0
0

19_indexes_document_loader.py 1.2 KB

123456789101112131415161718192021222324252627282930313233343536373839404142
  1. from pathlib import Path
  2. import sys
  3. sys.path.append(str(Path(__file__).resolve().parents[1]))
  4. from _chapter2_config import print_dependency_hint
  5. DATA_FILE = Path(__file__).with_name("data") / "衣服属性.txt"
  6. def main() -> None:
  7. try:
  8. from langchain_community.document_loaders import TextLoader
  9. except ModuleNotFoundError as exc:
  10. print_dependency_hint(exc)
  11. raise SystemExit(1)
  12. try:
  13. from langchain_unstructured import UnstructuredLoader
  14. except ModuleNotFoundError:
  15. UnstructuredLoader = None
  16. if UnstructuredLoader is not None:
  17. loader = UnstructuredLoader(str(DATA_FILE), encoding="utf8")
  18. docs = loader.load()
  19. print(f"unstructured docs-->{docs}")
  20. print(f"len-->{len(docs)}")
  21. print(f"第一行数据-->{docs[0].page_content}")
  22. print("*" * 100)
  23. else:
  24. print("未安装 langchain-unstructured,跳过 UnstructuredLoader 示例。")
  25. loader = TextLoader(str(DATA_FILE), encoding="utf8")
  26. docs = loader.load()
  27. print(f"TextLoader docs-->{docs}")
  28. print(f"len-->{len(docs)}")
  29. print("第一行数据-->{}".format(docs[0].page_content.split("\n")[0]))
  30. if __name__ == "__main__":
  31. main()