0.27.1

What's Changed

• #575 fix the case when POJO has final fields by @langchain4j in #620

Full Changelog: 0.27.0...0.27.1

0.27.0

New Integrations

Embedding Stores

• Integration with Infinispan by @karesti in #552
• Integration with MongoDB by @Martin7-1 in #535

Embedding Models

• NomicEmbeddingModel by @langchain4j in #592

Notable Updates

• Reworked support of AstraDB and Cassandra by @clun in #548
• Nested POJO extract by @tenpigs267 in #575
• Azure OpenAI: added support for responseFormat and other parameters by @jdubois in #605
• #588: Implemented a fallback strategy for LanguageModelQueryRouter by @langchain4j in #593

Other Updates

• Beautifying Maven output by @agoncal in #572
• [ci] publish snapshots with every commit to main by @Shastick in #16
• Docs updates by @amithkoujalgi in #599
• [Docs] Summary table of the integration capabilities by @czelabueno in #609
• Adding disabled implementation of all model interfaces by @edeandrea in #549
• Update language models and tests for Azure OpenAI by @jdubois in #573
• langchain/d.l.data.document test coverage. by @crutcher in #577
• IllegalConfigurationException coverage tests. by @crutcher in #578
• set default lifecycle plugins versions by @hboutemy in #569
• langchain4j/memory.chat test coverage. by @crutcher in #580
• langchain4j/model.output coverage tests. by @crutcher in #581
• Add IT for Azure Blob Storage with Testcontainers by @eddumelendez in #591
• Upgrade com.azure:azure-identity from 1.11.1 to 1.11.2 by @jdubois in #607
• Azure OpenAI : Configure the langchain4j user-agent for reporting by @jdubois in #611

New Contributors

• @agoncal made their first contribution in #572
• @karesti made their first contribution in #552
• @tenpigs267 made their first contribution in #575

Full Changelog: 0.26.1...0.27.0

0.26.1

New Core Features:

• Advanced RAG by @langchain4j in #538
• Multimodality: support image inputs by @langchain4j in #428
• ConversationalRetrievalChain: adding metadata in the prompt by @mateusscheper in #431

New Integrations:

Model providers:

• Mistral AI by @czelabueno in #514
• Wenxin Qianfan by @lane12138 in #499
• Cohere Rerank by @langchain4j in #539

Embedding Stores:

• Azure AI Search by @jdubois in #530
• Qdrant by @Anush008 in #542
• Vearch by @Martin7-1 in #525

Document Loaders:

• Azure Blob Storage by @jdubois in #439
• GitHub by @jdubois in #415
• Tencent COS by @Martin7-1 in #421

Notable Updates:

• OpenAI: support image inputs by @langchain4j in #428
• OpenAI: support shortened embeddings by @langchain4j in #555
• OpenAI: support native images by @langchain4j in #497
• Vertex AI Gemini: support image inputs by @langchain4j in #428
• Vertex AI: image generation by @glaforge in #456
• Ollama: support image inputs by @bidek in #462
• Ollama: list available models by @bidek in #533
• Ollama: Spring Boot starter by @langchain4j in langchain4j/langchain4j-spring#3
• Qwen: support image inputs by @jiangsier-xyz in #537

Technical Improvements and Bugfixes:

• Pre-built ollama models by @langchain4j in #419
• Image tests and docs. by @crutcher in #401
• Add documentation and tests for Document class by @crutcher in #376
• Add DocumentSplitterTest class by @crutcher in #378
• Add DocumentLoaderTest class and update DocumentLoader by @crutcher in #379
• Add DocumentTransformerTest class. by @crutcher in #380
• Add ChatLanguageModelTest class. by @crutcher in #381
• Add Javadoc to TokenCountEstimator and create TokenCountEstimatorTest by @crutcher in #382
• Retriever Interface Add a default method with memoryId by @leojames in #358
• Tests and docs for ImageModel api. by @crutcher in #383
• Tests and docs for JsonSchemaProperty api. by @crutcher in #387
• ToolExecutionRequest docs and tests. by @crutcher in #388
• Tests and docs for ToolParameters by @crutcher in #389
• Implement Metadata class tests and enhance documentation by @crutcher in #377
• ToolSpecification and ToolSpecifications docs and tests. by @crutcher in #390
• TextClassifier tests. by @crutcher in #391
• CosineSimilarity protection from zeros; full coverage tests. by @crutcher in #392
• InMemoryChatMemoryStore, tests by @crutcher in #393
• EmbeddingMatch tests by @crutcher in #394
• Utils test coverage and docs. by @crutcher in #396
• Exceptions utility docs and tests. by @crutcher in #397
• TokenUsage docs and tests. by @crutcher in #398
• Embedding tests and docs. by @crutcher in #399
• ChatMessage, UserMessage, AiMessage tests by @crutcher in #403
• Fix another Neo4jEmbeddingStoreTest error by @vga91 in #441
• Fix function without params for Azure Open AI by @kumlien in #443
• Added skeleton structure for docs by @amithkoujalgi in #458
• Full unittest coverage and docs for ValidationUtils. by @crutcher in #445
• Full test coverage of EmbeddingMatch. by @crutcher in #446
• Ensure that Redis pipelines are closed after usage by @gdarmont in #449
• Add support to Anthropic Claude v2.1 by @sergioa in #450
• Refactor ServiceHelper and add tests and docs. by @crutcher in #452
• Docs and default-case tests for Retriever. by @crutcher in #453
• Restructure ChatMessageType to structured enums. by @crutcher in #457
• set up documentation structure by @LizeRaes in #473
• Tests and Docs for default EmbeddingModel methods. by @crutcher in #454
• Test coverage for EmbeddingStore defaults. by @crutcher in #465
• Coverage test for Image.Builder.url(String) by @crutcher in #466
• GsonJsonCodec tests by @crutcher in #467
• TextSegment and TextSegmentTransformer tests. by @crutcher in #468
• Fix #460: no overlap when previous segment is shorter than maxOverlapSize by @langchain4j in #464
• UserMessage tests. by @crutcher in #469
• EmbeddingStoreIngestor test coverage. by @crutcher in #470
• Prompt test coverage. by @crutcher in #471
• Basic test for StreamingResponseHandler. by @crutcher in #475
• Utils.readBytes() test. by @crutcher in #472
• Add jacoco coverage targets to langchain4j-core, attached to the verify stage. by @crutcher in #474
• Tokenizer default methods tests. by @crutcher in #476
• ContentType migrated to structured enum. by @crutcher in #477
• Move RetryUtilsTest to langchain4j-core with RetryUtils. by @crutcher in #478
• Bump follow-redirects from 1.15.3 to 1.15.4 in /docs by @dependabot in #480
• Fix #444: Added missing GraalVM metadata for Gson and excluded commons-logging by @langchain4j in #482
• Fix #451: SentenceDetectorME is not thread-safe by @langchain4j in #481
• Document VertexAI authentication by @langchain4j in #463
• Add IntelliJ icon by @geoand in #495
• Updated Docs by @amithkoujalgi in #500
• Docu logos and domain by @amithkoujalgi in #503
• action to build and publish javadoc to pages by @LizeRaes in #504
• Flesh out RetryUtils with a policy mechanism. by @crutcher in #484
• ServiceHelper.{loadService, loadFactoryService}() by @crutcher in #485
• SystemMessage tests. by @crutcher in #486
• ImageContent tests. by @crutcher in #487
• ToolExecutionResultMessage tests. by @crutcher in #488
• Make PromptTemplateFactory.Input a FunctionalInterface. by @crutcher in #489
• LanguageModel and StreamingLanguageModel tests. by @crutcher in #491
• Moderation tests. by @crutcher in #492
• Response tests. by @crutcher in #493
• language.TokenCountEstimator tests. by @crutcher in #494
• StructuredPrompt refactor and tests. by @crutcher in #490
• Tighten coverage ratchet. by @crutcher in #483
• Unify langchang4j-core coverage to ...

0.25.0

What's Changed

• Update README.md by @aashish13 in #281
• Cleanup Testcontainers deps by @eddumelendez in #292
• Feature/elastic search restclient constructor by @aphethean1 in #291
• build most modules with jdk 8 by @langchain4j in #295
• Graal Integration for Local JS and Python Execution by @janglada in #263
• graal: cleanup by @langchain4j in #297
• Fix a token usage statistical issue in DefaultAiServices by @jiangsier-xyz in #280
• pass OpenSearchClient directly to OpenSearchEmbeddingStore by @sboeckelmann in #309
• #299 Add null check for ToolsParameters in tokenCountEstimation by @kumlien in #303
• Add note about Quarkus specific examples by @geoand in #329
• Fix PG Vector Index creation statement incorrect by @pascalconfluent in #302
• Use Testcontainers annotation by @eddumelendez in #310
• Use ElasticsearchContainer in Elasticsearch IT by @eddumelendez in #312
• Use Testcontainers in Chroma IT by @eddumelendez in #313
• Enable Milvus IT by using Testcontainers by @eddumelendez in #314
• Use Testcontainers in Ollama IT by @eddumelendez in #315
• Enhanced QWen Model Features and Tokenizer Implementation by @jiangsier-xyz in #317
• Use Testcontainers to run LocalAI by @eddumelendez in #331
• enable langchain4j-graal build in workflow by @shalk in #333
• Use Testcontainers for Weaviate IT by @eddumelendez in #332
• Migration of the AzureOpenAiChatModel to use the Azure OpenAI SDK by @jdubois in #328
• OpenAI: Support parallel tool calling by @langchain4j in #338
• spring-boot-starter add azure openai support by @QuantumBear in #339
• make OpenAI tokenizer more precise by @langchain4j in #346
• Support Java 21 by @jdubois in #336
• Bump org.json:json from 20230618 to 20231013 in /langchain4j-cassandra by @dependabot in #341
• Add Dev Container support by @jdubois in #337
• Correctly configure Maven wrapper by @jdubois in #348
• fix the thread safety issue in InMemoryEmbeddingStore. by @lexluo09 in #351
• Remove quarkus module by @geoand in #345
• Migrate to the new Azure OpenAI 1.0.0-beta.6 SDK by @jdubois in #356
• Update GitHub Actions versions by @jdubois in #357
• Extract document loaders and parsers into separate modules by @langchain4j in #354
• Allow for specifying the organization id in the configuration by @edeandrea in #364
• Fixes #241: Added support for Neo4j Vector Index by @vga91 in #282
• feat: add OllamaChatModel and its corresponding integration test by @fintanmm in #323
• OpenAI DALL·E support by @Heezer in #298
• bumped version to 0.25.0-SNAPSHOT by @langchain4j in #369
• Removed generic AiMessage ctor by @langchain4j in #370
• Fix Neo4jEmbeddingStoreTest error by @vga91 in #368
• Integration with ChatGLM by @Martin7-1 in #360
• Add image generation support with Azure OpenAI by @jdubois in #359
• #365: handle enum arguments in tools properly by @langchain4j in #367
• Add support for "stop" in Azure OpenAI by @jdubois in #372
• Azure OpenAI: add tests for GPT 4 by @jdubois in #371
• Extracted OpenAI Spring Boot starter into a separate module by @langchain4j in #361
• Update Ollama by @langchain4j in #373
• OpenAI: added missing parameters (logit_bias, response_format, seed, user) by @langchain4j in #374
• Support Google Vertex AI Gemini by @langchain4j in #402
• rename langchain4j-graal module by @langchain4j in #405
• Add support for the 3 authentication methods to Azure OpenAI by @jdubois in #375
• fix pinecone QueryRequest usage, upgrade from 0.2.1 to 0.6.0 by @shalk in #406
• fixed pinecone client CVEs by @langchain4j in #409
• Replace Mustache templates with simple implementation to reduce transitive dependencies by @langchain4j in #408
• 385 Metadata takes ownership of its construction Map<String, String> by @langchain4j in #412
• Metadata: deprecated potentially confusing add/from/metadata(String, Object) methods by @langchain4j in #413
• Fixed (#264) by @langchain4j in #414

New Contributors

• @aashish13 made their first contribution in #281
• @eddumelendez made their first contribution in #292
• @aphethean1 made their first contribution in #291
• @janglada made their first contribution in #263
• @sboeckelmann made their first contribution in #309
• @kumlien made their first contribution in #303
• @shalk made their first contribution in #333
• @jdubois made their first contribution in #328
• @QuantumBear made their first contribution in #339
• @dependabot made their first contribution in #341
• @lexluo09 made their first contribution in #351
• @edeandrea made their first contribution in #364
• @vga91 made their first contribution in #282
• @fintanmm made their first contribution in #323

Breaking Changes

• langchain4j-azure-open-ai module was completely reworked (using official Azure SDK now), there might be breaking changes when constructing models in the builders
• langchain4j-spring-boot-starter module is replaced by separate langchain4j-*-spring-boot-starter modules for each provider, e.g. langchain4j-open-ai-spring-boot-starter, all available starters are here
• Spring Boot properties changed: langchain4j.{model type}.{provider}.{property} -> langchain4j.{provider}.{model type}.{property}
• dev.langchain4j.data.document.FileSystemDocumentLoader -> dev.langchain4j.data.document.loader.FileSystemDocumentLoader
• Document parsers (PDF, MS Office) were extracted into their own modules, TextDocumentParser stays in langchain4j module
• Amazon S3 loaders migrated to langchain4j-document-loader-amazon-s3 module
• EmbeddingStoreIngestor moved into langchain4j-core, keeping the same package name
• potential problems: mustache templates were replaced with simple implementation, but should be backwards compatible unless you used advanced features. You can still plug in mustache using SPI.

Full Changelog: 0.24.0...0.25.0

0.24.0

• Integration with OpenSearch by @riferrei
• Vertex AI: embed in batches of 5
• Support of database name configuration for Milvus by @Heezer
• OpenAI: added option to setup a custom Tokenizer, increased default timeouts to 60 seconds
• Azure OpenAI: increased default timeouts to 60 seconds
• Add support for custom base url for OpenAI model in Spring Boot starter by @marlukcz
• Make some aspects of the project swappable by external integrations by @geoand
• In case a model like Vertex does not have a ToolSpecification do not use the generate method expecting one by @SimonVerhoeven
• Include metadata for custom Spring Boot properties by @ThomasVitale
• Add support for loading documents from S3 by @jmgang
• Add BOM to manage artifacts by @Martin7-1
• Integration with PGVector by @kevin-wu-os
• Add support for autocompletion of Boot configuration properties by @jkuipers
• Integration with Ollama by @Martin7-1
• Make AiServices swappable by external implementations by @geoand
• Introduce SPI for HuggingFace by @geoand
• Allow using another metadata key than text_segment in Pinecone by @Artur-
• Typo fix in VertexAI support by @Heezer
• Integration with Amazon Bedrock by @pascalconfluent
• Adding Memory Id to Tool Method Call by @benedictstrube

0.23.0

• Updates to models API: return Response<T> instead of T. Response<T> contains token usage and finish reason.
• All model and embedding store integrations now live in their own modules
• Integration with Vespa by @Heezer
• Integration with Elasticsearch by @Martin7-1
• Integration with Redis by @Martin7-1
• Integration with Milvus by @IuriiKoval
• Integration with Astra DB and Cassandra by @clun
• Added support for overlap in document splitters
• Some bugfixes and smaller improvements

0.22.0

• Integration with Google Vertex AI by @kuraleta
• Offline text classification with embeddings
• Reworked document splitters
• InMemoryEmbeddingStore can now be easily persisted and restored, see serializeToJson(), serializeToFile(), fromJson() and fromFile()
• Added option to easily extract metadata in HtmlTextExtractor
• Fixed #126 and #127

0.21.0

• Integration with Azure OpenAI by @kuraleta
• Integration with Qwen models (DashScope) by @jiangsier-xyz
• Integration with Chroma by @kuraleta
• Support for persistent ChatMemory

0.20.0

Added an option to setup a proxy for OpenAI models (#93)

Added more pre-packaged in-process embedding models (#91):

• bge-small-en
• bge-small-zh

InMemoryEmbeddingStore: return matches from highest to lowest (#90)

0.19.0

• Weaviate integration by @Heezer
• DOC, XLS and PPT loaders by @oognuyh
• Separate chat memory for each user
• Custom in-process embedding models
• Added lots of Javadoc
• Added DocumentTransformer and it's first implementation: HtmlTextExtractor
• OpenAiTokenizer is now more precise and can estimate tokens for tools/functions
• Added option to force tool/function execution in OpenAiChatModel and OpenAiStreamingChatModel
• Some bugfixes and improvements