        assert embedder.progress_bar is True
+        assert embedder.meta_fields_to_embed == []
+        assert embedder.embedding_separator == "\n"
+        assert embedder.timeout is None
+        assert embedder.max_retries is None
+        assert embedder.client is None
+        assert embedder.async_client is None
+
+    def test_init_with_parameters(self, monkeypatch):
+        monkeypatch.setenv("OPENAI_TIMEOUT", "100")
+        monkeypatch.setenv("OPENAI_MAX_RETRIES", "10")
+        embedder = OpenAIDocumentEmbedder(
+            api_key=Secret.from_token("fake-api-key-2"),
+            model="model",
+            organization="my-org",
+            prefix="prefix",
+            suffix="suffix",
+            batch_size=64,
+            progress_bar=False,
+            meta_fields_to_embed=["test_field"],
+            embedding_separator=" | ",
+            timeout=40.0,
+            max_retries=1,
+        )
+        assert embedder.api_key.resolve_value() == "fake-api-key-2"
+        assert embedder.organization == "my-org"
+        assert embedder.model == "model"
+        assert embedder.prefix == "prefix"
+        assert embedder.suffix == "suffix"
+        assert embedder.batch_size == 64
+        assert embedder.progress_bar is False
+        assert embedder.meta_fields_to_embed == ["test_field"]
+        assert embedder.embedding_separator == " | "
+        assert embedder.timeout == 40.0
+        assert embedder.max_retries == 1
+        assert embedder.client is None
+        assert embedder.async_client is None
+
+    def test_init_with_parameters_and_env_vars(self, monkeypatch):
+        monkeypatch.setenv("OPENAI_TIMEOUT", "100")
+        monkeypatch.setenv("OPENAI_MAX_RETRIES", "10")
+        embedder = OpenAIDocumentEmbedder(
+            api_key=Secret.from_token("fake-api-key-2"),
+            model="model",
+            organization="my-org",
+            prefix="prefix",
+            suffix="suffix",
+            batch_size=64,
+            progress_bar=False,
+            meta_fields_to_embed=["test_field"],
+            embedding_separator=" | ",
+        )
+        assert embedder.api_key.resolve_value() == "fake-api-key-2"
+        assert embedder.organization == "my-org"
+        assert embedder.model == "model"
+        assert embedder.prefix == "prefix"
+        assert embedder.suffix == "suffix"
+        assert embedder.batch_size == 64
+        assert embedder.progress_bar is False
+        assert embedder.meta_fields_to_embed == ["test_field"]
+        assert embedder.embedding_separator == " | "
+        assert embedder.timeout is None
+        assert embedder.max_retries is None
+        assert embedder.client is None
+        assert embedder.async_client is None
+
+    def test_to_dict(self, monkeypatch):
+        monkeypatch.setenv("OPENAI_API_KEY", "fake-api-key")
+        component = OpenAIDocumentEmbedder()
+        data = component.to_dict()
+        assert data == {
+            "type": "haystack.components.embedders.openai_document_embedder.OpenAIDocumentEmbedder",
+            "init_parameters": {
+                "api_key": {"env_vars": ["OPENAI_API_KEY"], "strict": True, "type": "env_var"},
+                "api_base_url": None,
+                "model": "text-embedding-ada-002",
+                "dimensions": None,
+                "organization": None,
+                "http_client_kwargs": None,
+                "prefix": "",
+                "suffix": "",
+                "batch_size": 32,
+                "progress_bar": True,
+                "meta_fields_to_embed": [],
+                "embedding_separator": "\n",
+                "timeout": None,
+                "max_retries": None,
+                "raise_on_failure": False,
+            },
+        }
+
+    def test_to_dict_with_custom_init_parameters(self, monkeypatch):
+        monkeypatch.setenv("ENV_VAR", "fake-api-key")
+        component = OpenAIDocumentEmbedder(
+            api_key=Secret.from_env_var("ENV_VAR", strict=False),
+            model="model",
+            organization="my-org",
+            http_client_kwargs={"proxy": "http://localhost:8080"},
+            prefix="prefix",
+            suffix="suffix",
+            batch_size=64,
+            progress_bar=False,
+            meta_fields_to_embed=["test_field"],
+            embedding_separator=" | ",
+            timeout=10.0,
+            max_retries=2,
+            raise_on_failure=True,
+        )
+        data = component.to_dict()
+        assert data == {
+            "type": "haystack.components.embedders.openai_document_embedder.OpenAIDocumentEmbedder",
+            "init_parameters": {
+                "api_key": {"env_vars": ["ENV_VAR"], "strict": False, "type": "env_var"},
+                "api_base_url": None,
+                "model": "model",
+                "dimensions": None,
+                "organization": "my-org",
+                "http_client_kwargs": {"proxy": "http://localhost:8080"},
+                "prefix": "prefix",
+                "suffix": "suffix",
+                "batch_size": 64,
+                "progress_bar": False,
+                "meta_fields_to_embed": ["test_field"],
+                "embedding_separator": " | ",
+                "timeout": 10.0,
+                "max_retries": 2,
+                "raise_on_failure": True,
+            },
+        }
+
+    def test_prepare_texts_to_embed_w_metadata(self):
+        documents = [
+            Document(id=f"{i}", content=f"document number {i}:\ncontent", meta={"meta_field": f"meta_value {i}"})
+            for i in range(5)
+        ]
+
+        embedder = OpenAIDocumentEmbedder(
+            api_key=Secret.from_token("fake-api-key"), meta_fields_to_embed=["meta_field"], embedding_separator=" | "
+        )
+
+        prepared_texts = embedder._prepare_texts_to_embed(documents)
+
+        assert prepared_texts == {
+            "0": "meta_value 0 | document number 0:\ncontent",
+            "1": "meta_value 1 | document number 1:\ncontent",
+            "2": "meta_value 2 | document number 2:\ncontent",
+            "3": "meta_value 3 | document number 3:\ncontent",
+            "4": "meta_value 4 | document number 4:\ncontent",
+        }
+
+    def test_prepare_texts_to_embed_w_suffix(self):
+        documents = [Document(id=f"{i}", content=f"document number {i}") for i in range(5)]
+
+        embedder = OpenAIDocumentEmbedder(
+            api_key=Secret.from_token("fake-api-key"), prefix="my_prefix ", suffix=" my_suffix"
+        )
+
+        prepared_texts = embedder._prepare_texts_to_embed(documents)
+
+        assert prepared_texts == {
+            "0": "my_prefix document number 0 my_suffix",
+            "1": "my_prefix document number 1 my_suffix",
+            "2": "my_prefix document number 2 my_suffix",
+            "3": "my_prefix document number 3 my_suffix",
+            "4": "my_prefix document number 4 my_suffix",
+        }
+
+    def test_run_wrong_input_format(self):
+        embedder = OpenAIDocumentEmbedder(api_key=Secret.from_token("fake-api-key"))
+
+        # wrong formats
+        string_input = "text"
+        list_integers_input = [1, 2, 3]
+
+        with pytest.raises(TypeError, match="OpenAIDocumentEmbedder expects a list of Documents as input"):
+            embedder.run(documents=string_input)  # type: ignore[arg-type]
+
+        with pytest.raises(TypeError, match="OpenAIDocumentEmbedder expects a list of Documents as input"):
+            embedder.run(documents=list_integers_input)  # type: ignore[arg-type]
+
+    def test_run_on_empty_list(self):
+        embedder = OpenAIDocumentEmbedder(api_key=Secret.from_token("fake-api-key"))
+
+        empty_list_input: list[Document] = []
+        result = embedder.run(documents=empty_list_input)
+
+        assert result["documents"] is not None
+        assert not result["documents"]  # empty list
+
+    def test_embed_batch_handles_exceptions_gracefully(self, caplog):
+        embedder = OpenAIDocumentEmbedder(api_key=Secret.from_token("fake_api_key"))
+        embedder.warm_up()
+        assert embedder.client is not None
+        fake_texts_to_embed = {"1": "text1", "2": "text2"}
+        with patch.object(
+            embedder.client.embeddings,
+            "create",
+            side_effect=APIError(message="Mocked error", request=Mock(), body=None),
+        ):
+            embedder._embed_batch(texts_to_embed=fake_texts_to_embed, batch_size=2)
+
+        assert len(caplog.records) == 1
+        assert "Failed embedding of documents 1, 2 caused by Mocked error" in caplog.records[0].msg
+
+    def test_run_handles_exceptions_gracefully(self, caplog):
+        embedder = OpenAIDocumentEmbedder(api_key=Secret.from_token("fake_api_key"), batch_size=1)
+        embedder.warm_up()
+        assert embedder.client is not None
+        docs = [
+            Document(content="I love cheese", meta={"topic": "Cuisine"}),
+            Document(content="A transformer is a deep learning architecture", meta={"topic": "ML"}),
+        ]
+
+        # Create a successful response for the second call
+        successful_response = Mock()
+        successful_response.data = [
+            Mock(embedding=[0.4, 0.5, 0.6])  # Mock embedding for second doc
+        ]
+        successful_response.model = "text-embedding-ada-002"
+        successful_response.usage = {"prompt_tokens": 10, "total_tokens": 10}
+
+        with patch.object(
+            embedder.client.embeddings,
+            "create",
+            side_effect=[
+                APIError(message="Mocked error", request=Mock(), body=None),  # First call fails
+                successful_response,  # Second call succeeds
+            ],
+        ):
+            result = embedder.run(documents=docs)
+        assert len(result["documents"]) == 2
+        assert result["documents"][0].embedding is None
+        assert result["documents"][1].embedding == [0.4, 0.5, 0.6]
+
+    def test_embed_batch_raises_exception_on_failure(self):
+        embedder = OpenAIDocumentEmbedder(api_key=Secret.from_token("fake_api_key"), raise_on_failure=True)
+        embedder.warm_up()
+        assert embedder.client is not None
+        fake_texts_to_embed = {"1": "text1", "2": "text2"}
+        with patch.object(
+            embedder.client.embeddings,
+            "create",
+            side_effect=APIError(message="Mocked error", request=Mock(), body=None),
+        ):
+            with pytest.raises(APIError, match="Mocked error"):
+                embedder._embed_batch(texts_to_embed=fake_texts_to_embed, batch_size=2)
+
+    @pytest.mark.skipif(os.environ.get("OPENAI_API_KEY", "") == "", reason="OPENAI_API_KEY is not set")
+    @pytest.mark.integration
+    def test_run(self):
+        docs = [
+            Document(content="I love cheese", meta={"topic": "Cuisine"}),
+            Document(content="A transformer is a deep learning architecture", meta={"topic": "ML"}),
+        ]
+
+        model = "text-embedding-ada-002"
+
+        embedder = OpenAIDocumentEmbedder(model=model, meta_fields_to_embed=["topic"], embedding_separator=" | ")
+
+        result = embedder.run(documents=docs)
+        assert embedder.client is not None
+        documents_with_embeddings = result["documents"]
+
+        assert isinstance(documents_with_embeddings, list)
+        assert len(documents_with_embeddings) == len(docs)
+        for doc, new_doc in zip(docs, documents_with_embeddings, strict=True):
+            assert doc.embedding is None
+            assert new_doc is not doc
+            assert isinstance(new_doc, Document)
+            assert isinstance(new_doc.embedding, list)
+            assert len(new_doc.embedding) == 1536
+            assert all(isinstance(x, float) for x in new_doc.embedding)
+
+        assert "text" in result["meta"]["model"] and "ada" in result["meta"]["model"], (
+            "The model name does not contain 'text' and 'ada'"
+        )
+
+        assert result["meta"]["usage"] == {"prompt_tokens": 15, "total_tokens": 15}, "Usage information does not match"
+
+    @pytest.mark.skipif(os.environ.get("OPENAI_API_KEY", "") == "", reason="OPENAI_API_KEY is not set")
+    @pytest.mark.integration
+    @pytest.mark.asyncio
+    async def test_run_async(self):
+        embedder = OpenAIDocumentEmbedder(
+            model="text-embedding-ada-002", meta_fields_to_embed=["topic"], embedding_separator=" | "
+        )
+        docs = [
+            Document(content="I love cheese", meta={"topic": "Cuisine"}),
+            Document(content="A transformer is a deep learning architecture", meta={"topic": "ML"}),
+        ]
+
+        result = await embedder.run_async(documents=docs)
+        assert embedder.async_client is not None
+        documents_with_embeddings = result["documents"]
+
+        assert isinstance(documents_with_embeddings, list)
+        assert len(documents_with_embeddings) == len(docs)
+        for doc, new_doc in zip(docs, documents_with_embeddings, strict=True):
+            assert doc.embedding is None
+            assert new_doc is not doc
+            assert isinstance(new_doc, Document)
+            assert isinstance(new_doc.embedding, list)
+            assert len(new_doc.embedding) == 1536
+            assert all(isinstance(x, float) for x in new_doc.embedding)
+
+        assert "text" in result["meta"]["model"] and "ada" in result["meta"]["model"], (
+            "The model name does not contain 'text' and 'ada'"
+        )
+
+        assert result["meta"]["usage"] == {"prompt_tokens": 15, "total_tokens": 15}, "Usage information does not match"
+
+        # Close async client; suppress RuntimeError if the event loop is already closed
+        with contextlib.suppress(RuntimeError):
+            await embedder.close_async()
+
+
+@pytest.fixture
+def mock_openai_clients(monkeypatch):
+    monkeypatch.setenv("OPENAI_API_KEY", "fake")
+    sync_cls = MagicMock(name="OpenAI")
+    async_cls = MagicMock(name="AsyncOpenAI")
+    async_cls.return_value.close = AsyncMock()
+    monkeypatch.setattr(openai_document_embedder_module, "OpenAI", sync_cls)
+    monkeypatch.setattr(openai_document_embedder_module, "AsyncOpenAI", async_cls)
+    return sync_cls, async_cls
+
+
+class TestComponentLifecycle:
+    def test_warm_up_uses_default_timeout_and_max_retries(self, monkeypatch):
+        monkeypatch.setenv("OPENAI_API_KEY", "fake-api-key")
+        embedder = OpenAIDocumentEmbedder()
+        embedder.warm_up()
+        assert embedder.client is not None
+        assert embedder.client.max_retries == 5
+        assert embedder.client.timeout == 30.0
+
+    def test_warm_up_uses_timeout_and_max_retries_from_parameters(self):
+        embedder = OpenAIDocumentEmbedder(api_key=Secret.from_token("fake-api-key"), timeout=40.0, max_retries=1)
+        embedder.warm_up()
+        assert embedder.client is not None
+        assert embedder.client.max_retries == 1
+        assert embedder.client.timeout == 40.0
+
+    def test_warm_up_uses_timeout_and_max_retries_from_env_vars(self, monkeypatch):
+        monkeypatch.setenv("OPENAI_TIMEOUT", "100")
+        monkeypatch.setenv("OPENAI_MAX_RETRIES", "10")
+        embedder = OpenAIDocumentEmbedder(api_key=Secret.from_token("fake-api-key"))
+        embedder.warm_up()
+        assert embedder.client is not None
+        assert embedder.client.max_retries == 10
+        assert embedder.client.timeout == 100.0
+
+    def test_key_resolved_at_warm_up_not_init(self, monkeypatch):
+        monkeypatch.delenv("OPENAI_API_KEY", raising=False)
+        embedder = OpenAIDocumentEmbedder()
+        with pytest.raises(ValueError, match="None of the .* environment variables are set"):
+            embedder.warm_up()
+
+    def test_sync_lifecycle(self, mock_openai_clients):
+        sync_cls, _ = mock_openai_clients
+        sync_client = sync_cls.return_value
+        embedder = OpenAIDocumentEmbedder()
+        assert embedder.client is None
+        assert embedder.async_client is None
+
+        embedder.warm_up()
+        assert embedder.client is sync_cls.return_value
+        assert embedder.async_client is None
+
+        embedder.close()
+        sync_client.close.assert_called_once()
+        assert embedder.client is None
+
+    async def test_async_lifecycle(self, mock_openai_clients):
+        _, async_cls = mock_openai_clients
+        async_client = async_cls.return_value
+        embedder = OpenAIDocumentEmbedder()
+
+        await embedder.warm_up_async()
+        assert embedder.async_client is async_cls.return_value
+        assert embedder.client is None
+
+        await embedder.close_async()
+        async_client.close.assert_awaited_once()
+        assert embedder.async_client is None
+
+    async def test_close_is_safe_without_warm_up(self, mock_openai_clients):
+        embedder = OpenAIDocumentEmbedder()
+        embedder.close()
+        await embedder.close_async()
+        assert embedder.client is None
+        assert embedder.async_client is None
+
+    async def test_close_and_close_async_are_independent(self, mock_openai_clients):
+        embedder = OpenAIDocumentEmbedder()
+        embedder.warm_up()
+        await embedder.warm_up_async()
+
+        embedder.close()
+        assert embedder.client is None
+        assert embedder.async_client is not None
+
+        await embedder.close_async()
+        assert embedder.async_client is None

test/components/embedders/test_openai_text_embedder.py

@@ -0,0 +1,316 @@
+# SPDX-FileCopyrightText: 2022-present deepset GmbH <info@deepset.ai>
+#
+# SPDX-License-Identifier: Apache-2.0
+
+import contextlib
+import os
+from unittest.mock import AsyncMock, MagicMock
+
+import pytest
+from openai.types import CreateEmbeddingResponse, Embedding
+from openai.types.create_embedding_response import Usage
+
+import haystack.components.embedders.openai_text_embedder as openai_text_embedder_module
+from haystack.components.embedders.openai_text_embedder import OpenAITextEmbedder
+from haystack.utils.auth import Secret
+
+
+class TestOpenAITextEmbedder:
+    def test_init_default(self, monkeypatch):
+        monkeypatch.setenv("OPENAI_API_KEY", "fake-api-key")
+        embedder = OpenAITextEmbedder()
+
+        assert embedder.api_key.resolve_value() == "fake-api-key"
+        assert embedder.model == "text-embedding-ada-002"
+        assert embedder.api_base_url is None
+        assert embedder.organization is None
+        assert embedder.prefix == ""
+        assert embedder.suffix == ""
+        assert embedder.timeout is None
+        assert embedder.max_retries is None
+        assert embedder.client is None
+        assert embedder.async_client is None
+
+    def test_init_with_parameters(self, monkeypatch):
+        monkeypatch.setenv("OPENAI_TIMEOUT", "100")
+        monkeypatch.setenv("OPENAI_MAX_RETRIES", "10")
+        embedder = OpenAITextEmbedder(
+            api_key=Secret.from_token("fake-api-key"),
+            model="model",
+            api_base_url="https://my-custom-base-url.com",
+            organization="fake-organization",
+            prefix="prefix",
+            suffix="suffix",
+            timeout=40.0,
+            max_retries=1,
+        )
+        assert embedder.api_key.resolve_value() == "fake-api-key"
+        assert embedder.model == "model"
+        assert embedder.api_base_url == "https://my-custom-base-url.com"
+        assert embedder.organization == "fake-organization"
+        assert embedder.prefix == "prefix"
+        assert embedder.suffix == "suffix"
+        assert embedder.timeout == 40.0
+        assert embedder.max_retries == 1
+        assert embedder.client is None
+        assert embedder.async_client is None
+
+    def test_init_with_parameters_and_env_vars(self, monkeypatch):
+        monkeypatch.setenv("OPENAI_TIMEOUT", "100")
+        monkeypatch.setenv("OPENAI_MAX_RETRIES", "10")
+        embedder = OpenAITextEmbedder(
+            api_key=Secret.from_token("fake-api-key"),
+            model="model",
+            api_base_url="https://my-custom-base-url.com",
+            organization="fake-organization",
+            prefix="prefix",
+            suffix="suffix",
+        )
+        assert embedder.api_key.resolve_value() == "fake-api-key"
+        assert embedder.model == "model"
+        assert embedder.api_base_url == "https://my-custom-base-url.com"
+        assert embedder.organization == "fake-organization"
+        assert embedder.prefix == "prefix"
+        assert embedder.suffix == "suffix"
+        assert embedder.timeout is None
+        assert embedder.max_retries is None
+        assert embedder.client is None
+        assert embedder.async_client is None
+
+    def test_to_dict(self, monkeypatch):
+        monkeypatch.setenv("OPENAI_API_KEY", "fake-api-key")
+        component = OpenAITextEmbedder()
+        data = component.to_dict()
+        assert data == {
+            "type": "haystack.components.embedders.openai_text_embedder.OpenAITextEmbedder",
+            "init_parameters": {
+                "api_key": {"env_vars": ["OPENAI_API_KEY"], "strict": True, "type": "env_var"},
+                "api_base_url": None,
+                "dimensions": None,
+                "model": "text-embedding-ada-002",
+                "organization": None,
+                "http_client_kwargs": None,
+                "prefix": "",
+                "suffix": "",
+                "timeout": None,
+                "max_retries": None,
+            },
+        }
+
+    def test_to_dict_with_custom_init_parameters(self, monkeypatch):
+        monkeypatch.setenv("ENV_VAR", "fake-api-key")
+        component = OpenAITextEmbedder(
+            api_key=Secret.from_env_var("ENV_VAR", strict=False),
+            model="model",
+            api_base_url="https://my-custom-base-url.com",
+            organization="fake-organization",
+            prefix="prefix",
+            suffix="suffix",
+            timeout=10.0,
+            max_retries=2,
+            http_client_kwargs={"proxy": "http://localhost:8080"},
+        )
+        data = component.to_dict()
+        assert data == {
+            "type": "haystack.components.embedders.openai_text_embedder.OpenAITextEmbedder",
+            "init_parameters": {
+                "api_key": {"env_vars": ["ENV_VAR"], "strict": False, "type": "env_var"},
+                "api_base_url": "https://my-custom-base-url.com",
+                "model": "model",
+                "dimensions": None,
+                "organization": "fake-organization",
+                "http_client_kwargs": {"proxy": "http://localhost:8080"},
+                "prefix": "prefix",
+                "suffix": "suffix",
+                "timeout": 10.0,
+                "max_retries": 2,
+            },
+        }
+
+    def test_from_dict(self, monkeypatch):
+        monkeypatch.setenv("OPENAI_API_KEY", "fake-api-key")
+        data = {
+            "type": "haystack.components.embedders.openai_text_embedder.OpenAITextEmbedder",
+            "init_parameters": {
+                "api_key": {"env_vars": ["OPENAI_API_KEY"], "strict": True, "type": "env_var"},
+                "model": "text-embedding-ada-002",
+                "api_base_url": "https://my-custom-base-url.com",
+                "organization": "fake-organization",
+                "http_client_kwargs": None,
+                "prefix": "prefix",
+                "suffix": "suffix",
+            },
+        }
+        component = OpenAITextEmbedder.from_dict(data)
+        assert component.api_key.resolve_value() == "fake-api-key"
+        assert component.model == "text-embedding-ada-002"
+        assert component.api_base_url == "https://my-custom-base-url.com"
+        assert component.organization == "fake-organization"
+        assert component.http_client_kwargs is None
+        assert component.prefix == "prefix"
+        assert component.suffix == "suffix"
+
+    def test_prepare_input(self, monkeypatch):
+        monkeypatch.setenv("OPENAI_API_KEY", "fake-api-key")
+        embedder = OpenAITextEmbedder(dimensions=1536)
+
+        inp = "The food was delicious"
+        prepared_input = embedder._prepare_input(inp)
+        assert prepared_input == {
+            "model": "text-embedding-ada-002",
+            "input": "The food was delicious",
+            "encoding_format": "float",
+            "dimensions": 1536,
+        }
+
+    def test_prepare_output(self, monkeypatch):
+        monkeypatch.setenv("OPENAI_API_KEY", "fake-api-key")
+
+        response = CreateEmbeddingResponse(
+            data=[Embedding(embedding=[0.1, 0.2, 0.3], index=0, object="embedding")],
+            model="text-embedding-ada-002",
+            object="list",
+            usage=Usage(prompt_tokens=6, total_tokens=6),
+        )
+
+        embedder = OpenAITextEmbedder()
+        result = embedder._prepare_output(result=response)
+        assert result == {
+            "embedding": [0.1, 0.2, 0.3],
+            "meta": {"model": "text-embedding-ada-002", "usage": {"prompt_tokens": 6, "total_tokens": 6}},
+        }
+
+    def test_run_wrong_input_format(self):
+        embedder = OpenAITextEmbedder(api_key=Secret.from_token("fake-api-key"))
+
+        list_integers_input = [1, 2, 3]
+
+        with pytest.raises(TypeError, match="OpenAITextEmbedder expects a string as an input"):
+            embedder.run(text=list_integers_input)  # type: ignore[arg-type]
+
+    @pytest.mark.skipif(os.environ.get("OPENAI_API_KEY", "") == "", reason="OPENAI_API_KEY is not set")
+    @pytest.mark.integration
+    def test_run(self):
+        model = "text-embedding-ada-002"
+
+        embedder = OpenAITextEmbedder(model=model, prefix="prefix ", suffix=" suffix")
+        result = embedder.run(text="The food was delicious")
+
+        assert len(result["embedding"]) == 1536
+        assert all(isinstance(x, float) for x in result["embedding"])
+
+        assert "text" in result["meta"]["model"] and "ada" in result["meta"]["model"], (
+            "The model name does not contain 'text' and 'ada'"
+        )
+
+        assert result["meta"]["usage"] == {"prompt_tokens": 6, "total_tokens": 6}, "Usage information does not match"
+
+    @pytest.mark.asyncio
+    @pytest.mark.skipif(os.environ.get("OPENAI_API_KEY", "") == "", reason="OPENAI_API_KEY is not set")
+    @pytest.mark.integration
+    async def test_run_async(self):
+        embedder = OpenAITextEmbedder(model="text-embedding-ada-002", prefix="prefix ", suffix=" suffix")
+        result = await embedder.run_async(text="The food was delicious")
+
+        assert len(result["embedding"]) == 1536
+        assert all(isinstance(x, float) for x in result["embedding"])
+
+        assert "text" in result["meta"]["model"] and "ada" in result["meta"]["model"], (
+            "The model name does not contain 'text' and 'ada'"
+        )
+
+        assert result["meta"]["usage"] == {"prompt_tokens": 6, "total_tokens": 6}, "Usage information does not match"
+
+        # Close async client; suppress RuntimeError if the event loop is already closed
+        with contextlib.suppress(RuntimeError):
+            await embedder.close_async()
+
+
+@pytest.fixture
+def mock_openai_clients(monkeypatch):
+    monkeypatch.setenv("OPENAI_API_KEY", "fake")
+    sync_cls = MagicMock(name="OpenAI")
+    async_cls = MagicMock(name="AsyncOpenAI")
+    async_cls.return_value.close = AsyncMock()
+    monkeypatch.setattr(openai_text_embedder_module, "OpenAI", sync_cls)
+    monkeypatch.setattr(openai_text_embedder_module, "AsyncOpenAI", async_cls)
+    return sync_cls, async_cls
+
+
+class TestComponentLifecycle:
+    def test_warm_up_uses_default_timeout_and_max_retries(self, monkeypatch):
+        monkeypatch.setenv("OPENAI_API_KEY", "fake-api-key")
+        embedder = OpenAITextEmbedder()
+        embedder.warm_up()
+        assert embedder.client is not None
+        assert embedder.client.max_retries == 5
+        assert embedder.client.timeout == 30.0
+
+    def test_warm_up_uses_timeout_and_max_retries_from_parameters(self):
+        embedder = OpenAITextEmbedder(api_key=Secret.from_token("fake-api-key"), timeout=40.0, max_retries=1)
+        embedder.warm_up()
+        assert embedder.client is not None
+        assert embedder.client.max_retries == 1
+        assert embedder.client.timeout == 40.0
+
+    def test_warm_up_uses_timeout_and_max_retries_from_env_vars(self, monkeypatch):
+        monkeypatch.setenv("OPENAI_TIMEOUT", "100")
+        monkeypatch.setenv("OPENAI_MAX_RETRIES", "10")
+        embedder = OpenAITextEmbedder(api_key=Secret.from_token("fake-api-key"))
+        embedder.warm_up()
+        assert embedder.client is not None
+        assert embedder.client.max_retries == 10
+        assert embedder.client.timeout == 100.0
+
+    def test_key_resolved_at_warm_up_not_init(self, monkeypatch):
+        monkeypatch.delenv("OPENAI_API_KEY", raising=False)
+        embedder = OpenAITextEmbedder()
+        with pytest.raises(ValueError, match="None of the .* environment variables are set"):
+            embedder.warm_up()
+
+    def test_sync_lifecycle(self, mock_openai_clients):
+        sync_cls, _ = mock_openai_clients
+        sync_client = sync_cls.return_value
+        embedder = OpenAITextEmbedder()
+        assert embedder.client is None
+        assert embedder.async_client is None
+
+        embedder.warm_up()
+        assert embedder.client is sync_cls.return_value
+        assert embedder.async_client is None
+
+        embedder.close()
+        sync_client.close.assert_called_once()
+        assert embedder.client is None
+
+    async def test_async_lifecycle(self, mock_openai_clients):
+        _, async_cls = mock_openai_clients
+        async_client = async_cls.return_value
+        embedder = OpenAITextEmbedder()
+
+        await embedder.warm_up_async()
+        assert embedder.async_client is async_cls.return_value
+        assert embedder.client is None
+
+        await embedder.close_async()
+        async_client.close.assert_awaited_once()
+        assert embedder.async_client is None
+
+    async def test_close_is_safe_without_warm_up(self, mock_openai_clients):
+        embedder = OpenAITextEmbedder()
+        embedder.close()
+        await embedder.close_async()
+        assert embedder.client is None
+        assert embedder.async_client is None
+
+    async def test_close_and_close_async_are_independent(self, mock_openai_clients):
+        embedder = OpenAITextEmbedder()
+        embedder.warm_up()
+        await embedder.warm_up_async()
+
+        embedder.close()
+        assert embedder.client is None
+        assert embedder.async_client is not None
+
+        await embedder.close_async()
+        assert embedder.async_client is None