Add backwards compatibility methods to the Builder class in OllamaOptions
to maintain compatibility with code using the old withX style methods.
Each deprecated method delegates to its new counterpart and includes
proper Javadoc directing users to the new method name.
All deprecated methods are marked for removal in 1.0.0-M5, giving users
time to migrate to the new style while maintaining functionality.
Add backwards compatibility methods from FunctionCallingOptionsBuilder
to the Builder interface in FunctionCallingOptions. This allows users
to gradually migrate from the old builder style (with* methods) to the
new style while maintaining the same functionality.
Each deprecated method is marked for removal in 1.0.0-M5 and includes
proper Javadoc directing users to the new method names. The methods
are implemented as default methods that delegate to their new
counterparts.
This change streamlines the builder implementation by removing generics
that was complicating the implementation and providing hard to
debug checkstyle warnings.
It adopts a simpler, more direct builder pattern. Key changes:
- Remove generic type parameters from builder interfaces
- Switch to concrete builder implementations with direct field access
- Make all collection getters return unmodifiable views
- Ensure proper copy semantics in builders and options
- Add comprehensive test coverage for builder behavior
Update documentation examples to show correct builder usage across all vector store implementations.
Demonstrate passing required parameters directly in builder() method.
- Rename all specific builder inner classes (PineconeBuilder, MongoDBBuilder, etc.)
to simply Builder for consistency across vector store implementations
- Update code references to use the new standardized Builder class names
The change establishes a consistent naming convention for builder classes
across the vector store implementations, improving code uniformity.
- Remove references of embedding from Document
- Update the javadoc for Document#score to reflect the embedding of either text/media instead of Document embedding
Fixes: #1913
Issue: https://github.com/spring-projects/spring-ai/issues/1913
- Add lineSeparator field to ExtractedTextFormatter with configurable override
- Update deleteTopTextLines and deleteBottomTextLines methods to use custom separator
- Mark old methods as deprecated in favor of new ones with separator parameter
- Update PDF test to use explicit line separator for Windows compatibility
* Introduce CompressionQueryTransformer and RewriteQueryTransformer in the Pre-Retrieval Module.
* Extend Query with conversation history and context.
* Remove QueryRouter sub-module. New design will be considered in the next milestone.
* Update RetrievalAugmentationAdvisor accordingly.
* Add initial documentation for Modular RAG, and establish dedicated page for RAG content.
Fixes gh-1850
Signed-off-by: Thomas Vitale <ThomasVitale@users.noreply.github.com>
Following Amazon Bedrock's recommendations, deprecate individual model chat implementations (Anthropic, Cohere, Llama, Titan, etc.)
in favor of the unified BedrockProxyChatModel. This change promotes the use of Bedrock's Converse API which offers
benefits like model flexibility, extended functionality, and native tool support.
The Converse API does not support embedding operations, so these will remain in the current API and the embedding
model functionality in the existing InvokeModel API will be maintained
- Replace Document.getContent() with getText() across all vector store implementations
- Fix incorrect package declarations in package-info.java files
- Make builder constructors private and implement proper builder patterns
- Add @Nullable annotations for better null safety
- Simplify PineconeVectorStore builder API by requiring essential parameters in factory method
- Make static Map fields final
- Clean up code and improve JavaDoc documentation
The changes focus on making the vector store APIs more consistent,
type-safe, and maintainable while following best practices for
builder patterns and null safety.
This commit refactors the builder pattern implementation across all VectorStore
implementations to make the EmbeddingModel a required constructor parameter
rather than an optional builder method. Key changes include:
- Move embeddingModel from being a builder method to a required constructor parameter
- Make embeddingModel final in AbstractVectorStoreBuilder
- Remove redundant validate() methods since EmbeddingModel validation now happens
in constructor
- Update all VectorStore builder instantiations to pass EmbeddingModel in builder
creation
- Add @Nullable annotations to appropriate methods in VectorStore interface
This change improves the API design by:
1. Enforcing that EmbeddingModel is provided at builder creation time
2. Removing the possibility of forgotten EmbeddingModel configuration
3. Simplifying the builder implementation by moving validation to construction
4. Making the dependency on EmbeddingModel more explicit in the API
Breaking Changes:
- VectorStore builders must now be created with an EmbeddingModel parameter
- The embeddingModel() builder method has been removed from all implementations
Introduce multimodality support for Mistral AI, which currently supports text and vision modalities.
Added integration tests and documentation for the new capability.
Signed-off-by: Thomas Vitale <ThomasVitale@users.noreply.github.com>
- VectorStore implementations may still be returning the embedding values and mapping onto
Document.class with Jackson
This change ignores that field, but vector store implementations should change to not return
the embedding data that is no longer stored in Document
The FactCheckingEvaluator was initially designed for the Bespoke Minicheck
model which doesn't require explicit instructions. This change adds support
for general-purpose LLMs like Claude and GPT-4 that need clear evaluation
instructions in their prompts.
Key changes:
- Introduce separate prompts for Bespoke and general LLMs
- Add factory method for Bespoke Minicheck configuration
- Make evaluation prompt configurable via constructor
This maintains backward compatibility while enabling the evaluator to work
effectively with any LLM implementation.
The OpenAiImageOptions class previously had inconsistent behavior when
setting width and height properties. This change ensures that the size
property is only computed when both dimensions are available, preventing
invalid states in the API requests.
Key changes:
- Update setWidth/setHeight to only set size when both are non-null
- Add test cases to verify null size when dimensions are incomplete
- Improve code organization with proper import conventions
This change aligns with OpenAI's API requirements for image generation
where both dimensions must be specified for valid requests.
- This PR removes the reference of `embedding` from the `Document` object.
- The Document is designed to contain the content and its metadata but not the embedding representation of its content. This provides clear separation of concern.
- Prior to this change, all the vector stores are updated by not depending on the embedding from Document. Instead, use the embedding values returned by the Embedding Models directly: https://github.com/spring-projects/spring-ai/issues/1826Resolves#1781
Removes duplicate state by reusing parent's options field for both function
calling and chat options. This simplifies the inheritance structure and
prevents potential state inconsistencies.
- Remove functionCallingOptions field from DefaultFunctionCallingOptionsBuilder
- Set parent class options with FunctionCallingOptions in child constructor
- Add protected constructor in DefaultChatOptionsBuilder for inheritance
- Initialize options through constructors instead of direct assignment
The Ollama options builder API has been refactored to follow standard Java
builder pattern conventions. This change deprecates all builder methods
prefixed with 'with' in favor of more concise method names, improving API
consistency and usability.
The deprecated methods are marked for removal in version 1.0.0-M5, giving
users time to migrate to the new builder pattern. This change aligns with
our goal of providing a more intuitive and maintainable API surface.
Breaking Changes:
* builder() method now returns Builder instead of OllamaOptions
* Clients using the old fluent API will need to migrate to the new builder pattern
Refactor Ollama options builder methods
This commit deprecates builder methods with the "with" prefix in the VectorStoreObservationContext class
and introduces new methods without the prefix for a cleaner API. For example:
withCollectionName() → collectionName()
withDimensions() → dimensions()
withNamespace() → namespace()
The old methods are marked as deprecated for maintaining backward compatibility.
All vector store implementations have been updated to use the new method names.
- Create SimpleVectorStoreBuilder which can be used to create the SimpleVectorStore instance
- Remove the field embeddingModel from SimpleVectorStore as it can be retrieved from AbstractObservationVectorStore
- Deprecate the existing SimpleVectorStore constructors and use the builder to instantiate it
- Fix the tests
