fastembed-haystack 1.2.0__tar.gz → 1.4.0__tar.gz

fastembed_haystack-1.4.0/CHANGELOG.md

@@ -0,0 +1,72 @@
+# Changelog
+
+## [integrations/fastembed-v1.3.0] - 2024-10-07
+
+### 🚀 Features
+
+- Introduce `model_kwargs` in Sparse Embedders (can be used for BM25 parameters) (#1126)
+
+### 🧪 Testing
+
+- Do not retry tests in `hatch run test` command (#954)
+
+### ⚙️ Miscellaneous Tasks
+
+- Retry tests to reduce flakyness (#836)
+- Update ruff invocation to include check parameter (#853)
+- Update ruff linting scripts and settings (#1105)
+
+### Fix
+
+- Typo on Sparse embedders. The parameter should be "progress_bar" … (#814)
+
+## [integrations/fastembed-v1.1.0] - 2024-05-15
+
+## [integrations/fastembed-v1.0.0] - 2024-05-06
+
+## [integrations/fastembed-v0.1.0] - 2024-04-10
+
+### 🚀 Features
+
+- *(FastEmbed)* Support for SPLADE Sparse Embedder (#579)
+
+### 📚 Documentation
+
+- Disable-class-def (#556)
+
+## [integrations/fastembed-v0.0.6] - 2024-03-07
+
+### 📚 Documentation
+
+- Review and normalize docstrings - `integrations.fastembed` (#519)
+- Small consistency improvements (#536)
+
+## [integrations/fastembed-v0.0.5] - 2024-02-20
+
+### 🐛 Bug Fixes
+
+- Fix order of API docs (#447)
+
+This PR will also push the docs to Readme
+
+### 📚 Documentation
+
+- Update category slug (#442)
+
+## [integrations/fastembed-v0.0.4] - 2024-02-16
+
+## [integrations/fastembed-v0.0.3] - 2024-02-12
+
+### 🐛 Bug Fixes
+
+- From numpy float to float (#391)
+
+### 📚 Documentation
+
+- Update paths and titles (#397)
+
+## [integrations/fastembed-v0.0.2] - 2024-02-11
+
+## [integrations/fastembed-v0.0.1] - 2024-02-10
+
+<!-- generated by git-cliff -->

{fastembed_haystack-1.2.0 → fastembed_haystack-1.4.0}/PKG-INFO

@@ -1,13 +1,12 @@
 Metadata-Version: 2.3
 Name: fastembed-haystack
-Version: 1.2.0
+Version: 1.4.0
 Summary: Haystack 2.x component to embed strings and Documents using fastembed embedding model
 Project-URL: Source, https://github.com/deepset-ai/haystack-core-integrations
 Project-URL: Documentation, https://github.com/deepset-ai/haystack-core-integrations/blob/main/integrations/fastembed/README.md
 Project-URL: Issues, https://github.com/deepset-ai/haystack-core-integrations/issues
 Author-email: deepset GmbH <info@deepset.ai>
-License-Expression: Apache-2.0
-License-File: LICENSE.txt
+License: Apache-2.0
 Classifier: Development Status :: 4 - Beta
 Classifier: License :: OSI Approved :: Apache Software License
 Classifier: Programming Language :: Python
@@ -19,7 +18,7 @@ Classifier: Programming Language :: Python :: 3.12
 Classifier: Programming Language :: Python :: Implementation :: CPython
 Classifier: Programming Language :: Python :: Implementation :: PyPy
 Requires-Python: >=3.8
-Requires-Dist: fastembed>=0.2.5
+Requires-Dist: fastembed>=0.4.2
 Requires-Dist: haystack-ai>=2.0.1
 Description-Content-Type: text/markdown
 

{fastembed_haystack-1.2.0 → fastembed_haystack-1.4.0}/examples/example.py

@@ -1,6 +1,7 @@
 from haystack import Document, Pipeline
 from haystack.components.retrievers.in_memory import InMemoryEmbeddingRetriever
 from haystack.document_stores.in_memory import InMemoryDocumentStore
+
 from haystack_integrations.components.embedders.fastembed import FastembedDocumentEmbedder, FastembedTextEmbedder
 
 document_store = InMemoryDocumentStore(embedding_similarity_function="cosine")

fastembed_haystack-1.4.0/examples/ranker_example.py

@@ -0,0 +1,22 @@
+from haystack import Document
+
+from haystack_integrations.components.rankers.fastembed import FastembedRanker
+
+query = "Who is maintaining Qdrant?"
+documents = [
+    Document(
+        content="This is built to be faster and lighter than other embedding libraries e.g. Transformers, Sentence-Transformers, etc."
+    ),
+    Document(content="fastembed is supported by and maintained by Qdrant."),
+]
+
+ranker = FastembedRanker(model_name="Xenova/ms-marco-MiniLM-L-6-v2")
+ranker.warm_up()
+reranked_documents = ranker.run(query=query, documents=documents)["documents"]
+
+
+print(reranked_documents["documents"][0])
+
+# Document(id=...,
+#  content: 'fastembed is supported by and maintained by Qdrant.',
+#  score: 5.472434997558594..)

{fastembed_haystack-1.2.0 → fastembed_haystack-1.4.0}/examples/sparse_example.py

@@ -4,6 +4,7 @@
 # involving indexing and retrieval of documents.
 
 from haystack import Document
+
 from haystack_integrations.components.embedders.fastembed import FastembedSparseDocumentEmbedder
 
 document_list = [

{fastembed_haystack-1.2.0 → fastembed_haystack-1.4.0}/pydoc/config.yml

@@ -6,7 +6,8 @@ loaders:
         "haystack_integrations.components.embedders.fastembed.fastembed_document_embedder",
         "haystack_integrations.components.embedders.fastembed.fastembed_text_embedder",
         "haystack_integrations.components.embedders.fastembed.fastembed_sparse_document_embedder",
-        "haystack_integrations.components.embedders.fastembed.fastembed_sparse_text_embedder"
+        "haystack_integrations.components.embedders.fastembed.fastembed_sparse_text_embedder",
+        "haystack_integrations.components.rankers.fastembed.ranker"
       ]
     ignore_when_discovered: ["__init__"]
 processors:

{fastembed_haystack-1.2.0 → fastembed_haystack-1.4.0}/pyproject.toml

@@ -23,7 +23,7 @@ classifiers = [
   "Programming Language :: Python :: Implementation :: CPython",
   "Programming Language :: Python :: Implementation :: PyPy",
 ]
-dependencies = ["haystack-ai>=2.0.1", "fastembed>=0.2.5"]
+dependencies = ["haystack-ai>=2.0.1", "fastembed>=0.4.2"]
 
 [project.urls]
 Source = "https://github.com/deepset-ai/haystack-core-integrations"
@@ -42,6 +42,7 @@ root = "../.."
 git_describe_command = 'git describe --tags --match="integrations/fastembed-v[0-9]*"'
 
 [tool.hatch.envs.default]
+installer = "uv"
 dependencies = [
   "coverage[toml]>=6.5",
   "pytest",
@@ -50,22 +51,25 @@ dependencies = [
   "haystack-pydoc-tools",
 ]
 [tool.hatch.envs.default.scripts]
-test = "pytest --reruns 3 --reruns-delay 30 -x {args:tests}"
-test-cov = "coverage run -m pytest --reruns 3 --reruns-delay 30 -x {args:tests}"
+test = "pytest {args:tests}"
+test-cov = "coverage run -m pytest {args:tests}"
+test-cov-retry = "test-cov --reruns 3 --reruns-delay 30 -x"
 cov-report = ["- coverage combine", "coverage report"]
 cov = ["test-cov", "cov-report"]
+cov-retry = ["test-cov-retry", "cov-report"]
 docs = ["pydoc-markdown pydoc/config.yml"]
 
 [[tool.hatch.envs.all.matrix]]
 python = ["3.8", "3.9", "3.10", "3.11", "3.12"]
 
 [tool.hatch.envs.lint]
+installer = "uv"
 detached = true
-dependencies = ["black>=23.1.0", "mypy>=1.0.0", "ruff>=0.0.243", "numpy"]
+dependencies = ["pip", "black>=23.1.0", "mypy>=1.0.0", "ruff>=0.0.243", "numpy"]
 [tool.hatch.envs.lint.scripts]
 typing = "mypy --install-types --non-interactive --explicit-package-bases {args:src/ tests}"
 style = ["ruff check {args:.}", "black --check --diff {args:.}"]
-fmt = ["black {args:.}", "ruff --fix {args:.}", "style"]
+fmt = ["black {args:.}", "ruff check --fix {args:.}", "style"]
 all = ["style", "typing"]
 
 [tool.black]
@@ -76,6 +80,8 @@ skip-string-normalization = true
 [tool.ruff]
 target-version = "py38"
 line-length = 120
+
+[tool.ruff.lint]
 select = [
   "A",
   "ARG",
@@ -126,13 +132,13 @@ unfixable = [
   "F401",
 ]
 
-[tool.ruff.isort]
-known-first-party = ["src"]
+[tool.ruff.lint.isort]
+known-first-party = ["haystack_integrations"]
 
-[tool.ruff.flake8-tidy-imports]
+[tool.ruff.lint.flake8-tidy-imports]
 ban-relative-imports = "parents"
 
-[tool.ruff.per-file-ignores]
+[tool.ruff.lint.per-file-ignores]
 # Tests can use magic values, assertions, and relative imports
 "tests/**/*" = ["PLR2004", "S101", "TID252"]
 # examples can contain "print" commands
@@ -148,6 +154,10 @@ omit = ["*/tests/*", "*/__init__.py"]
 show_missing = true
 exclude_lines = ["no cov", "if __name__ == .__main__.:", "if TYPE_CHECKING:"]
 
+[tool.pytest.ini_options]
+minversion = "6.0"
+markers = ["unit: unit tests", "integration: integration tests"]
+
 [[tool.mypy.overrides]]
 module = [
   "haystack.*",

{fastembed_haystack-1.2.0 → fastembed_haystack-1.4.0}/src/haystack_integrations/components/embedders/fastembed/embedding_backend/fastembed_backend.py

@@ -1,4 +1,4 @@
-from typing import ClassVar, Dict, List, Optional
+from typing import Any, ClassVar, Dict, List, Optional
 
 from haystack.dataclasses.sparse_embedding import SparseEmbedding
 from tqdm import tqdm
@@ -73,14 +73,19 @@ class _FastembedSparseEmbeddingBackendFactory:
         cache_dir: Optional[str] = None,
         threads: Optional[int] = None,
         local_files_only: bool = False,
+        model_kwargs: Optional[Dict[str, Any]] = None,
     ):
-        embedding_backend_id = f"{model_name}{cache_dir}{threads}"
+        embedding_backend_id = f"{model_name}{cache_dir}{threads}{local_files_only}{model_kwargs}"
 
         if embedding_backend_id in _FastembedSparseEmbeddingBackendFactory._instances:
             return _FastembedSparseEmbeddingBackendFactory._instances[embedding_backend_id]
 
         embedding_backend = _FastembedSparseEmbeddingBackend(
-            model_name=model_name, cache_dir=cache_dir, threads=threads, local_files_only=local_files_only
+            model_name=model_name,
+            cache_dir=cache_dir,
+            threads=threads,
+            local_files_only=local_files_only,
+            model_kwargs=model_kwargs,
         )
         _FastembedSparseEmbeddingBackendFactory._instances[embedding_backend_id] = embedding_backend
         return embedding_backend
@@ -97,9 +102,16 @@ class _FastembedSparseEmbeddingBackend:
         cache_dir: Optional[str] = None,
         threads: Optional[int] = None,
         local_files_only: bool = False,
+        model_kwargs: Optional[Dict[str, Any]] = None,
     ):
+        model_kwargs = model_kwargs or {}
+
         self.model = SparseTextEmbedding(
-            model_name=model_name, cache_dir=cache_dir, threads=threads, local_files_only=local_files_only
+            model_name=model_name,
+            cache_dir=cache_dir,
+            threads=threads,
+            local_files_only=local_files_only,
+            **model_kwargs,
         )
 
     def embed(self, data: List[List[str]], progress_bar=True, **kwargs) -> List[SparseEmbedding]:

{fastembed_haystack-1.2.0 → fastembed_haystack-1.4.0}/src/haystack_integrations/components/embedders/fastembed/fastembed_sparse_document_embedder.py

@@ -62,6 +62,7 @@ class FastembedSparseDocumentEmbedder:
         local_files_only: bool = False,
         meta_fields_to_embed: Optional[List[str]] = None,
         embedding_separator: str = "\n",
+        model_kwargs: Optional[Dict[str, Any]] = None,
     ):
         """
         Create an FastembedDocumentEmbedder component.
@@ -81,6 +82,7 @@ class FastembedSparseDocumentEmbedder:
         :param local_files_only: If `True`, only use the model files in the `cache_dir`.
         :param meta_fields_to_embed: List of meta fields that should be embedded along with the Document content.
         :param embedding_separator: Separator used to concatenate the meta fields to the Document content.
+        :param model_kwargs: Dictionary containing model parameters such as `k`, `b`, `avg_len`, `language`.
         """
 
         self.model_name = model
@@ -92,6 +94,7 @@ class FastembedSparseDocumentEmbedder:
         self.local_files_only = local_files_only
         self.meta_fields_to_embed = meta_fields_to_embed or []
         self.embedding_separator = embedding_separator
+        self.model_kwargs = model_kwargs
 
     def to_dict(self) -> Dict[str, Any]:
         """
@@ -110,6 +113,7 @@ class FastembedSparseDocumentEmbedder:
             local_files_only=self.local_files_only,
             meta_fields_to_embed=self.meta_fields_to_embed,
             embedding_separator=self.embedding_separator,
+            model_kwargs=self.model_kwargs,
         )
 
     def warm_up(self):
@@ -122,6 +126,7 @@ class FastembedSparseDocumentEmbedder:
                 cache_dir=self.cache_dir,
                 threads=self.threads,
                 local_files_only=self.local_files_only,
+                model_kwargs=self.model_kwargs,
             )
 
     def _prepare_texts_to_embed(self, documents: List[Document]) -> List[str]:

{fastembed_haystack-1.2.0 → fastembed_haystack-1.4.0}/src/haystack_integrations/components/embedders/fastembed/fastembed_sparse_text_embedder.py

@@ -35,6 +35,7 @@ class FastembedSparseTextEmbedder:
         progress_bar: bool = True,
         parallel: Optional[int] = None,
         local_files_only: bool = False,
+        model_kwargs: Optional[Dict[str, Any]] = None,
     ):
         """
         Create a FastembedSparseTextEmbedder component.
@@ -50,6 +51,7 @@ class FastembedSparseTextEmbedder:
                 If 0, use all available cores.
                 If None, don't use data-parallel processing, use default onnxruntime threading instead.
         :param local_files_only: If `True`, only use the model files in the `cache_dir`.
+        :param model_kwargs: Dictionary containing model parameters such as `k`, `b`, `avg_len`, `language`.
         """
 
         self.model_name = model
@@ -58,6 +60,7 @@ class FastembedSparseTextEmbedder:
         self.progress_bar = progress_bar
         self.parallel = parallel
         self.local_files_only = local_files_only
+        self.model_kwargs = model_kwargs
 
     def to_dict(self) -> Dict[str, Any]:
         """
@@ -74,6 +77,7 @@ class FastembedSparseTextEmbedder:
             progress_bar=self.progress_bar,
             parallel=self.parallel,
             local_files_only=self.local_files_only,
+            model_kwargs=self.model_kwargs,
         )
 
     def warm_up(self):
@@ -86,6 +90,7 @@ class FastembedSparseTextEmbedder:
                 cache_dir=self.cache_dir,
                 threads=self.threads,
                 local_files_only=self.local_files_only,
+                model_kwargs=self.model_kwargs,
             )
 
     @component.output_types(sparse_embedding=SparseEmbedding)

fastembed_haystack-1.4.0/src/haystack_integrations/components/rankers/fastembed/__init__.py

@@ -0,0 +1,3 @@
+from .ranker import FastembedRanker
+
+__all__ = ["FastembedRanker"]

fastembed_haystack-1.4.0/src/haystack_integrations/components/rankers/fastembed/ranker.py

@@ -0,0 +1,202 @@
+from typing import Any, Dict, List, Optional
+
+from haystack import Document, component, default_from_dict, default_to_dict, logging
+
+from fastembed.rerank.cross_encoder import TextCrossEncoder
+
+logger = logging.getLogger(__name__)
+
+
+@component
+class FastembedRanker:
+    """
+    Ranks Documents based on their similarity to the query using
+    [Fastembed models](https://qdrant.github.io/fastembed/examples/Supported_Models/).
+
+    Documents are indexed from most to least semantically relevant to the query.
+
+    Usage example:
+    ```python
+    from haystack import Document
+    from haystack_integrations.components.rankers.fastembed import FastembedRanker
+
+    ranker = FastembedRanker(model_name="Xenova/ms-marco-MiniLM-L-6-v2", top_k=2)
+
+    docs = [Document(content="Paris"), Document(content="Berlin")]
+    query = "What is the capital of germany?"
+    output = ranker.run(query=query, documents=docs)
+    print(output["documents"][0].content)
+
+    # Berlin
+    ```
+    """
+
+    def __init__(
+        self,
+        model_name: str = "Xenova/ms-marco-MiniLM-L-6-v2",
+        top_k: int = 10,
+        cache_dir: Optional[str] = None,
+        threads: Optional[int] = None,
+        batch_size: int = 64,
+        parallel: Optional[int] = None,
+        local_files_only: bool = False,
+        meta_fields_to_embed: Optional[List[str]] = None,
+        meta_data_separator: str = "\n",
+    ):
+        """
+        Creates an instance of the 'FastembedRanker'.
+
+        :param model_name: Fastembed model name. Check the list of supported models in the [Fastembed documentation](https://qdrant.github.io/fastembed/examples/Supported_Models/).
+        :param top_k: The maximum number of documents to return.
+        :param cache_dir: The path to the cache directory.
+                Can be set using the `FASTEMBED_CACHE_PATH` env variable.
+                Defaults to `fastembed_cache` in the system's temp directory.
+        :param threads: The number of threads single onnxruntime session can use. Defaults to None.
+        :param batch_size: Number of strings to encode at once.
+        :param parallel:
+                If > 1, data-parallel encoding will be used, recommended for offline encoding of large datasets.
+                If 0, use all available cores.
+                If None, don't use data-parallel processing, use default onnxruntime threading instead.
+        :param local_files_only: If `True`, only use the model files in the `cache_dir`.
+        :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.
+        """
+        if top_k <= 0:
+            msg = f"top_k must be > 0, but got {top_k}"
+            raise ValueError(msg)
+
+        self.model_name = model_name
+        self.top_k = top_k
+        self.cache_dir = cache_dir
+        self.threads = threads
+        self.batch_size = batch_size
+        self.parallel = parallel
+        self.local_files_only = local_files_only
+        self.meta_fields_to_embed = meta_fields_to_embed or []
+        self.meta_data_separator = meta_data_separator
+        self._model = None
+
+    def to_dict(self) -> Dict[str, Any]:
+        """
+        Serializes the component to a dictionary.
+
+        :returns:
+            Dictionary with serialized data.
+        """
+        return default_to_dict(
+            self,
+            model_name=self.model_name,
+            top_k=self.top_k,
+            cache_dir=self.cache_dir,
+            threads=self.threads,
+            batch_size=self.batch_size,
+            parallel=self.parallel,
+            local_files_only=self.local_files_only,
+            meta_fields_to_embed=self.meta_fields_to_embed,
+            meta_data_separator=self.meta_data_separator,
+        )
+
+    @classmethod
+    def from_dict(cls, data: Dict[str, Any]) -> "FastembedRanker":
+        """
+        Deserializes the component from a dictionary.
+
+        :param data:
+            The dictionary to deserialize from.
+        :returns:
+            The deserialized component.
+        """
+        return default_from_dict(cls, data)
+
+    def warm_up(self):
+        """
+        Initializes the component.
+        """
+        if self._model is None:
+            self._model = TextCrossEncoder(
+                model_name=self.model_name,
+                cache_dir=self.cache_dir,
+                threads=self.threads,
+                local_files_only=self.local_files_only,
+            )
+
+    def _prepare_fastembed_input_docs(self, documents: List[Document]) -> List[str]:
+        """
+        Prepare the input by concatenating the document text with the metadata fields specified.
+        :param documents: The list of Document objects.
+
+        :return: A list of strings to be given as input to Fastembed model.
+        """
+        concatenated_input_list = []
+        for doc in documents:
+            meta_values_to_embed = [
+                str(doc.meta[key]) for key in self.meta_fields_to_embed if key in doc.meta and doc.meta.get(key)
+            ]
+            concatenated_input = self.meta_data_separator.join([*meta_values_to_embed, doc.content or ""])
+            concatenated_input_list.append(concatenated_input)
+
+        return concatenated_input_list
+
+    @component.output_types(documents=List[Document])
+    def run(self, query: str, documents: List[Document], top_k: Optional[int] = None):
+        """
+        Returns a list of documents ranked by their similarity to the given query, using FastEmbed.
+
+        :param query:
+            The input query to compare the documents to.
+        :param documents:
+            A list of documents to be ranked.
+        :param top_k:
+            The maximum number of documents to return.
+
+        :returns:
+            A dictionary with the following keys:
+            - `documents`: A list of documents closest to the query, sorted from most similar to least similar.
+
+        :raises ValueError: If `top_k` is not > 0.
+        """
+        if not isinstance(documents, list) or documents and not isinstance(documents[0], Document):
+            msg = "FastembedRanker expects a list of Documents as input. "
+            raise TypeError(msg)
+        if query == "":
+            msg = "No query provided"
+            raise ValueError(msg)
+
+        if not documents:
+            return {"documents": []}
+
+        top_k = top_k or self.top_k
+        if top_k <= 0:
+            msg = f"top_k must be > 0, but got {top_k}"
+            raise ValueError(msg)
+
+        if self._model is None:
+            msg = "The ranker model has not been loaded. Please call warm_up() before running."
+            raise RuntimeError(msg)
+
+        fastembed_input_docs = self._prepare_fastembed_input_docs(documents)
+
+        scores = list(
+            self._model.rerank(
+                query=query,
+                documents=fastembed_input_docs,
+                batch_size=self.batch_size,
+                parallel=self.parallel,
+            )
+        )
+
+        # Combine the two lists into a single list of tuples
+        doc_scores = list(zip(documents, scores))
+
+        # Sort the list of tuples by the score in descending order
+        sorted_doc_scores = sorted(doc_scores, key=lambda x: x[1], reverse=True)
+
+        # Get the top_k documents
+        top_k_documents = []
+        for doc, score in sorted_doc_scores[:top_k]:
+            doc.score = score
+            top_k_documents.append(doc)
+
+        return {"documents": top_k_documents}

{fastembed_haystack-1.2.0 → fastembed_haystack-1.4.0}/tests/test_fastembed_backend.py

@@ -2,6 +2,7 @@ from unittest.mock import patch
 
 from haystack_integrations.components.embedders.fastembed.embedding_backend.fastembed_backend import (
     _FastembedEmbeddingBackendFactory,
+    _FastembedSparseEmbeddingBackendFactory,
 )
 
 
@@ -44,3 +45,28 @@ def test_embedding_function_with_kwargs(mock_instructor):  # noqa: ARG001
     embedding_backend.model.embed.assert_called_once_with(data)
     # restore the factory stateTrue
     _FastembedEmbeddingBackendFactory._instances = {}
+
+
+@patch("haystack_integrations.components.embedders.fastembed.embedding_backend.fastembed_backend.SparseTextEmbedding")
+def test_model_kwargs_initialization(mock_instructor):
+    bm25_config = {
+        "k": 1.2,
+        "b": 0.75,
+        "avg_len": 300.0,
+        "language": "english",
+        "token_max_length": 40,
+    }
+
+    # Invoke the backend factory with the BM25 configuration
+    _FastembedSparseEmbeddingBackendFactory.get_embedding_backend(
+        model_name="Qdrant/bm25",
+        model_kwargs=bm25_config,
+    )
+
+    # Check if SparseTextEmbedding was called with the correct arguments
+    mock_instructor.assert_called_once_with(
+        model_name="Qdrant/bm25", cache_dir=None, threads=None, local_files_only=False, **bm25_config
+    )
+
+    # Restore factory state after the test
+    _FastembedSparseEmbeddingBackendFactory._instances = {}

{fastembed_haystack-1.2.0 → fastembed_haystack-1.4.0}/tests/test_fastembed_document_embedder.py

@@ -3,6 +3,7 @@ from unittest.mock import MagicMock, patch
 import numpy as np
 import pytest
 from haystack import Document, default_from_dict
+
 from haystack_integrations.components.embedders.fastembed.fastembed_document_embedder import (
     FastembedDocumentEmbedder,
 )

fastembed_haystack-1.4.0/tests/test_fastembed_ranker.py

@@ -0,0 +1,292 @@
+from unittest.mock import MagicMock
+
+import pytest
+from haystack import Document, default_from_dict
+
+from haystack_integrations.components.rankers.fastembed.ranker import (
+    FastembedRanker,
+)
+
+
+class TestFastembedRanker:
+    def test_init_default(self):
+        """
+        Test default initialization parameters for FastembedRanker.
+        """
+        ranker = FastembedRanker(model_name="BAAI/bge-reranker-base")
+        assert ranker.model_name == "BAAI/bge-reranker-base"
+        assert ranker.top_k == 10
+        assert ranker.cache_dir is None
+        assert ranker.threads is None
+        assert ranker.batch_size == 64
+        assert ranker.parallel is None
+        assert not ranker.local_files_only
+        assert ranker.meta_fields_to_embed == []
+        assert ranker.meta_data_separator == "\n"
+
+    def test_init_with_parameters(self):
+        """
+        Test custom initialization parameters for FastembedRanker.
+        """
+        ranker = FastembedRanker(
+            model_name="BAAI/bge-reranker-base",
+            top_k=64,
+            cache_dir="fake_dir",
+            threads=2,
+            batch_size=50,
+            parallel=1,
+            local_files_only=True,
+            meta_fields_to_embed=["test_field"],
+            meta_data_separator=" | ",
+        )
+        assert ranker.model_name == "BAAI/bge-reranker-base"
+        assert ranker.top_k == 64
+        assert ranker.cache_dir == "fake_dir"
+        assert ranker.threads == 2
+        assert ranker.batch_size == 50
+        assert ranker.parallel == 1
+        assert ranker.local_files_only
+        assert ranker.meta_fields_to_embed == ["test_field"]
+        assert ranker.meta_data_separator == " | "
+
+    def test_init_with_incorrect_input(self):
+        """
+        Test for checking incorrect input format on init
+        """
+        with pytest.raises(
+            ValueError,
+            match="top_k must be > 0, but got 0",
+        ):
+            FastembedRanker(model_name="Xenova/ms-marco-MiniLM-L-12-v2", top_k=0)
+
+        with pytest.raises(
+            ValueError,
+            match="top_k must be > 0, but got -3",
+        ):
+            FastembedRanker(model_name="Xenova/ms-marco-MiniLM-L-12-v2", top_k=-3)
+
+    def test_to_dict(self):
+        """
+        Test serialization of FastembedRanker to a dictionary, using default initialization parameters.
+        """
+        ranker = FastembedRanker(model_name="BAAI/bge-reranker-base")
+        ranker_dict = ranker.to_dict()
+        assert ranker_dict == {
+            "type": "haystack_integrations.components.rankers.fastembed.ranker.FastembedRanker",
+            "init_parameters": {
+                "model_name": "BAAI/bge-reranker-base",
+                "top_k": 10,
+                "cache_dir": None,
+                "threads": None,
+                "batch_size": 64,
+                "parallel": None,
+                "local_files_only": False,
+                "meta_fields_to_embed": [],
+                "meta_data_separator": "\n",
+            },
+        }
+
+    def test_to_dict_with_custom_init_parameters(self):
+        """
+        Test serialization of FastembedRanker to a dictionary, using custom initialization parameters.
+        """
+        ranker = FastembedRanker(
+            model_name="BAAI/bge-reranker-base",
+            cache_dir="fake_dir",
+            threads=2,
+            top_k=5,
+            batch_size=50,
+            parallel=1,
+            local_files_only=True,
+            meta_fields_to_embed=["test_field"],
+            meta_data_separator=" | ",
+        )
+        ranker_dict = ranker.to_dict()
+        assert ranker_dict == {
+            "type": "haystack_integrations.components.rankers.fastembed.ranker.FastembedRanker",
+            "init_parameters": {
+                "model_name": "BAAI/bge-reranker-base",
+                "cache_dir": "fake_dir",
+                "threads": 2,
+                "top_k": 5,
+                "batch_size": 50,
+                "parallel": 1,
+                "local_files_only": True,
+                "meta_fields_to_embed": ["test_field"],
+                "meta_data_separator": " | ",
+            },
+        }
+
+    def test_from_dict(self):
+        """
+        Test deserialization of FastembedRanker from a dictionary, using default initialization parameters.
+        """
+        ranker_dict = {
+            "type": "haystack_integrations.components.rankers.fastembed.ranker.FastembedRanker",
+            "init_parameters": {
+                "model_name": "BAAI/bge-reranker-base",
+                "cache_dir": None,
+                "threads": None,
+                "top_k": 5,
+                "batch_size": 50,
+                "parallel": None,
+                "local_files_only": False,
+                "meta_fields_to_embed": [],
+                "meta_data_separator": "\n",
+            },
+        }
+        ranker = default_from_dict(FastembedRanker, ranker_dict)
+        assert ranker.model_name == "BAAI/bge-reranker-base"
+        assert ranker.cache_dir is None
+        assert ranker.threads is None
+        assert ranker.top_k == 5
+        assert ranker.batch_size == 50
+        assert ranker.parallel is None
+        assert not ranker.local_files_only
+        assert ranker.meta_fields_to_embed == []
+        assert ranker.meta_data_separator == "\n"
+
+    def test_from_dict_with_custom_init_parameters(self):
+        """
+        Test deserialization of FastembedRanker from a dictionary, using custom initialization parameters.
+        """
+        ranker_dict = {
+            "type": "haystack_integrations.components.rankers.fastembed.ranker.FastembedRanker",
+            "init_parameters": {
+                "model_name": "BAAI/bge-reranker-base",
+                "cache_dir": "fake_dir",
+                "threads": 2,
+                "top_k": 5,
+                "batch_size": 50,
+                "parallel": 1,
+                "local_files_only": True,
+                "meta_fields_to_embed": ["test_field"],
+                "meta_data_separator": " | ",
+            },
+        }
+        ranker = default_from_dict(FastembedRanker, ranker_dict)
+        assert ranker.model_name == "BAAI/bge-reranker-base"
+        assert ranker.cache_dir == "fake_dir"
+        assert ranker.threads == 2
+        assert ranker.top_k == 5
+        assert ranker.batch_size == 50
+        assert ranker.parallel == 1
+        assert ranker.local_files_only
+        assert ranker.meta_fields_to_embed == ["test_field"]
+        assert ranker.meta_data_separator == " | "
+
+    def test_run_incorrect_input_format(self):
+        """
+        Test for checking incorrect input format.
+        """
+        ranker = FastembedRanker(model_name="Xenova/ms-marco-MiniLM-L-12-v2")
+        ranker._model = "mock_model"
+
+        query = "query"
+        string_input = "text"
+        list_integers_input = [1, 2, 3]
+        list_document = [Document("Document 1")]
+
+        with pytest.raises(
+            TypeError,
+            match="FastembedRanker expects a list of Documents as input.",
+        ):
+            ranker.run(query=query, documents=string_input)
+
+        with pytest.raises(
+            TypeError,
+            match="FastembedRanker expects a list of Documents as input.",
+        ):
+            ranker.run(query=query, documents=list_integers_input)
+
+        with pytest.raises(
+            ValueError,
+            match="No query provided",
+        ):
+            ranker.run(query="", documents=list_document)
+
+        with pytest.raises(
+            ValueError,
+            match="top_k must be > 0, but got -3",
+        ):
+            ranker.run(query=query, documents=list_document, top_k=-3)
+
+    def test_run_no_warmup(self):
+        """
+        Test for checking error when calling without a warmup.
+        """
+        ranker = FastembedRanker(model_name="Xenova/ms-marco-MiniLM-L-12-v2")
+
+        query = "query"
+        list_document = [Document("Document 1")]
+
+        with pytest.raises(
+            RuntimeError,
+        ):
+            ranker.run(query=query, documents=list_document)
+
+    def test_run_empty_document_list(self):
+        """
+        Test for no error when sending no documents.
+        """
+        ranker = FastembedRanker(model_name="Xenova/ms-marco-MiniLM-L-12-v2")
+        ranker._model = "mock_model"
+
+        query = "query"
+        list_document = []
+
+        result = ranker.run(query=query, documents=list_document)
+        assert len(result["documents"]) == 0
+
+    def test_embed_metadata(self):
+        """
+        Tests the embedding of metadata fields in document content for ranking.
+        """
+        ranker = FastembedRanker(
+            model_name="model_name",
+            meta_fields_to_embed=["meta_field"],
+        )
+        ranker._model = MagicMock()
+
+        documents = [Document(content=f"document-number {i}", meta={"meta_field": f"meta_value {i}"}) for i in range(5)]
+        query = "test"
+        ranker.run(query=query, documents=documents)
+
+        ranker._model.rerank.assert_called_once_with(
+            query=query,
+            documents=[
+                "meta_value 0\ndocument-number 0",
+                "meta_value 1\ndocument-number 1",
+                "meta_value 2\ndocument-number 2",
+                "meta_value 3\ndocument-number 3",
+                "meta_value 4\ndocument-number 4",
+            ],
+            batch_size=64,
+            parallel=None,
+        )
+
+    @pytest.mark.integration
+    def test_run(self):
+        ranker = FastembedRanker(model_name="Xenova/ms-marco-MiniLM-L-6-v2", top_k=2)
+        ranker.warm_up()
+
+        query = "Who is maintaining Qdrant?"
+        documents = [
+            Document(
+                content="This is built to be faster and lighter than other embedding \
+libraries e.g. Transformers, Sentence-Transformers, etc."
+            ),
+            Document(content="This is some random input"),
+            Document(content="fastembed is supported by and maintained by Qdrant."),
+        ]
+
+        result = ranker.run(query=query, documents=documents)
+
+        assert len(result["documents"]) == 2
+        first_document = result["documents"][0]
+        second_document = result["documents"][1]
+
+        assert isinstance(first_document, Document)
+        assert isinstance(second_document, Document)
+        assert first_document.content == "fastembed is supported by and maintained by Qdrant."
+        assert first_document.score > second_document.score

{fastembed_haystack-1.2.0 → fastembed_haystack-1.4.0}/tests/test_fastembed_sparse_document_embedder.py

@@ -4,6 +4,7 @@ import numpy as np
 import pytest
 from haystack import Document, default_from_dict
 from haystack.dataclasses.sparse_embedding import SparseEmbedding
+
 from haystack_integrations.components.embedders.fastembed.fastembed_sparse_document_embedder import (
     FastembedSparseDocumentEmbedder,
 )
@@ -68,6 +69,7 @@ class TestFastembedSparseDocumentEmbedderDoc:
                 "local_files_only": False,
                 "embedding_separator": "\n",
                 "meta_fields_to_embed": [],
+                "model_kwargs": None,
             },
         }
 
@@ -99,6 +101,7 @@ class TestFastembedSparseDocumentEmbedderDoc:
                 "local_files_only": True,
                 "meta_fields_to_embed": ["test_field"],
                 "embedding_separator": " | ",
+                "model_kwargs": None,
             },
         }
 
@@ -173,7 +176,11 @@ class TestFastembedSparseDocumentEmbedderDoc:
         mocked_factory.get_embedding_backend.assert_not_called()
         embedder.warm_up()
         mocked_factory.get_embedding_backend.assert_called_once_with(
-            model_name="prithvida/Splade_PP_en_v1", cache_dir=None, threads=None, local_files_only=False
+            model_name="prithvida/Splade_PP_en_v1",
+            cache_dir=None,
+            threads=None,
+            local_files_only=False,
+            model_kwargs=None,
         )
 
     @patch(
@@ -274,6 +281,52 @@ class TestFastembedSparseDocumentEmbedderDoc:
             parallel=None,
         )
 
+    def test_init_with_model_kwargs_parameters(self):
+        """
+        Test initialization of FastembedSparseDocumentEmbedder with model_kwargs parameters.
+        """
+        bm25_config = {
+            "k": 1.2,
+            "b": 0.75,
+            "avg_len": 300.0,
+            "language": "english",
+            "token_max_length": 50,
+        }
+
+        embedder = FastembedSparseDocumentEmbedder(
+            model="Qdrant/bm25",
+            model_kwargs=bm25_config,
+        )
+
+        assert embedder.model_kwargs == bm25_config
+
+    @pytest.mark.integration
+    def test_run_with_model_kwargs(self):
+        """
+        Integration test to check the embedding with model_kwargs parameters.
+        """
+        bm42_config = {
+            "alpha": 0.2,
+        }
+
+        embedder = FastembedSparseDocumentEmbedder(
+            model="Qdrant/bm42-all-minilm-l6-v2-attentions",
+            model_kwargs=bm42_config,
+        )
+        embedder.warm_up()
+
+        doc = Document(content="Example content using BM42")
+
+        result = embedder.run(documents=[doc])
+        embedding = result["documents"][0].sparse_embedding
+        embedding_dict = embedding.to_dict()
+
+        assert isinstance(embedding, SparseEmbedding)
+        assert isinstance(embedding_dict["indices"], list)
+        assert isinstance(embedding_dict["values"], list)
+        assert isinstance(embedding_dict["indices"][0], int)
+        assert isinstance(embedding_dict["values"][0], float)
+
     @pytest.mark.integration
     def test_run(self):
         embedder = FastembedSparseDocumentEmbedder(

{fastembed_haystack-1.2.0 → fastembed_haystack-1.4.0}/tests/test_fastembed_sparse_text_embedder.py

@@ -4,6 +4,7 @@ import numpy as np
 import pytest
 from haystack import default_from_dict
 from haystack.dataclasses.sparse_embedding import SparseEmbedding
+
 from haystack_integrations.components.embedders.fastembed.fastembed_sparse_text_embedder import (
     FastembedSparseTextEmbedder,
 )
@@ -53,6 +54,7 @@ class TestFastembedSparseTextEmbedder:
                 "progress_bar": True,
                 "parallel": None,
                 "local_files_only": False,
+                "model_kwargs": None,
             },
         }
 
@@ -78,6 +80,7 @@ class TestFastembedSparseTextEmbedder:
                 "progress_bar": False,
                 "parallel": 1,
                 "local_files_only": True,
+                "model_kwargs": None,
             },
         }
 
@@ -134,7 +137,11 @@ class TestFastembedSparseTextEmbedder:
         mocked_factory.get_embedding_backend.assert_not_called()
         embedder.warm_up()
         mocked_factory.get_embedding_backend.assert_called_once_with(
-            model_name="prithvida/Splade_PP_en_v1", cache_dir=None, threads=None, local_files_only=False
+            model_name="prithvida/Splade_PP_en_v1",
+            cache_dir=None,
+            threads=None,
+            local_files_only=False,
+            model_kwargs=None,
         )
 
     @patch(
@@ -194,6 +201,54 @@ class TestFastembedSparseTextEmbedder:
         with pytest.raises(TypeError, match="FastembedSparseTextEmbedder expects a string as input"):
             embedder.run(text=list_integers_input)
 
+    def test_init_with_model_kwargs_parameters(self):
+        """
+        Test initialization of FastembedSparseTextEmbedder with model_kwargs parameters.
+        """
+        bm25_config = {
+            "k": 1.2,
+            "b": 0.75,
+            "avg_len": 300.0,
+            "language": "english",
+            "token_max_length": 50,
+        }
+
+        embedder = FastembedSparseTextEmbedder(
+            model="Qdrant/bm25",
+            model_kwargs=bm25_config,
+        )
+
+        assert embedder.model_kwargs == bm25_config
+
+    @pytest.mark.integration
+    def test_run_with_model_kwargs(self):
+        """
+        Integration test to check the embedding with model_kwargs parameters.
+        """
+        bm25_config = {
+            "k": 1.2,
+            "b": 0.75,
+            "avg_len": 256.0,
+        }
+
+        embedder = FastembedSparseTextEmbedder(
+            model="Qdrant/bm25",
+            model_kwargs=bm25_config,
+        )
+        embedder.warm_up()
+
+        text = "Example content using BM25"
+
+        result = embedder.run(text=text)
+        embedding = result["sparse_embedding"]
+        embedding_dict = embedding.to_dict()
+
+        assert isinstance(embedding, SparseEmbedding)
+        assert isinstance(embedding_dict["indices"], list)
+        assert isinstance(embedding_dict["values"], list)
+        assert isinstance(embedding_dict["indices"][0], int)
+        assert isinstance(embedding_dict["values"][0], float)
+
     @pytest.mark.integration
     def test_run(self):
         embedder = FastembedSparseTextEmbedder(

{fastembed_haystack-1.2.0 → fastembed_haystack-1.4.0}/tests/test_fastembed_text_embedder.py

@@ -3,6 +3,7 @@ from unittest.mock import MagicMock, patch
 import numpy as np
 import pytest
 from haystack import default_from_dict
+
 from haystack_integrations.components.embedders.fastembed.fastembed_text_embedder import (
     FastembedTextEmbedder,
 )