splitter.py 593 B

123456789101112131415161718192021
  1. from langchain_text_splitters import RecursiveCharacterTextSplitter
  2. def get_knowledge_splitter():
  3. """知识库文档分块器"""
  4. return RecursiveCharacterTextSplitter(
  5. chunk_size=500,
  6. chunk_overlap=50,
  7. separators=["\n\n", "\n", "。", "!", "?", ",", " ", ""],
  8. length_function=len,
  9. )
  10. def get_summary_splitter():
  11. """摘要分块器 (用于对话记忆)"""
  12. return RecursiveCharacterTextSplitter(
  13. chunk_size=1000,
  14. chunk_overlap=100,
  15. separators=["\n\n", "\n", "。", " ", ""],
  16. length_function=len,
  17. )