v0.44.0

Feature

• Use new sem index, gen ctx extractors (#174) (cde4183)

ℹ️ Breaking changes

• CorpusRAGQuery and DocumentRAGQuery replaced by RAGQuery
• CorpusSemanticQuery and DocumentSemanticQuery replaced by SemanticQuery
• semantic_ingest, RAGQuery, and SemanticQuery all require a data_source: deepsearch.cps.client.components.documents.DataSource (check deepsearch-examples for usage samples)
• the above now return a CPSTask which does not contain the project key, so invocations of api.tasks.wait_for should be updated accordingly — to sth like api.tasks.wait_for(EXPLICIT_PROJ_KEY, task.task_id)
• RAGResult / SearchResult:
  • passage renamed to chunk
  • path_in_doc replaced by main_path (+ path_group)
  • RAGGroundingInfo.items renamed to RAGGroundingInfo.retr_items

v0.43.0

Feature

• Generate sdk from new v2 api (#173) (0abbf7f)

v0.42.0

Feature

• Add RAG exceptions (#171) (533e4c5)

v0.41.0

Feature

• Enable control over chunking of URL lists (#170) (536666e)

v0.40.0

Feature

• Expose gen AI params (#165) (dc3db56)

v0.39.2

Fix

• Update dependencies (#164) (aaa9aff)

v0.39.1

Fix

• Align records type with API validation (#163) (8d0f7be)

v0.39.0

Feature

• Expose search results to RAG, update types (#161) (49ec255)

v0.38.1

Fix

• Py38 compatibility (#160) (af32e11)

v0.38.0

Feature

• Expose sem. params, default to hybrid search (#159) (5fa0963)