v0.6.0: Provider-aware Token Limits

🚀 Release v0.6.0: Provider-aware Token Limits

✨ Features

• Provider-aware chunking: Implement intelligent chunking that respects embedding model token limits
• Auto-configuration: Cache automatically detects and sets MaxTokens from the embedding provider
• Efficient processing: Only chunks text when it exceeds the model's token capacity, avoiding unnecessary overhead

🔧 API Changes

• Added GetMaxTokens() method to EmbeddingProvider interface
• OpenAI provider now exposes model-specific token limits (8191 tokens for supported models)

🧪 Testing

• Added comprehensive tests for chunking behavior with MaxTokens thresholds
• Updated mock providers to implement new interface methods
• All tests pass with 90%+ coverage

📝 Changes

• Modified cache.Set() to count tokens first before chunking decisions
• Updated chunker logic to use MaxTokens instead of ChunkSize for threshold checks
• Enhanced error handling and validation for token counting operations

🔄 Migration Notes

This is a minor version bump with backward-compatible changes. Existing code will continue to work, but chunking behavior is now more efficient by avoiding unnecessary processing of texts that fit within model limits.

v0.5.2 - Go 1.25.3 and Dependency Updates

What's Changed

This patch release updates the Go version and brings in the latest dependency updates.

Go Version Update

• Go 1.25.3: Updated from Go 1.24 to 1.25.3
• CI Updates: Updated GitHub Actions workflows to use Go 1.25

Dependency Updates

• Redis Client: github.com/redis/go-redis/v9 from 9.14.0 → 9.17.0
• Crypto Library: golang.org/x/crypto updated to latest version
• CI Actions: actions/checkout from v5 → v6

Maintenance

• All tests pass with the new Go version
• CI pipeline updated to use Go 1.25
• Security and performance improvements from dependency updates

Full Changelog: v0.5.1...v0.5.2

v0.5.1 - Simplified Chunking API

What's Changed

This patch release simplifies the chunking configuration API for better usability.

Refactoring

• Simplified API: Replaced WithChunking(config) and WithoutChunking() with a single WithChunking(enabled bool, config ...chunker.ChunkConfig) function
• Backward Compatible: Default chunking behavior remains unchanged (enabled by default)
• Flexible Configuration: Optional config parameter allows custom chunking settings when enabling

API Changes

Before:

// Enable with custom config
options.WithChunking(customConfig)

// Disable chunking  
options.WithoutChunking()

After:

// Enable with defaults (unchanged default behavior)
options.WithChunking(true)

// Enable with custom config
options.WithChunking(true, customConfig)

// Disable chunking
options.WithChunking(false)

Testing

• All existing tests pass
• Updated test code to use new API
• No breaking changes to functionality

Full Changelog: v0.5.0...v0.5.1

v0.5.0 - Automatic Text Chunking

What's New

This release adds automatic text chunking with aggregate embeddings for handling documents that exceed embedding model token limits.

Features

• Automatic Text Chunking: New chunker package with configurable strategies

  • FixedOverlapChunker: splits text with configurable size and overlap
  • Uses tiktoken for accurate token counting (cl100k_base encoding)
  • Default: 512 token chunks with 50 token overlap, 8191 max tokens

• Batch Embedding Support: Efficient multi-text embedding

  • New BatchEmbeddingProvider interface
  • Implemented in OpenAI provider (up to 2048 texts per request)
  • Automatic fallback to individual embeddings if batch not supported

• Aggregate Embedding Approach

  • Chunks text when exceeding token limits
  • Embeds all chunks using batch API for performance
  • Averages chunk embeddings into single aggregate embedding
  • Stores as single entry (no derived keys needed)

• Improved Precision: Converted from float32 to float64

  • Uses OpenAI's native float64 format
  • Updated all similarity functions and backends
  • Better precision for similarity calculations

Configuration

• Chunking enabled by default with sensible defaults
• WithChunking() to customize chunk size/overlap/strategy
• WithoutChunking() to disable if not needed
• Zero-config works out of the box

Testing

• Comprehensive chunker package tests
• Cache integration tests for aggregate embeddings
• All existing tests pass

Full Changelog: https://github.com/botirkhaltaev/semanticcache/compare/v0.4.0...v0.5.0

v0.4.0 - Async Operations Support

feat: add async methods for all backends

- Add asynchronous SetAsync and GetAsync methods for FIFO, LFU, LRU backends
- Add async support for Redis backend
- Improve performance and responsiveness with non-blocking operations
- Add comprehensive examples and performance guide
- Update documentation with async operations section

v0.3.0: Major restructuring to modular architecture

🏗️ Major Restructuring to Modular Architecture

This release represents a significant architectural improvement, transforming the semantic cache library from a monolithic to a clean, modular structure.

✨ New Features

• Modular Package Structure: Split functionality into dedicated packages
  • similarity/ - All similarity algorithms (cosine, euclidean, etc.)
  • options/ - Functional configuration options
• Enhanced Documentation: Added comprehensive AI agent instructions
  • CLAUDE.md - Claude-specific development guide
  • AGENTS.md - Universal AI agent instructions

🔧 Improvements

• Cleaner Imports: More intuitive package organization
• Better Code Organization: Logical separation of concerns
• Enhanced Developer Experience: Clearer API structure
• Comprehensive Testing: All tests passing with zero linting errors

📖 Updated Usage

Before (v0.2.x):

cache := semanticcache.New(
    semanticcache.WithLRUBackend(100),
    semanticcache.WithSimilarityComparator(semanticcache.CosineSimilarity),
)

After (v0.3.0):

import (
    "github.com/botirk38/semanticcache"
    "github.com/botirk38/semanticcache/options"
    "github.com/botirk38/semanticcache/similarity"
)

cache := semanticcache.New(
    options.WithLRUBackend(100),
    options.WithSimilarityComparator(similarity.CosineSimilarity),
)

⚠️ Breaking Changes

• Import paths changed: Update your imports to use the new package structure
• Function locations: Options moved to options package, similarity functions to similarity package
• API compatibility: Core functionality remains the same, only import paths changed

🔧 Migration Guide

1. Update imports to use options.With*() functions
2. Update similarity function calls to use similarity.*Similarity
3. All other API usage remains identical

🛠️ Development

• Zero linting issues (golangci-lint run)
• All tests passing (go test ./...)
• Clean modular architecture
• Ready for production use

📦 Installation

go get github.com/botirk38/semanticcache@v0.3.0

Release v0.2.0

Major Features

• 🔌 Pluggable backends architecture
• 🔌 Pluggable providers architecture
• 🚀 Redis backend support with vector search
• 💾 Multiple in-memory cache strategies (LRU, LFU, FIFO)

Security Fixes

• 🔒 Fix TLS MinVersion vulnerability in Redis backend

Enhancements

• 📡 Redis URL parsing with TLS configuration
• 📚 Improved documentation and project structure
• 🧪 Enhanced testing coverage
• ⚙️ CI/CD improvements with dependency management