v0.1.1: LangChain & LlamaIndex Framework Adapters + Multi-Domain Support

Release v0.1.1: Official Framework Integrations & Multi-Domain Evaluation

🚀 What's New

• LangChain Integration (SDCATextSplitter): Drop-in replacement for LangChain text splitters conforming to BaseDocumentTransformer. Preserves table geometries and hierarchical breadcrumbs.
• LlamaIndex Integration (SDCANodeParser): Drop-in node parser conforming to LlamaIndex NodeParser. Directly parses PDFs or LlamaIndex Documents into structured TextNode collections.
• Multi-Domain Enterprise Evaluation: 100% table preservation and 100% section context retention validated across Financial SEC Form 10-K disclosures and commercial legal contracts.
• Test Suite Expansion: 125/125 tests passing cleanly across extraction, baselines, SDCA V1/V2/V3 (STAIR), multi-domain processing, and framework adapters.
• PyPI Release: Live at https://pypi.org/project/sdca-rag/0.1.1/ (pip install sdca-rag==0.1.1).

v0.1.0: Structure-Aware Document Chunking Algorithm (SDCA)

Initial release of SDCA (Structure-Aware Document Chunking Algorithm for RAG Systems).

• Evaluated on Qasper 40-paper benchmark under strict 1,500-token context budget.
• Achieves 60.81% Evidence Recall (+10.8% relative gain over LangChain recursive chunking).
• Achieves 70.40% Hit Rate and 0.4739 NDCG@k.
• Published to PyPI: https://pypi.org/project/sdca-rag/

Authors: Subodh Butt & Ashish Dhillon