cohere-haystack 10.0.1__tar.gz → 10.1.0__tar.gz

{cohere_haystack-10.0.1 → cohere_haystack-10.1.0}/CHANGELOG.md

@@ -1,5 +1,12 @@
 # Changelog
 
+## [integrations/cohere-v10.0.1] - 2026-09-10
+
+### 🐛 Bug Fixes
+
+- Batch CohereDocumentEmbedder.run_async like the sync path (#3926)
+
+
 ## [integrations/cohere-v10.0.0] - 2026-09-04
 
 ### 🚜 Refactor

{cohere_haystack-10.0.1 → cohere_haystack-10.1.0}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.5
 Name: cohere-haystack
-Version: 10.0.1
+Version: 10.1.0
 Project-URL: Documentation, https://github.com/deepset-ai/haystack-core-integrations/tree/main/integrations/cohere#readme
 Project-URL: Issues, https://github.com/deepset-ai/haystack-core-integrations/issues
 Project-URL: Source, https://github.com/deepset-ai/haystack-core-integrations/tree/main/integrations/cohere

{cohere_haystack-10.0.1 → cohere_haystack-10.1.0}/pyproject.toml

@@ -72,7 +72,8 @@ integration-cov-append-retry = 'pytest --cov=haystack_integrations --cov-append
 
 types = """mypy -p haystack_integrations.components.embedders.cohere \
 -p haystack_integrations.components.generators.cohere \
--p haystack_integrations.components.rankers.cohere {args}"""
+-p haystack_integrations.components.rankers.cohere \
+-p haystack_integrations.utils.cohere {args}"""
 
 [tool.mypy]
 install_types = true

{cohere_haystack-10.0.1 → cohere_haystack-10.1.0}/src/haystack_integrations/components/embedders/cohere/document_embedder.py

@@ -8,6 +8,7 @@ from haystack import Document, component, default_from_dict, default_to_dict
 from haystack.utils import Secret, deserialize_secrets_inplace
 
 from cohere import AsyncClientV2, ClientV2
+from haystack_integrations.utils.cohere import validate_api_base_url
 
 from .embedding_types import EmbeddingTypes
 from .utils import get_async_response, get_response
@@ -68,7 +69,7 @@ class CohereDocumentEmbedder:
             Read [Cohere documentation](https://docs.cohere.com/docs/models#embed) for a list of all supported models.
         :param input_type: specifies the type of input you're giving to the model. Supported values are
             "search_document", "search_query", "classification" and "clustering".
-        :param api_base_url: the Cohere API Base url.
+        :param api_base_url: the Cohere API Base url. The Cohere client appends the endpoint path to it.
         :param truncate: truncate embeddings that are too long from start or end, ("NONE"|"START"|"END").
             Passing "START" will discard the start of the input. "END" will discard the end of the input. In both
             cases, input is discarded until the remaining input is exactly the maximum input token length for the model.
@@ -81,7 +82,10 @@ class CohereDocumentEmbedder:
         :param embedding_separator: separator used to concatenate the meta fields to the Document text.
         :param embedding_type: the type of embeddings to return. Defaults to float embeddings.
             Note that int8, uint8, binary, and ubinary are only valid for v3 models.
+
+        :raises ValueError: If `api_base_url` is a full endpoint URL rather than a base URL.
         """
+        validate_api_base_url(api_base_url)
 
         self.api_key = api_key
         self.model = model

{cohere_haystack-10.0.1 → cohere_haystack-10.1.0}/src/haystack_integrations/components/embedders/cohere/document_image_embedder.py

@@ -17,6 +17,7 @@ from haystack.utils.auth import Secret, deserialize_secrets_inplace
 from tqdm import tqdm
 
 from cohere import AsyncClientV2, ClientV2
+from haystack_integrations.utils.cohere import validate_api_base_url
 
 from .embedding_types import EmbeddingTypes
 
@@ -101,7 +102,7 @@ class CohereDocumentImageEmbedder:
             The Cohere model to use for calculating embeddings.
             Read [Cohere documentation](https://docs.cohere.com/docs/models#embed) for a list of all supported models.
         :param api_base_url:
-            The Cohere API base URL.
+            The Cohere API base URL. The Cohere client appends the endpoint path to it.
         :param timeout:
             Request timeout in seconds.
         :param embedding_dimension:
@@ -114,7 +115,10 @@ class CohereDocumentImageEmbedder:
         :param progress_bar:
             Whether to show a progress bar or not. Can be helpful to disable in production deployments
             to keep the logs clean.
+
+        :raises ValueError: If `api_base_url` is a full endpoint URL rather than a base URL.
         """
+        validate_api_base_url(api_base_url)
 
         self.file_path_meta_field = file_path_meta_field
         self.root_path = root_path or ""

{cohere_haystack-10.0.1 → cohere_haystack-10.1.0}/src/haystack_integrations/components/embedders/cohere/text_embedder.py

@@ -7,6 +7,7 @@ from haystack import component, default_from_dict, default_to_dict
 from haystack.utils import Secret, deserialize_secrets_inplace
 
 from cohere import AsyncClientV2, ClientV2
+from haystack_integrations.utils.cohere import validate_api_base_url
 
 from .embedding_types import EmbeddingTypes
 from .utils import get_async_response, get_response
@@ -60,7 +61,7 @@ class CohereTextEmbedder:
             Read [Cohere documentation](https://docs.cohere.com/docs/models#embed) for a list of all supported models.
         :param input_type: specifies the type of input you're giving to the model. Supported values are
         "search_document", "search_query", "classification" and "clustering".
-        :param api_base_url: the Cohere API Base url.
+        :param api_base_url: the Cohere API Base url. The Cohere client appends the endpoint path to it.
         :param truncate: truncate embeddings that are too long from start or end, ("NONE"|"START"|"END").
             Passing "START" will discard the start of the input. "END" will discard the end of the input. In both
             cases, input is discarded until the remaining input is exactly the maximum input token length for the model.
@@ -68,7 +69,10 @@ class CohereTextEmbedder:
         :param timeout: request timeout in seconds.
         :param embedding_type: the type of embeddings to return. Defaults to float embeddings.
             Note that int8, uint8, binary, and ubinary are only valid for v3 models.
+
+        :raises ValueError: If `api_base_url` is a full endpoint URL rather than a base URL.
         """
+        validate_api_base_url(api_base_url)
 
         self.api_key = api_key
         self.model = model

{cohere_haystack-10.0.1 → cohere_haystack-10.1.0}/src/haystack_integrations/components/generators/cohere/chat/chat_generator.py

@@ -29,6 +29,7 @@ from httpx import AsyncHTTPTransport, HTTPTransport
 from httpx import Client as HTTPXClient
 
 from cohere import AsyncClientV2, ChatResponse, ClientV2, StreamedChatResponseV2
+from haystack_integrations.utils.cohere import validate_api_base_url
 
 logger = logging.getLogger(__name__)
 
@@ -544,7 +545,7 @@ class CohereChatGenerator:
         :param streaming_callback: A callback function that is called when a new token is received from the stream.
             The callback function accepts [StreamingChunk](https://docs.haystack.deepset.ai/docs/data-classes#streamingchunk)
             as an argument.
-        :param api_base_url: The base URL of the Cohere API.
+        :param api_base_url: The base URL of the Cohere API. The Cohere client appends the endpoint path to it.
         :param generation_kwargs: Other parameters to use for the model during generation. For a list of parameters,
             see [Cohere Chat endpoint](https://docs.cohere.com/reference/chat).
             Some of the parameters are:
@@ -564,11 +565,13 @@ class CohereChatGenerator:
             Maximum number of retries to attempt for failed requests. If not set, it defaults to the default set by
             the Cohere client.
 
+        :raises ValueError: If `api_base_url` is a full endpoint URL rather than a base URL.
         """
         _check_duplicate_tool_names(flatten_tools_or_toolsets(tools))
 
         if not api_base_url:
             api_base_url = "https://api.cohere.com"
+        validate_api_base_url(api_base_url)
 
         self.api_key = api_key
         self.model = model

{cohere_haystack-10.0.1 → cohere_haystack-10.1.0}/src/haystack_integrations/components/rankers/cohere/ranker.py

@@ -5,6 +5,7 @@ from haystack import Document, component, default_from_dict, default_to_dict, lo
 from haystack.utils import Secret, deserialize_secrets_inplace
 
 from cohere import AsyncClientV2, ClientV2
+from haystack_integrations.utils.cohere import validate_api_base_url
 
 logger = logging.getLogger(__name__)
 
@@ -48,19 +49,22 @@ class CohereRanker:
         :param model: Cohere model name. Check the list of supported models in the [Cohere documentation](https://docs.cohere.com/docs/models).
         :param top_k: The maximum number of documents to return.
         :param api_key: Cohere API key.
-        :param api_base_url: the base URL of the Cohere API.
+        :param api_base_url: the base URL of the Cohere API. The Cohere client appends the endpoint path to it.
         :param meta_fields_to_embed: List of meta fields that should be concatenated
             with the document content for reranking.
         :param meta_data_separator: Separator used to concatenate the meta fields
             to the Document content.
         :param max_tokens_per_doc: The maximum number of tokens to embed for each document defaults to 4096.
 
-        :raises ValueError: If `top_k` is not > 0.
+        :raises ValueError: If `top_k` is not > 0, or if `api_base_url` is a full endpoint URL rather than a
+            base URL.
         """
         if top_k <= 0:
             msg = f"top_k must be > 0, but got {top_k}"
             raise ValueError(msg)
 
+        validate_api_base_url(api_base_url)
+
         self.model_name = model
         self.api_key = api_key
         self.api_base_url = api_base_url

cohere_haystack-10.1.0/src/haystack_integrations/utils/cohere/__init__.py

@@ -0,0 +1,6 @@
+# SPDX-FileCopyrightText: 2023-present deepset GmbH <info@deepset.ai>
+#
+# SPDX-License-Identifier: Apache-2.0
+from .api_base_url import validate_api_base_url
+
+__all__ = ["validate_api_base_url"]

cohere_haystack-10.1.0/src/haystack_integrations/utils/cohere/api_base_url.py

@@ -0,0 +1,29 @@
+# SPDX-FileCopyrightText: 2023-present deepset GmbH <info@deepset.ai>
+#
+# SPDX-License-Identifier: Apache-2.0
+import re
+
+# Cohere clients build a request URL by appending the endpoint path to the configured base URL, so a base URL that
+# already ends with one of these paths produces a doubled path and a generic API error at request time.
+_COHERE_ENDPOINT_PATH = re.compile(r"/v\d+/(?:rerank|embed|chat)$")
+
+
+def validate_api_base_url(api_base_url: str) -> None:
+    """
+    Check that `api_base_url` is a base URL rather than a full endpoint URL.
+
+    :param api_base_url: The base URL to validate.
+    :raises ValueError: If `api_base_url` ends with a Cohere endpoint path, which the Cohere client appends itself.
+    """
+    trimmed = api_base_url.rstrip("/")
+    match = _COHERE_ENDPOINT_PATH.search(trimmed)
+    if match is None:
+        return
+
+    base = trimmed[: match.start()]
+    hint = f" and pass '{base}' instead" if base else ""
+    msg = (
+        f"'api_base_url' must be a base URL, not a full endpoint URL, but got {api_base_url!r}. "
+        f"The Cohere client appends the endpoint path itself, so remove the trailing '{match.group()}'{hint}."
+    )
+    raise ValueError(msg)

cohere_haystack-10.1.0/tests/test_api_base_url.py

@@ -0,0 +1,75 @@
+# SPDX-FileCopyrightText: 2023-present deepset GmbH <info@deepset.ai>
+#
+# SPDX-License-Identifier: Apache-2.0
+import pytest
+
+from haystack_integrations.components.embedders.cohere import (
+    CohereDocumentEmbedder,
+    CohereDocumentImageEmbedder,
+    CohereTextEmbedder,
+)
+from haystack_integrations.components.generators.cohere import CohereChatGenerator
+from haystack_integrations.components.rankers.cohere import CohereRanker
+from haystack_integrations.utils.cohere import validate_api_base_url
+
+
+class TestValidateApiBaseUrl:
+    @pytest.mark.parametrize(
+        "api_base_url",
+        [
+            "https://api.cohere.com",
+            "https://api.cohere.com/",
+            "https://resource.services.ai.azure.com/models",
+            # only a trailing endpoint path is rejected, not one in the middle of the base path
+            "https://gateway.example.com/v2/rerank/proxy",
+        ],
+    )
+    def test_accepts_base_urls(self, api_base_url):
+        validate_api_base_url(api_base_url)
+
+    @pytest.mark.parametrize(
+        "api_base_url",
+        [
+            "https://api.cohere.com/v1/rerank",
+            "https://api.cohere.com/v2/embed",
+            "https://api.cohere.com/v2/chat",
+            # a future API version is rejected without needing a new entry in the pattern
+            "https://api.cohere.com/v3/rerank",
+        ],
+    )
+    def test_rejects_endpoint_urls(self, api_base_url):
+        with pytest.raises(ValueError, match="must be a base URL"):
+            validate_api_base_url(api_base_url)
+
+    @pytest.mark.parametrize("suffix", ["", "/", "//"])
+    def test_error_names_the_endpoint_path_and_the_corrected_url(self, suffix):
+        url = f"https://resource.services.ai.azure.com/models/v2/rerank{suffix}"
+        with pytest.raises(ValueError) as exc_info:
+            validate_api_base_url(url)
+
+        message = str(exc_info.value)
+        assert "remove the trailing '/v2/rerank'" in message
+        assert "pass 'https://resource.services.ai.azure.com/models' instead" in message
+
+    def test_omits_the_hint_when_nothing_would_remain(self):
+        with pytest.raises(ValueError) as exc_info:
+            validate_api_base_url("/v2/rerank")
+
+        assert "instead" not in str(exc_info.value)
+
+
+@pytest.mark.parametrize(
+    "component_cls",
+    [
+        CohereRanker,
+        CohereTextEmbedder,
+        CohereDocumentEmbedder,
+        CohereDocumentImageEmbedder,
+        CohereChatGenerator,
+    ],
+)
+def test_components_reject_full_endpoint_urls(component_cls, monkeypatch):
+    monkeypatch.setenv("COHERE_API_KEY", "test-api-key")
+
+    with pytest.raises(ValueError, match="must be a base URL"):
+        component_cls(api_base_url="https://resource.services.ai.azure.com/models/v2/rerank")